# Research for cx-listing-playbook This directory is the research base for the pi package cx-listing-playbook: the skills listing-review (which carries references/writing.md), listing-change-log, effect-review, listing-create and featured-offer, the script scripts/report.sh, and two templates, /listing-review and /listing-create. /listing-watch (which writes the standing instruction) moved in 0.3.0 to its own package, `@comchatx/cx-listing-watch`; the rows and positions below that name it are the research its template rests on. Every file here was read on 2026-10-14, except method/listing-create.md, read on 2026-10-27, and method/featured-offer.md, read on 2026-10-28. Sources are weighed in this order and tagged by type: Amazon's own words first, tagged [official] (Seller Central help, SP-API docs and models, Amazon Ads, aboutamazon.com, sell.amazon.com, Amazon staff posts in Seller Forums; Keepa's own documentation and code, and regulators' pages, carry the same tag); then named practitioners and vendors, tagged [opinion]; then published research and Amazon Science, tagged [research]; the researching seat's own reasoning is tagged [seat], and no skill states a [seat] item as fact. Two limits were kept: no Amazon or Keepa API host was called (no `sellingpartnerapi-*`, no `api.keepa.com`; documentation pages only), and no seller data is held here (no store names, seller ids, real seller ASINs or store figures; `` and `` stand in). Seller Central help pages were read through the help hub's own content call, as recorded in method/help-page-access.md, and the standing use of that call is for the office to rule. ## How to read this - Start with the positions below: they are what the skills will say and do, numbered L, W, C, E, S and T so Jack can answer by number. - Each position names the file and P-id that carries its sources; `P--N` is in that file's *Positions* section, and the pages behind it are in the same file's *Pages read* table. - Figures are for human readers and are stated by no skill: they sit under *Facts for human readers only*, and a skill reads the schema, the page or the seller's console at run time instead. ## Files Every research file has one layout: what it covers; Amazon's rules; what practitioners add; published research; what is stable enough for a skill; what cx must read at run time; pages read; open questions; positions; verification. ### angles/ | file | what it covers | funnel stage | skill(s) it feeds | |---|---|---|---| | account-health.md | account status, Account Health rating, policy items, appeals, Account Health Assurance, FBM offer-level deactivation, as constraints on every listing | gate, offer | listing-review; effect-review; /listing-watch | | ai-shopping-surfaces.md | Alexa for Shopping (formerly Rufus), Amazon's AI title rewrites, Review listing changes, Sponsored prompts, AI-drafted copy, synthetic-performer tagging | click-through, conversion | listing-review; listing-change-log; effect-review; writing.md | | brand-protection-and-catalog-contribution.md | contribution precedence, reconciled Catalog against submitted listing, brand roles, other contributors' edits, Report Abuse and violation routes, ASIN identity | all stages (what is shown) | listing-review; listing-change-log; effect-review; writing.md | | competitor-discovery.md | finding rivals: SQP and Search Terms, Keepa search and `listedSince`, the limits of BSR | exposure | listing-review; /listing-watch | | competitor-set.md | building, freezing and reading a competitor set; Keepa snapshots; the set as a control | all (control) | listing-review; effect-review; /listing-watch; scripts/report.sh | | inventory-and-fulfillment.md | stock, buyability, stranded units, stock-out precursors, the delivery promise, FBA reports, fee previews and fee bands (the merge of two seats' files on this angle) | gate, offer | listing-review; listing-change-log; effect-review; /listing-watch; scripts/report.sh | | listing-content-and-health.md | reading a listing: `getListingsItem`, status, issues, Catalog against Listings, suppression reports, non-copy causes | gate, all | listing-review; listing-change-log; writing.md | | marketplaces-and-languages.md | one marketplace per series, regional endpoints, language tags, Build International Listings, JP counting, EU and UK rules | all | listing-change-log; effect-review; writing.md | | off-listing-traffic.md | Brand Store, Amazon Live, Attribution, tailored promotions, influencer video, external traffic as sources of sessions outside search | exposure (non-search) | listing-review; effect-review | | offer-competition-and-hijackers.md | other sellers on the ASIN, Featured Offer share and its denominators, violation routes, Project Zero, Transparency, commingling | offer | listing-review; listing-change-log; effect-review; /listing-watch | | own-traffic-funnel.md | the seller's own sessions, unit session percentage, ratios, denominators, partial days | conversion, all | listing-review; effect-review; scripts/report.sh | | paid-keywords.md | Sponsored Products reports, the keyword gap, ads as confounder or mediator, ad spend from Economics | exposure | listing-review; listing-change-log; effect-review | | price-position.md | price, Featured Offer, promotions, reference prices, fees | offer, click-through | listing-review; listing-change-log; effect-review | | returns.md | return rates by kind, return reasons, return badges, return lag | conversion (post-purchase guardrail) | listing-review; effect-review | | reviews-and-qa.md | ratings, review topics, review sharing, Vine, solicitation rules | click-through, conversion | listing-review; effect-review; writing.md | | search-query-performance.md | SQP, Search Catalog Performance and Search Terms reports: fields, periods, limits, roles | exposure, click-through | listing-review; effect-review; scripts/report.sh | | search-rank-and-volume.md | why no keyword rank is available and what to offer instead | exposure | listing-review | | seasonality-and-events.md | sales events, run-ups and ramp-downs, promotions and their eligibility floors, year-over-year alignment, moderation lead times | all (confounder) | effect-review; listing-change-log; /listing-watch | | vendor-and-hybrid.md | 1P and hybrid brands: account type, the retail offer, vendor write semantics, precedence, vendor A+, vendor reports and clocks (the merge of the angle and element seats' files) | offer, all | listing-review; listing-change-log; effect-review; scripts/report.sh | ### elements/ | file | what it covers | funnel stage | skill(s) it feeds | |---|---|---|---| | a-plus-content.md | A+ and Premium A+: API, eligibility, publish records, validation, content rules, MYE | conversion | listing-review; writing.md; listing-change-log; effect-review | | attributes-and-browse-nodes.md | product-type schema, attributes, browse nodes, item type keyword, compliance attributes, restrictions | findable gate, exposure | listing-review; writing.md; listing-change-log; effect-review | | badges-and-placements.md | Amazon's Choice, Best Seller, New Release, deal and coupon badges, return badges, Climate Pledge Friendly, Prime | click-through | listing-review; writing.md; listing-change-log; effect-review | | brand-story-and-store.md | Brand Story, Brand Store, Store insights, tailored promotions, Posts (discontinued) | conversion, off-listing | listing-review; listing-change-log; effect-review | | bullet-points.md | bullet schema, policy, indexing, what "good" means, change timing | conversion (exposure secondary) | listing-review; writing.md; effect-review | | description.md | description against A+, schema, HTML, contribution, assistant reading | conversion | listing-review; writing.md; effect-review | | images-and-video.md | image rules, media attributes, global images, video, suppression, MYE | click-through (main), conversion (gallery) | listing-review; writing.md; listing-change-log; effect-review | | search-terms.md | `generic_keyword`: write limit against indexing limit, byte counting, policy, whose terms apply | exposure | listing-review; writing.md; effect-review | | title.md | `item_name` and `title_differentiation`, Item Highlights, AI rewrites, title issue codes, policy | exposure, click-through | listing-review; writing.md; listing-change-log; effect-review | | variations.md | families, themes, review sharing, family changes and their risk | findable gate, conversion | listing-review; listing-change-log; effect-review | ### method/ | file | what it covers | funnel stage | skill(s) it feeds | |---|---|---|---| | amazon-change-calendar.md | dated Amazon-side changes that break a before/after reading | all (confounder) | effect-review; listing-change-log | | effect-measurement.md | MYE, windows, controls, confounders, verdicts, noise, placebo, durability | all | effect-review; listing-change-log | | featured-offer.md | how Amazon picks the Featured Offer, reading who holds it and why not, which causes a Listings Items write reaches, the Support API | offer | featured-offer | | funnel-and-search.md | the funnel, its gates, which element serves which stage, Amazon search's own words | all | listing-review | | help-page-access.md | how Seller Central help pages were read, how to cite them, and what each captured page says, by owning file | none (method) | research upkeep; listing-review; writing.md | | (not in this repository) | 54 dated text captures of Seller Central help pages, each named `-v`: Amazon's own text, so the office keeps them and this repository cites them as `capture -v` | none (evidence) | research upkeep | | listing-create.md | creating a listing: stores, new item against existing ASIN, product type and schema, identity and GTIN exemption, the request, variations, preview, write, read-back, sandbox, Build International Listings | none (write path) | listing-create | | listing-policies.md | claims, restricted words, regulators' rules, post-acceptance issues | all (policy gate) | writing.md; listing-change-log | | listing-writes.md | schema conformance, preview, patch and put, feeds, deletes, recording a write | none (write path) | writing.md; listing-change-log | | playbook-defaults.md | three lines the charter once held as rules and Jack moved into the playbook on 2026-08-27 as defaults: how a series is shown after a chart, A+ as conversion knowledge, the store's language as the default the seller's word overrides | none (method) | listing-review; listing-create | | report-lifecycle.md | report and Data Kiosk lifecycles: create, poll, terminal statuses, documents, pagination, schedules, cancels, rate limits, landed receipts | none (data path) | scripts/report.sh; effect-review | | seller-central-only-tools.md | MYE, RLC, VOC, Seller Assistant and other tools without an API; how handoffs are kept | all | listing-review; effect-review; /listing-watch | | sp-api-access.md | roles, rate limits, 429s, account identity | none (access) | scripts/report.sh; listing-review | | triggers.md | trigger families, cadence, alert rules, the standing instruction | all | listing-review; /listing-watch | ## Positions the skills will take Each position is one sentence, followed by the P-ids that carry it and the names of its source pages, Amazon's first. Duplicates across files are merged; where files disagreed, the position takes the side that won (see *Where the files disagreed*). Positions whose sources did not hold after verification and reconciliation are left out. ### listing-review #### The funnel - **L1.** A review first names the stage that fell (findable, exposure, click-through, conversion, or the offer) and the file and field that show it. — P-funnel-and-search-1, 2; P-triggers-1. [official] Search Catalog Performance schema; Sales and Traffic schema; My ASIN is unsearchable. - **L2.** Before any wording change, cx checks the gates (parent ASIN, browse node, buyable offer, launch date, search suppression, `issues` with BUYABLE and DISCOVERABLE, the Featured Offer holder, own stock) and reports a gate failure first, because copy fixes none of them. — P-funnel-and-search-3; P-triggers-2; P-search-terms-13; P-price-position-1; P-inventory-and-fulfillment-1, 3. [official] My ASIN is unsearchable; Suppressed listings; Manage Listings Issues; Becoming the Featured Offer; Understanding Amazon listing status and seller-fulfilled inventory management; getInventorySummaries. - **L3.** Before blaming copy, cx reports any non-copy cause Amazon names: stranded FBA units, inactive or split children, an approval requirement, a pricing-error deactivation, main-image suppression, an adult classification, or an Account Health listing-policy violation. — P-listing-content-and-health-22, 23, 24; P-funnel-and-search-23, 24; P-images-and-video-2. [official] Fix stranded inventory; Pricing Health; Image suppression; Account health rating. - **L4.** The account layer is read before the listing (participation and suspended listings from `getMarketplaceParticipations`, the seller-performance report and the Account Health rating), and an account cause is named before any content edit is proposed. — P-account-health-1; P-sp-api-access-14; P-triggers-23; P-listing-content-and-health-29. [official] Account health rating; Performance Reports; accountHealthReport-2020-11-18.json; sellers.json; Review Account Health updates for FBM offers. - **L5.** Every account-health threshold, status and score is taken from what Amazon returns (`targetValue`, `targetCondition`, `status`, the rating fields), never quoted from memory. — P-account-health-3. [official] Performance Reports; accountHealthReport-2020-11-18.json; Account health rating. - **L6.** An empty `issues` array is not proof that a listing is fine, because Amazon says not every problem has a defined issue and policy removals show elsewhere. — P-account-health-2; P-listing-content-and-health-3. [official] Notification Type Values; Appeal an account deactivation or listing removal; Listings APIs FAQ. - **L7.** A policy item that names the detail page itself (a trademark on the page, a prohibited claim, a page that does not match the product) may get a content fix, while one about the product, its packaging, authenticity, safety or documents is told plainly as something no listing edit resolves. — P-account-health-5, 6. [official] Intellectual Property Policy; Additional information … Restricted Products policy; Invoice requirements for appealing a policy violation; Safety, compliance, and product restrictions. - **L8.** cx never advises closing, deleting or zeroing stock on a listing as a fix for an account-health violation. — P-account-health-7. [official] Account health rating; Additional information … Restricted Products policy. - **L9.** cx tells the seller which account-health facts the connection cannot read (Product Policy Compliance details, Performance Notifications, Manage Your Compliance) and asks for them rather than guessing. — P-account-health-12. [official] Appeal an account deactivation or listing removal; Account health FAQ; Account Health Assurance FAQ. - **L10.** An appeal drafted for the seller follows Amazon's structure (the violation and product, the cause, the reactive steps, the proactive steps), and cx never says Account Health Assurance protects a listing, because Amazon says it covers the account. — P-account-health-13, 14. [official] Additional information … Restricted Products policy; Complete an appeal questionnaire to reinstate selling privileges; Account Health Assurance FAQ; Seller Challenge FAQ. - **L11.** Exposure rests on the title, search terms, item type keyword, browse node and attributes, and Amazon says bullets and description are not always indexed, so they count for search only conditionally. — P-funnel-and-search-4, 5; P-attributes-and-browse-nodes-20, 21; P-description-11, 20. [official] Optimize your product discoverability; Amazon search glossary; Use search terms effectively; Classify your products using Browse Tree Guides; Keyword attributes explained. [research] Semantic Product Search (Nigam et al., KDD 2019). - **L12.** Click-through is read against the main image, the title with Item Highlights, price, badges, rating and the delivery message, and conversion against the detail page together with the offer. — P-funnel-and-search-6, 7, 8; P-title-11, 28. [official] Amazon search glossary; Product title requirements and guidelines; Becoming the Featured Offer. - **L13.** No element is credited or blamed for a stage it does not serve: search terms serve exposure only, the main image click-through, the gallery and video conversion, bullets conversion first, the title exposure and click-through, and A+ and Brand Story conversion only. — P-search-terms-1; P-images-and-video-1; P-bullet-points-1; P-title-11; P-a-plus-content-1; P-brand-story-and-store-1; P-funnel-and-search-30. [official] Use search terms effectively; Product bullet point requirements; A+ Premium Modules; SQP and SCP schemas. - **L14.** Featured Offer factors are stated only as Amazon states them (price, delivery speed and cost, stock, returns, seller performance), never as star rating, Prime, fulfillment method, title or description. — P-price-position-4; P-account-health-9; P-funnel-and-search-8. [official] Becoming the Featured Offer. - **L15.** Amazon no longer keeps a seller eligibility requirement for the Featured Offer, so cx never reports an "ineligible seller", names the store and the page it read, and says a new seller's offer is judged on the same criteria as the brand's. — P-price-position-3; P-triggers-18, 20; P-offer-competition-and-hijackers-15; P-amazon-change-calendar-4. [official] Featured Offer eligibility (G200418100 v23); Updates to Featured Offer eligibility requirements; Becoming the Featured Offer. - **L16.** An offer deactivated for a suspected pricing error is kept apart from a live offer that cannot be featured at its price, and PRICING_HEALTH is a constraint, not a competitor move or a content problem. — P-price-position-2, 10; P-triggers-19; P-offer-competition-and-hijackers-6. [official] Pricing Health; Notification Type Values. - **L17.** cx says "Amazon search", never "A9" or "A10", and states no ranking weight as Amazon's, because Amazon publishes none. — P-funnel-and-search-17. [official] Amazon search glossary. [opinion] Seller Labs; Canopy; Eva. - **L18.** The assistant is called Alexa for Shopping (formerly Rufus), and cx says only that Amazon states it answers from the detail page's listing details, customer reviews and community Q&A, naming no field it reads and neither claiming nor denying that a given field feeds it. — P-ai-shopping-surfaces-1, 2; P-funnel-and-search-18; P-bullet-points-12, 29; P-description-12; P-a-plus-content-11; P-reviews-and-qa-15, 26; P-amazon-change-calendar-15. [official] 'Amazon Rufus' AI experience comes to the Amazon Shopping app; Meet Alexa for Shopping; Alexa for shopping. [research] The technology behind Amazon's GenAI-powered shopping assistant, Rufus (Amazon Science). - **L19.** Complete, accurate structured attributes, the one element Amazon names as helping the assistant, are treated as both exposure and conversion content. — P-ai-shopping-surfaces-3; P-attributes-and-browse-nodes-13. [official] Attributes guide; Alexa for shopping. - **L20.** When the assistant answers wrongly about the product, cx first checks that the Catalog view holds the right facts and then follows the remedy Amazon's assistant page names. — P-ai-shopping-surfaces-4. [official] Alexa for shopping. - **L21.** cx promises no placement in Help Me Decide, AI Shopping Guides or assistant recommendations, and never offers "earn Amazon's Choice", "earn Overall Pick" or "earn Best Seller" as an action. — P-ai-shopping-surfaces-17; P-badges-and-placements-1. [official] Amazon's Choice FAQs; Sales Rank FAQ; Meet Alexa for Shopping; Amazon's new AI Shopping Guides. - **L22.** cx reports no keyword rank, because no Amazon API gives one and a position in `searchCatalogItems`, Keepa search or Creators API results is not search rank, and offers SQP impressions and impression share instead. — P-search-rank-and-volume-1, 2, 7; P-competitor-discovery-4. [official] catalogItems_2022-04-01.json; SQP schema; Seller Analytics Dataset Use Case Guide; Keepa Request.java; Conditions of Use. - **L23.** BSR is a relative rank within a named category, not a unit count, not organic search rank and not the browse node, and a change of rank category or browse node breaks the old series. — P-competitor-set-14; P-competitor-discovery-1; P-triggers-28; P-badges-and-placements-2; P-attributes-and-browse-nodes-22. [official] Keepa Statistics Object; catalogItems_2022-04-01.json; Sales Rank FAQ. [research] Measuring Prices and Price Competition Online. - **L24.** A content change neither cx nor the seller made is a finding first, and any title, the seller's or a competitor's, is read as possibly Amazon-written, because Amazon's AI title rewrites, another contributor's edit and Review listing changes proposals all change what the shopper sees. — P-ai-shopping-surfaces-9, 11; P-title-10, 12; P-funnel-and-search-16; P-brand-protection-and-catalog-contribution-9. [official] Review listing changes; Updates to improve your product titles begin on July 27; Answers to your product title update questions; Notification Type Values. - **L25.** A brand owner hears that Amazon's AI-proposed changes in Review listing changes publish automatically when the review window ends unless addressed, that image and browse-node changes are not shown there, and that View Change History is documented for AI-applied changes only. — P-ai-shopping-surfaces-9, 10; P-title-12; P-brand-protection-and-catalog-contribution-11. [official] Review listing changes; Notification Type Values; Engage With Amazon: Get your Product Title Update Questions Answered! #### Trigger families - **L26.** A trigger names its family (competitor, demand, own-listing, constraint), the stage it threatens and the file that shows it, and never by itself justifies an edit. — P-triggers-1. [official] Manage Listings Issues. [opinion] Google SRE ch. 6. - **L27.** A fall in SQP search volume or totals is demand, and a fall in the ASIN's share while totals hold is competition or the listing. — P-triggers-3; P-search-query-performance-1, 6; P-competitor-discovery-3; P-seasonality-and-events-10. [official] SQP schema; Analytics Reports. [research] Seasonal relevance in e-commerce search (Yang et al. 2021). [opinion] Intentwise; ZonGuru. - **L28.** Constraint triggers are named and handled before any content review: a new seller holding the Featured Offer, a lost Featured Offer, PRICING_HEALTH, an ASIN redirect or merge read from Catalog or Keepa, a change of rank category, a fee-tier change, stranded units with a removal deadline, and a stock or capacity risk. — P-triggers-17, 19, 24, 25, 26, 27, 28; P-inventory-and-fulfillment-6, 7, 20. [official] Becoming the Featured Offer; Pricing Health; Notification Type Values; Fix stranded inventory; Stranded inventory required removals; Fulfillment by Amazon (FBA) Reports; FBA capacity limits; Keepa Product.java. - **L29.** Returns and Voice of the Customer are own-listing triggers, a listing cause is named only when comments, Listing Insight or review topics agree, and a page-misdescription complaint gets an observation window in weeks. — P-returns-1, 15; P-account-health-10; P-seller-central-only-tools-10; P-funnel-and-search-28. [official] Voice of the Customer; Resolve your CX Health issues; Frequently returned item badge. - **L30.** Sponsored Products search-term reports feed a keyword-gap trigger, but a term's absence from that report is not evidence that it had no impressions. — P-paid-keywords-1, 2, 3. [official] Search term reports [v3]; Sponsored ads reporting FAQ (version 3). [opinion] Perpetua; Scale Insights. - **L31.** A session change without a matching SQP change is read as traffic from outside search, and off-listing activity (Store changes, Sponsored Brands or display, Live, tailored promotions, claim codes) is asked about before a listing gets credit or blame. — P-off-listing-traffic-1, 2. [official] Search Query Performance dashboard; sellerSalesAndTrafficReport.json; Business reports glossary. - **L32.** Competitor moves (price, promotions, stock, new sellers, content) are read from cx's kept copies and reported as findings for the seller to classify. — P-offer-competition-and-hijackers-1; P-brand-protection-and-catalog-contribution-8; P-triggers-5, 22; P-competitor-set-11. [official] productPricingV0.json; Promotions API; Keepa Product.java. [research] An Empirical Analysis of Algorithmic Pricing on Amazon Marketplace. #### The offer and stock - **L33.** Before any content verdict, cx reads the offers on the ASIN with `getItemOffers` (counts by condition and channel, the Featured Offer holder), and treats a `getCompetitiveSummary` or Keepa Buy Box reading as a snapshot, not a share. — P-offer-competition-and-hijackers-1, 4; P-brand-protection-and-catalog-contribution-8. [official] productPricingV0.json; productPricing_2022-05-01.json; Keepa Stats.java. [research] Chen, Mislove, Wilson (WWW 2016). - **L34.** When the seller's Featured Offer share falls, stock and Pricing Health are checked before the other offers. — P-offer-competition-and-hijackers-6. [official] Becoming the Featured Offer; Pricing Health; Notification Type Values. - **L35.** A new seller is sorted by case before a route is named: a confirmed counterfeit goes to Report a Violation, a changed detail page to Report Abuse or the contribution route, and a distribution or pricing agreement stays outside Amazon. — P-offer-competition-and-hijackers-9; P-brand-protection-and-catalog-contribution-6, 7. [official] Intellectual property for Rights Owners and Registered Agents; Report a violation; Suggest changes to your product detail page; Project Zero. [opinion] Top 3 Mistakes when Reporting a Violation through Brand Registry. - **L36.** Project Zero is named only for a confirmed counterfeit, Transparency only as blocking uncoded or mismatched units on enrolled products, and each route is given with the Brand Registry role it needs. — P-offer-competition-and-hijackers-10, 11, 12. [official] Project Zero; Transparency; Report a Violation: all reports in one place. - **L37.** Reviews on an ASIN are shared by every seller on it, so a "not as described" wave can come from another seller's units, and commingling is not presented as a current cause of mixed inventory. — P-offer-competition-and-hijackers-7, 8. [official] FBA Barcode Choice and Labeling FAQ; FBA barcode requirements. - **L38.** cx quotes no fixed price band for Featured Offer rotation, because Amazon states none, and on an EU store it checks whether a second offer is displayed before reading a competitor's share as zero. — P-offer-competition-and-hijackers-16, 17. [official] Becoming the Featured Offer; European Commission IP/22/7777. [opinion] Jungle Scout (as the contradicted source). - **L39.** Live quantity comes from `fulfillmentAvailability` in `getListingsItem` and, for FBA, fulfillable units from `getInventorySummaries`, never from the `fulfillment_availability` attribute, which echoes the last submission. — P-inventory-and-fulfillment-1, 2, 4. [official] Listings Items API; Understanding Amazon listing status…; FBA Inventory API; fbaInventory.json; Inventory Buyability. - **L40.** A drop that coincides with the seller's offer being out of stock, inactive or not shippable is reported as an offer problem, and no copy change is proposed for it. — P-inventory-and-fulfillment-3; P-funnel-and-search-9. [official] My ASIN is unsearchable; Becoming the Featured Offer. - **L41.** On a SKU with both FBA and MFN records, an FBA stock-out is read as a switch to the MFN promise, and the MFN quantity and handling time are checked before calling the offer unbuyable. — P-inventory-and-fulfillment-16. [official] Manage Amazon Haul, Advanced Multiple-Offer, and Multiple-Fulfillment Use Cases; Modify handling time. - **L42.** A single view of the detail page is not evidence of the delivery promise or the Featured Offer, because both vary with the shopper's location and Prime status. — P-inventory-and-fulfillment-18. [official] Becoming the Featured Offer; Inventory Buyability. - **L43.** cx states no lasting search-rank penalty from a past stock-out, because Amazon's pages state only the immediate effects. — P-inventory-and-fulfillment-13. [official] My ASIN is unsearchable; Inventory performance. [opinion] Decoding the Amazon Algorithm in 2026 (as the unsupported claim). - **L44.** A seller running thin on stock hears Amazon's stated consequences and the opposing pressures (capacity limits, excess and aged-inventory costs, placement fees) as the seller's trade-off, not as neglect. — P-inventory-and-fulfillment-19. [official] Low-inventory-level fee; Inventory Buyability; FBA capacity limits; Inventory performance; FBA inbound placement service fee. - **L45.** A change in FBA fees after a re-measurement or a tier-table change is a fee event, not a listing effect or listing problem. — P-inventory-and-fulfillment-20. [official] Product size tiers; FBA fees reimbursement policy: Weight and dimensions; Fulfillment by Amazon (FBA) Reports; analytics_economics_2024_03_15.graphql. #### Vendor and hybrid - **L46.** Before any review or write, cx establishes whether the connection is a seller or a vendor account and records it. — P-vendor-and-hybrid-1. [official] Listings Items API; Authorize your Application to Access Vendor Groups; Selling Partner API Roles. - **L47.** cx never takes the account's own id from Keepa's Buy Box seller, and on a vendor connection asks for the vendor code that owns the SKU rather than inferring it. — P-vendor-and-hybrid-2. [official] Keepa Product.java; Offer.java; listingsItems_2021-08-01.json; Authorize your Application to Access Vendor Groups. - **L48.** Before reviewing or changing a listing, cx establishes whether Amazon retail offers the ASIN and who holds the Featured Offer, and records the answer. — P-vendor-and-hybrid-3. [official] Keepa Product.java, Stats.java, Offer.java; productPricing_2022-05-01.json; Product Pricing API. - **L49.** A missing BUYABLE status on a vendor listing is not a defect. — P-vendor-and-hybrid-5. [official] listingsItems_2021-08-01.json. - **L50.** On a vendor account the listing is placed in the funnel with vendor glance views and ordered units, with the note that these count only retail-featured views and that no sessions or search-query data exist for the vendor. — P-vendor-and-hybrid-17. [official] Analytics Reports; vendorTrafficReport.json; vendorSalesReport.json; Vendor Analytics Dataset Use Case Guide. - **L51.** cx proposes no change to Amazon's retail price, and before a seller-side price proposal on a hybrid ASIN names the retail price, the Featured Offer price and Pricing Health's reference to Amazon's own recent price. — P-vendor-and-hybrid-9, 10. [official] Responses to Questions for the Record (2019); Listings APIs FAQ; Pricing Health; productPricing_2022-05-01.json. #### Marketplaces and off-listing content - **L52.** cx reads the seller's stores from `marketplaceParticipations` and sends each store's calls to its regional endpoint. — P-marketplaces-and-languages-2. [official] SP-API Endpoints; Marketplace IDs. - **L53.** An EU listing review checks GPSR content (manufacturer contact, EU Responsible Person, warning and safety information), and safety images are proposed only in Amazon's dedicated safety-image slots. — P-marketplaces-and-languages-11, 12. [official] General Product Safety Regulation (GPSR); Submitting manufacturer, Responsible Person, and warning and safety information; Update: Support for GPSR compliance in Listings Items API. - **L54.** cx does not tell a seller UKCA is required in Great Britain, because Amazon's marking guidance says CE and UKCA are both accepted there. — P-marketplaces-and-languages-13. [official] EU and UK product marking guidance: CE and UKCA marking; Market Surveillance Regulation. - **L55.** A detail page may carry influencer videos the seller did not add, so a conversion change is not credited to the seller's own content when such video may have appeared. — P-off-listing-traffic-10. [official] Associates Central help, storefront videos; Earning Onsite Commissions. - **L56.** Store visits are not sessions or page views, and Store, Live, Attribution and ad-attributed sales are never added to Sales and Traffic units or compared day by day. — P-off-listing-traffic-3, 6; P-brand-story-and-store-11. [official] Business reports glossary; Metrics and traffic sources | Stores; Orders, sales and units sold | Stores; Brand Store insights metrics. #### What a good listing shows in Amazon's own answers - **L57.** A listing is called healthy when its status holds BUYABLE and DISCOVERABLE in each store, no issue is ERROR or carries a NOT_EXEMPT enforcement, and its Catalog view matches what was submitted or differs for a reason cx can name. — P-listing-content-and-health-3, 4, 6; P-listing-writes-7. [official] Understanding Amazon listing status; Listings APIs FAQ; Submit modular titles; Notification Type Values. - **L58.** A listing is read at `getListingsItem` on `{sellerId}/` with `summaries`, `attributes` and `issues`, and an `` is turned into the seller's SKUs through `searchListingsItems`, because an ASIN is not a listing address. — P-listing-content-and-health-1, 2. [official] listingsItems_2021-08-01.json; Retrieve details about a listing; Search for listings items. - **L59.** The Catalog view is reconciled and can rightly differ from the submission (brand-owner precedence, other contributors, a vendor's or Amazon's data, a title split, an AI rewrite), so an accepted value that does not show is reported as another contributor winning, never as a failed write, and the same value is not resubmitted. — P-listing-content-and-health-6; P-brand-protection-and-catalog-contribution-1; P-attributes-and-browse-nodes-10; P-title-7; P-bullet-points-9; P-description-1, 24; P-vendor-and-hybrid-6, 7. [official] Listings APIs FAQ; Suggest changes to your product detail page; Product detail page rules; Submit modular titles. [research] Using LLMs to improve Amazon product listings. - **L60.** Before proposing content, cx records the seller's relationship to the brand (brand owner, Brand Representative, reseller or none), and for a seller without a brand role says a content write is a suggestion that will most likely not display. — P-brand-protection-and-catalog-contribution-2, 3; P-title-16; P-search-terms-17. [official] Suggest changes to your product detail page; Guide to Brand Registry roles; Listings Items API Issues Troubleshooting. - **L61.** "Good" bullets in Amazon's own answers means no issue names `bullet_point`, the Catalog bullets equal the submitted ones, the count meets the minimum Amazon's page or schema states, and none repeats the title, description, Item Highlights or an attribute already displayed. — P-bullet-points-10, 16, 27, 28. [official] Product bullet point requirements; Listings APIs FAQ; Product detail page rules. - **L62.** A title audit answers each check from a source cx reads: the schema limit and its unit, the policy screen, information order, parent and child consistency, submitted against Catalog, issues with enforcements, pending AI proposals, and whether the seller's top SQP phrases survive in `item_name` or Item Highlights. — P-title-2, 21, 24, 35. [official] Product title requirements and guidelines; Submit modular titles; Listings Items API Issues Troubleshooting; Variation relationship FAQ. - **L63.** A title issue is read by its enforcement and exemption as well as its code, accepting old and new codes for the same rule, and `SEARCH_SUPPRESSED` is reported as hidden from search, not as a ranking loss. — P-title-31, 32; P-attributes-and-browse-nodes-19. [official] Listings Items API Issues Troubleshooting; Manage Listings Issues. - **L64.** Image health is checked store-wide from the suppressed listings report first and then `issues` per SKU, and a main image breaching the rules is an exposure risk as well as a click-through matter. — P-images-and-video-2, 22. [official] Report Type Values – Inventory Reports; listingsItems_2021-08-01.json; Image suppression; Manage Listings Issues. - **L65.** What shoppers see in images is read only from `getCatalogItem` images, because Amazon picks among contributors and may show another seller's image. — P-images-and-video-7, 18. [official] catalogItems_2022-04-01.json; Listings APIs FAQ; Product image guide. - **L66.** When attributes suddenly read as missing, cx first checks whether Amazon switched the product type, and compares the listing's product type with `searchDefinitionsProductTypes`' recommendation for its title. — P-attributes-and-browse-nodes-4, 5; P-bullet-points-23. [official] Get Product Type Definition recommendations; Building Listings Management Workflows Guide; Listings APIs FAQ; Warning code 18367. - **L67.** For an exposure drop with stable demand, classification (browse classification, product type, `item_type_keyword`, browse refinements from the browse tree report) is checked before any text is rewritten. — P-attributes-and-browse-nodes-7, 8; P-search-terms-25. [official] Browse Tree Reports; BrowseTreeReport.xsd; catalogItems_2022-04-01.json; Seller Forums staff guide "Why Isn't My Listing Showing in Search Results?". - **L68.** Attribute completeness is read from `issues`, including WARNING issues that name missing attributes. — P-attributes-and-browse-nodes-9. [official] Manage Product Listings with the Selling Partner API; listingsItems_2021-08-01.json. - **L69.** The description is read twice (as submitted and as shown), and where seller A+ is published its plain description is not what shoppers see, while vendor A+ sits under "From the Manufacturer". — P-description-1, 3, 4, 21, 22. [official] aplusContent_2020-11-01.json; A+ Premium Modules; Listings APIs FAQ. - **L70.** A+ is read through `searchContentPublishRecords` then `getContentDocument`, and eligibility is established before A+ work is proposed, with a submit 403 read as an eligibility answer. — P-a-plus-content-2, 3, 4; P-brand-story-and-store-2. [official] A+ Content API; Create, edit, and publish A+ content; A+ Content Troubleshooting Guide. - **L71.** A family is reviewed from both sides (the seller's `relationships` and Catalog `relationships`), each child is checked against the non-varying attributes and the same-function test, and a functionally different child is flagged as a policy risk with a separate page proposed. — P-variations-1, 4, 14. [official] Building Listings Management Workflows Guide; Product variations overview; Create and manage product variations. - **L72.** When a child's displayed rating drops with no new reviews, the review-sharing rule applied to the family's theme is the first hypothesis, and Amazon names no appeal for a review-sharing decision. — P-variations-6, 22. [official] Changes to review sharing across product variations; Guidelines for reviews shared across product variations; Check review sharing eligibility with Seller Assistant. - **L73.** cx never proposes adding a product to a family to borrow reviews, rating or rank, and reports a competitor doing so as a possible abuse case. — P-variations-5. [official] Product variations overview; FTC Bountiful case, as cited in variations.md. - **L74.** Featured Offer, sessions and rank are judged per child, because the Featured Offer is decided per variation and a child visit is not a parent visit. — P-variations-16. [official] Becoming the Featured Offer; Business reports glossary; Sales Rank FAQ. - **L75.** A BSR of 1 is a candidate for the Best Seller badge, not proof, and a New Release question is answered from the release and launch date attributes. — P-badges-and-placements-2, 3. [official] Sales Rank FAQ; catalogItems_2022-04-01.json. - **L76.** Prime and the delivery promise are offer and fulfilment facts read per offer from `getItemOffers`, never listing content. — P-badges-and-placements-6. [official] Guidelines for Seller Fulfilled Prime offers missing Prime branding; Manage your delivery time; productPricingV0.json. - **L77.** For badges no API carries, cx reports "seen on " or "unknown", never "absent", and the Black-Owned Business badge follows the featured offer, so on a shared ASIN its change is not attributed to the listing. — P-badges-and-placements-10, 15. [official] Keepa Product Object; Track and prevent product return badges; Black-Owned Business badge. - **L78.** The frequently-returned badge is the one negative badge content can address, it is relative to similar ASINs, and Keepa `returnRate` is the only API trace of it. — P-badges-and-placements-11, 12; P-returns-12. [official] Frequently returned item badge; Track and prevent product return badges; Keepa Product Object. - **L79.** Brand Store presence is read from Keepa's `brandStoreUrl` as a present-state read, and Store insights, followers and the Store quality rating come from Amazon Ads or the seller's console, not SP-API. — P-brand-story-and-store-10, 12. [official] Keepa Product.java; Brand Store insights metrics; Amazon Ads API release notes. #### Which file or report carries each figure - **L80.** Unit session percentage is units ordered over sessions from Sales and Traffic, never orders over sessions. — P-own-traffic-funnel-1; P-bullet-points-18; P-a-plus-content-12. [official] sellerSalesAndTrafficReport.json; analytics_salesAndTraffic_2024_04_24.graphql. - **L81.** The "conversion" ratios are kept apart by file and field (unit session percentage; the SCP report's `conversionRate`; Data Kiosk seller analytics `conversionRatePct` on search clicks; Data Kiosk `search.catalog.conversionRate` on impressions), and the deprecated `conversationRatePct` is never used. — P-own-traffic-funnel-3; P-effect-measurement-24; P-funnel-and-search-20; P-seller-central-only-tools-9; P-a-plus-content-12. [official] sellerSalesAndTrafficReport.json; sellingPartnerSearchCatalogPerformanceReport.json; analytics_sellerAnalytics_2025_03_31.graphql. - **L82.** CTR is clicks over impressions, never SQP's `totalClickRate`, which divides by search volume. — P-own-traffic-funnel-6; P-search-query-performance-3. [official] SQP schema; SCP schema. - **L83.** A per-ASIN daily series comes from Data Kiosk (`salesAndTrafficTrends` at child and day, or seller analytics by ASIN and day), with one Sales and Traffic report per day only as the fallback. — P-own-traffic-funnel-4. [official] analytics_salesAndTraffic_2024_04_24.graphql; Seller Analytics Dataset Use Case Guide. - **L84.** Every ratio is stated with its numerator, its denominator (page views, glance views, sessions, impressions or clicks) and the file it came from. — P-own-traffic-funnel-2; P-funnel-and-search-20; P-effect-measurement-24. [official] sellerSalesAndTrafficReport.json; Business reports glossary. [research] A Dirty Dozen (KDD 2017). - **L85.** Featured Offer share names its source, because Business Reports' percentage is over page views and Data Kiosk's `featureOfferPct` over glance views, and the two are never mixed. — P-price-position-30; P-offer-competition-and-hijackers-2, 3; P-funnel-and-search-21. [official] Becoming the Featured Offer; Business reports glossary; analytics_sellerAnalytics_2025_03_31.graphql. - **L86.** Sales and Traffic `averageOfferCount` is the seller's own listing count by date, not the number of sellers on an ASIN. — P-offer-competition-and-hijackers-14. [official] sellerSalesAndTrafficReport.json. - **L87.** A return rate is named by kind (overall, fee rate, badge rate, NCX), and one kind is never compared with another. — P-returns-3, 18. [official] Returns and Recovery Insights; Voice of the Customer; Fulfillment by Amazon (FBA) Reports. - **L88.** cx never says it read a product's reviews: it reads review topics from the Customer Feedback API and rating counts from Data Kiosk, per child against its parent and browse node, and never presents topics as Amazon's "Customers say" text. — P-reviews-and-qa-1, 2, 3, 4; P-ai-shopping-surfaces-14. [official] customerFeedback_2024-06-01.json; analytics_sellerAnalytics_2025_03_31.graphql; How Amazon continues to improve the customer reviews experience with generative AI. - **L89.** Search terms are read only from the seller's own `getListingsItem` attributes, because Amazon treats them as seller-specific and the Catalog does not return them, and cx claims no knowledge of a competitor's search terms. — P-listing-content-and-health-7; P-search-terms-9, 16, 23. [official] SP-API Release Notes (`generic_keyword` change); Listings APIs FAQ; Keepa Product.java. - **L90.** No report shows assistant-origin traffic, so cx never claims one does. — P-ai-shopping-surfaces-5. [official] Business reports glossary; Search Query Performance dashboard; sellingPartnerSearchQueryPerformanceReport.json; analytics_sellerAnalytics_2025_03_31.graphql. - **L91.** Where an API covers the same facts (suppression, detail-page sales and traffic, review topics, the seller's own content), cx reads the API and does not ask for a paste. — P-seller-central-only-tools-2; P-own-traffic-funnel-20. [official] Inventory Reports; Analytics Reports; Customer Feedback API. #### Keepa fields for listing content - **L92.** Keepa and Amazon's APIs keep no history of listing text, so each competitor read is kept as cx's own dated copy and content changes are found by comparing two copies. — P-competitor-set-7, 8; P-triggers-6; P-description-14; P-a-plus-content-14. [official] Keepa Product Object; Keepa Product.java; catalogItems_2022-04-01.json. - **L93.** A title diff compares `title` together with `itemHighlights`, `parentTitle` and `parentHighlights`, and an over-long title split between them is not read as the competitor's own choice. — P-competitor-set-9, 10; P-title-17. [official] Submit modular titles; Keepa Product Object. - **L94.** Bullets are diffed from `features`, description and `aPlus` from snapshots, and images and videos from `images` filenames and `videos` requested through raw `/product`, while on the seller's own brand ASINs `BRANDED_ITEM_CONTENT_CHANGE` says only that some content changed. — P-bullet-points-15; P-description-4, 13, 14; P-a-plus-content-14, 16; P-images-and-video-10. [official] Keepa Product.java; Keepa Product Request; Notification Type Values; BrandedItemContentChangeNotification.json. - **L95.** `monthlySold` is Amazon's bracketed "bought in past month" count across all sellers, covers the family when shown with a price range, and its absence is not zero sales. — P-badges-and-placements-4; P-competitor-set-15. [official] Sales Trend FAQ; Keepa Product Object. - **L96.** `listedSince`, `parentAsinHistory`, `brandStoreUrl`, `availabilityAmazon` and `csv[17]` are read for listing age, family moves, store presence, Amazon's retail offer and the rating count, with freshness from `lastRatingUpdate` and the variation-specific rating count known to be frozen. — P-competitor-discovery-6, 15; P-variations-9; P-brand-story-and-store-10; P-vendor-and-hybrid-3; P-reviews-and-qa-16, 19, 20. [official] Keepa Product Object; Product.java; Stats.java. - **L97.** Keepa's `productType` and `type` are Keepa's own fields, not Amazon's product type, and classification is compared through `itemTypeKeyword` and `categoryTree`. — P-attributes-and-browse-nodes-12. [official] Keepa Product Object. - **L98.** A competitor's stock-out history is read from Keepa out-of-stock percentages and Buy Box availability, which describe whether any offer of a price type existed, not the competitor's own stock. — P-inventory-and-fulfillment-11. [official] Keepa Stats.java; fbaInventory.json. [research] Structural Estimation of the Effect of Out-of-Stocks. - **L99.** Keepa timestamps are converted with Keepa's documented offset, and a price series that straddles Keepa's change of price-slot definition is flagged. — P-competitor-set-12, 18. [official] Keepa Product Object; KeepaTime.java. - **L100.** Keepa's `/deal` endpoint is not evidence of an Amazon promotion; deal badges come from the product's `deals` field. — P-seasonality-and-events-8. [official] DealRequest.java; Keepa Product Object. ### references/writing.md #### The three facts - **W1.** The product-type schema that `getDefinitionsProductType` returns is what Amazon validates a write against, limits included, so every limit, required attribute, enum, `editable` flag and selector comes from it, with the unit its keyword names (characters for `maxLength`, UTF-8 bytes for `maxUtf8ByteLength`). — P-listing-writes-1, 2; P-title-1; P-attributes-and-browse-nodes-1, 2; P-bullet-points-2, 3; P-description-2; P-search-terms-2; P-help-page-access-6. [official] Product Type Definitions API; Manage Listings Issues; Listings Items API Issues Troubleshooting; Amazon Product Type Definitions Meta-Schema (v1). - **W2.** `VALIDATION_PREVIEW` runs Amazon's checks without persisting anything and returns the issues, so it changes nothing on the listing, and it is used for a few SKUs at a time, not as a bulk pipeline. — P-listing-writes-5; P-title-5, 6; P-bullet-points-5; P-listing-content-and-health-10; P-description-6. [official] Listings APIs FAQ; Manage Listings Issues; Preview errors before partially updating a listing. - **W3.** A title over the limit may be accepted and then split by Amazon across `item_name` and `title_differentiation` or rewritten in the Catalog, so a title proposal is a pair inside the schema's limits. — P-listing-writes-13; P-listing-content-and-health-11; P-title-3, 4, 7. [official] Submit modular titles; Listings APIs FAQ; Product title requirements and guidelines. #### The write path and its shape - **W4.** A PATCH body is validated against the schema fetched with `requirementsEnforced=NOT_ENFORCED`, and a full put against `ENFORCED`. — P-attributes-and-browse-nodes-2; P-listing-writes-3. [official] definitionsProductTypes_2020-09-01.json; Partially update a listing; Building Listings Management Workflows Guide. - **W5.** Values are written in Amazon's shape: an array of objects with `value` and the schema's selectors (`language_tag` in the schema's snake_case form taken from the target store, and `marketplace_id`), with images in their named locator attributes. — P-listing-writes-4; P-title-30; P-marketplaces-and-languages-7; P-images-and-video-4. [official] Retrieve details about a listing; Submit modular titles; Submit media attributes; definitionsProductTypes_2020-09-01.json. - **W6.** `ACCEPTED` means accepted for processing, never done, so cx re-reads `issues` and then the Catalog view before saying a change took, and confirms by read-back rather than by waiting a fixed time. — P-listing-writes-6, 7; P-listing-policies-13; P-images-and-video-8, 19; P-variations-10, 23; P-help-page-access-5. [official] Manage Listings Issues; Listings Items API; Listings APIs FAQ; Product image guide; Variation relationship FAQ. - **W7.** A content write goes one store per request in that store's language, after that store's schema, preview and issues are read, and a change proven in one store is never copied to another without that check. — P-listing-writes-22; P-marketplaces-and-languages-8, 9; P-title-23; P-bullet-points-25; P-description-25. [official] Submit modular titles; Listings APIs FAQ; Product title requirements and guidelines (store layers); Use search terms effectively (store layers). - **W8.** `patchListingsItem` replaces a whole top-level attribute with the full intended array, and `merge` is used only where Amazon documents it. — P-search-terms-10, 22; P-bullet-points-4; P-description-7. [official] listingsItems_2021-08-01.json (`PatchOperation`); Partially update a listing. - **W9.** A seller `putListingsItem` replaces product facts while a vendor put merges and cannot remove an attribute, so cx applies the write semantics of the account in use. — P-vendor-and-hybrid-4. [official] Building Listings Management Workflows Guide. - **W10.** A sandbox answer proves only a request's shape and route, never the content's validity. — P-listing-writes-11. [official] Selling Partner API Sandbox. - **W11.** Whether an attribute is editable is read from the schema before a change is proposed, and on issue 4005033, 4005011 or 4005039 cx stops writing that attribute. — P-brand-protection-and-catalog-contribution-4. [official] Amazon Product Type Definitions Meta-Schema (v1); Listings Items API Issues Troubleshooting. - **W12.** Lengths are counted with a shell command in the unit the schema names, and search terms are counted as UTF-8 bytes of the whole string with spaces (`printf %s "$t" | wc -c`), shown with every proposal. — P-title-5; P-search-terms-5, 20. [official] Amazon Product Type Definitions Meta-Schema (v1); Use search terms effectively. - **W13.** For Japanese text, title and Item Highlights lengths are counted in characters and search terms in UTF-8 bytes. — P-marketplaces-and-languages-10. [official] Product title requirements and guidelines (JP layer); Use search terms effectively (JP layer). - **W14.** The schema's write limit and Amazon Search's indexing limit are different limits, a `VALID` preview proves only the write limit, and Amazon says terms over the indexing limit are ignored as a whole. — P-search-terms-3, 4, 19, 21. [official] Keyword attributes explained; Use search terms effectively; Listings Items API Issues Troubleshooting. - **W15.** A term in the field is never presented as proof of indexing, because Amazon may decline to use any supplied keyword. — P-search-terms-12. [official] Use search terms effectively. - **W16.** Search terms are written on the child listings shoppers can find, never relying on the parent's. — P-search-terms-24. [official] My ASIN is unsearchable; Use search terms effectively. - **W17.** Channel codes, accepted fields and shipping-template keys come from the seller-specific schema, and a template is written by its `enum` key, never its display name. — P-inventory-and-fulfillment-15. [official] Manage Amazon Haul…; Tutorial: Retrieve Merchant Shipping Templates. - **W18.** Only `item_type_keyword` and browse-node values from the schema's `enum` are proposed, and `item_type_keyword` is kept apart from `item_type_name`. — P-attributes-and-browse-nodes-6, 21. [official] Get recommended browse nodes or item type keywords; Listings Items API Issues Troubleshooting; Keyword attributes explained. - **W19.** Allowed variation themes and their required attributes come from the schema at the listing's `parentageLevel`, deprecated enum values are never proposed, and a theme is never changed in place but rebuilt as a new family. — P-variations-2, 15; P-title-25. [official] Building Listings Management Workflows Guide; definitionsProductTypes_2020-09-01.json; Update to irrelevant or redundant variation theme removal; Variation relationship FAQ. - **W20.** Price-per-unit and size-normalised product types get `unit_count` and its type rather than a typed size, and apparel size sets come from the schema. — P-attributes-and-browse-nodes-14, 15, 24; P-variations-18. [official] Guidance for Complex Attributes; Shoe size standards; Apparel size standards. - **W21.** Compliance values (dangerous goods, battery facts, safety data sheet, Prop 65, CPSIA) are never inferred from copy, a Prop 65 warning goes in the compliance attributes on child SKUs only, and a title or brand change on an FBA product with an SDS is flagged for the SDS match rule. — P-attributes-and-browse-nodes-16, 17, 18; P-bullet-points-21. [official] attribute and compliance help pages as cited in attributes-and-browse-nodes.md; Product bullet point requirements. - **W22.** Gating is checked with `getListingsRestrictions` before a new offer, a category change or a product-type change. — P-attributes-and-browse-nodes-23. [official] Categories and products that require approval; Listings Restrictions API. - **W23.** A+ and Brand Story are read and written only through the A+ Content API, never through a Listings patch, and a draft is judged by `validateContentDocumentAsinRelations` errors and warnings, never by an HTTP 200. — P-a-plus-content-2, 5; P-brand-story-and-store-1, 4. [official] A+ Content API; Create, edit, and publish A+ content; A+ Content Troubleshooting Guide. - **W24.** A Brand Story write replaces the live story across its whole ASIN set, so the live publish record and full ASIN set are read first and the seller hears it. — P-brand-story-and-store-3. [official] A+ Content Troubleshooting Guide; Manage A+ content; aplusContent_2020-11-01.json. - **W25.** A new image is a new file at a new URL, a new main image shows the same product, and image writes wait until the seller's tree holds a URL Amazon can fetch. — P-images-and-video-5, 20. [official] Submit media attributes; Product image guide. - **W26.** An image showing a photorealistic AI-generated person is flagged for the `contains-synthetic-performer` tag before upload. — P-images-and-video-25; P-ai-shopping-surfaces-18. [official] Product image guide; How to tag media that contains an AI-generated person. - **W27.** Video, 3D models and Image-Manager-only image kinds are Seller Central tasks the seller does. — P-images-and-video-9, 24. [official] Add product images and video; SP-API Release Notes. #### Policy screens - **W28.** A claim about performance, composition, origin, health, safety, environment, certification, award or survey is flagged as a compliance change that needs the seller's proof, never offered as a search or conversion gain. — P-listing-policies-2, 3, 4, 16; P-bullet-points-7, 20; P-description-19. [official] Misleading and prohibited claims; General Listing Restrictions; Product bullet point requirements; Product detail page rules. - **W29.** No broad environmental words appear: only a specific, qualified claim backed by proof, with certification as the route to a sustainability badge. — P-listing-policies-5, 22; P-badges-and-placements-5. [official] Environmental Marketing Guidelines (16 CFR Part 260); General Listing Restrictions; Enroll products in our Climate Pledge Friendly program. - **W30.** "Made in USA" and similar origin claims need proof, the qualified forms are offered otherwise, and a conflict with the country-of-origin data is flagged. — P-listing-policies-21, 28. [official] 16 CFR Part 323; Complying with the Made in USA Standard. - **W31.** Copy carries no sales-rank or subjective commentary, no imitation badge, no Amazon badge in an image, and no "was" price or savings text. — P-listing-policies-6, 26; P-badges-and-placements-13. [official] Selling policies and seller code of conduct; Product detail page rules; Product image guide; 16 CFR § 233.1. - **W32.** Keyword repetition and stuffing are a policy risk, and a title draft is screened for prohibited characters, repetition, promotional phrases and barred content, with each hit listed. — P-listing-policies-9; P-title-33; P-title-21. [official] Selling policies and seller code of conduct; Product title requirements and guidelines; Product detail page rules. - **W33.** A brand's special characters belong in the Brand field, not the title, size and colour belong in child titles, not the parent's, and no fixed visible-character count is stated. — P-title-13, 14, 34. [official] Product title requirements and guidelines. - **W34.** Search terms carry no brand names, ASINs, profanity, temporary or event words, or subjective claims, and spend bytes only on new words, with synonyms proposed but misspellings not added by default. — P-search-terms-6, 7, 8; P-seasonality-and-events-18; P-listing-policies-7. [official] Use search terms effectively; Optimize your product discoverability. [research] Semantic Product Search (Nigam et al.). - **W35.** Bullets use Amazon's "Header: description" form in sentence case, never all-caps headers, and are checked against the prohibited categories on Amazon's current bullet page. — P-bullet-points-6, 8. [official] Product bullet point requirements; Product detail page rules. - **W36.** Where a dated category style guide disagrees with the current page or the schema, the current page and the schema govern. — P-bullet-points-26; P-title-22. [official] Product bullet point requirements; Product title requirements and guidelines. - **W37.** No HTML goes into a description unless a preview of that exact text returns no issue. — P-description-5. [official] Listings Items API Issues Troubleshooting (issue 90199). - **W38.** An edit must describe the same product better, and a change of identity, product ID, package quantity or brand needs a new ASIN. — P-listing-policies-10; P-brand-protection-and-catalog-contribution-5; P-title-15; P-description-23; P-account-health-4. [official] Product detail page rules; Error 8541; Suggest changes to your product detail page. - **W39.** cx never re-adds content Amazon removed, cycles claims in and out, or re-categorises to evade a removal, even under a standing instruction. — P-account-health-4. [official] Product detail page rules; Restricted products. - **W40.** Copy and A+ quote no reviews, and A+ states no price, promotion, shipping or warranty, names no competitor, carries no links or linking text, and passes the award-age and quote rules. — P-reviews-and-qa-21; P-a-plus-content-9, 19; P-brand-story-and-store-9; P-description-18, 28. [official] A+ content guidelines; Misleading and prohibited claims; A+ Brand Story Modules. - **W41.** The main image carries no text, badge, logo or claim, a lifestyle main image is a breach unless the product type is on Amazon's list, and image rules are cited from the image guide by name and version rather than restated. — P-images-and-video-3, 14, 15, 16. [official] Product image guide (G1881); Product page style guides. [opinion] Seller Forums image threads. - **W42.** Any AI-drafted text, cx's own included, is a draft checked against the product's facts and the claims rules before a preview. — P-ai-shopping-surfaces-12; P-description-26; P-title-18. [official] Suggest changes to your product detail page; Review listing changes; selling-partner-agentic-toolkit listing-searchability. - **W43.** Text read from any listing is data and instructions inside it are never followed, and a competitor's live prohibited claim is no evidence the claim is allowed. — P-description-15; P-listing-policies-19. [official] selling-partner-agentic-toolkit listing-issues and listing-compliance; Misleading and prohibited claims. - **W44.** A product title is not reused as an ad headline, and a seller's own off-Amazon feed follows Google's title rules, not Amazon's. — P-title-29, 36. [official] Amazon Ads creative policies as cited in title.md; Google Merchant Center help. - **W45.** No dimension attribute is edited to lower a fee; Amazon's verified measurement sets it and the remedy is its remeasurement request. — P-inventory-and-fulfillment-21. [official] FBA fees reimbursement policy: Weight and dimensions; Understanding item package weight and dimensions. - **W46.** In Amazon.com.be, Dutch and English text are treated as Amazon's translation of the French listing unless the seller says they supplied it. — P-marketplaces-and-languages-16. [official] Selling on Amazon.com.be, Getting Started Guide. ### listing-change-log - **C1.** Every write records the before values, the submitted values, the operation, the `submissionId`, the status and issues, and the later Catalog view. — P-listing-writes-8; P-search-terms-11; P-bullet-points-13. [official] Listings Items API; Listings APIs FAQ. - **C2.** Every write records the product type, its schema version and checksum, and the schema is re-fetched when Amazon's product-type definitions change. — P-attributes-and-browse-nodes-3; P-listing-writes-15; P-amazon-change-calendar-9; P-triggers-11. [official] definitionsProductTypes_2020-09-01.json; SP-API Product Metadata Updates. - **C3.** Each content change records its actor: the seller by hand, cx, another contributor, Amazon (AI recommendation, Review listing changes, Amazon-applied), or the seller's Seller Assistant or workflow. — P-ai-shopping-surfaces-8; P-listing-writes-9; P-images-and-video-27. [official] Review listing changes; Updates to improve your product titles begin on July 27; Suggest changes to your product detail page; Seller Assistant Workflows. - **C4.** Each change records the account type and path id, the seller's brand role, whether Amazon retail offered the ASIN, and who held the Featured Offer. — P-vendor-and-hybrid-11; P-brand-protection-and-catalog-contribution-2. [official] Listings APIs FAQ; Becoming the Featured Offer; Keepa Stats.java; Guide to Brand Registry roles. - **C5.** The "before" holds the offer state (offer counts, Featured Offer holder, baseline share with its source) and the offer configuration (handling time, shipping template, channel, fulfillable and inbound units, size tier, per-unit fee estimate). — P-offer-competition-and-hijackers-2; P-inventory-and-fulfillment-14. [official] productPricingV0.json; productPricing_2022-05-01.json; getInventorySummaries; Modify handling time; getMyFeesEstimateForSKU. - **C6.** A vendor ASIN's content "before" is recorded at change time, because the vendor dataset keeps no historical catalog attributes. — P-vendor-and-hybrid-12. [official] Vendor Analytics Dataset Use Case Guide. - **C7.** Before the write, the log pre-registers what the change will be judged on: the targeted stage, its metric and orders, their sources, the expected direction and range, the smallest move that would count, both windows by date, the dated control set, the expected lag, and what a cannot-tell verdict leads to. — P-effect-measurement-6, 39. [research] Sufficient trial size to inform clinical practice; Equivalence Tests; The preregistration revolution. - **C8.** The expected range for a content change starts small unless the seller's own pooled history says otherwise, and Amazon's "up to" figures are never logged as the expected effect. — P-effect-measurement-38, 2; P-description-17; P-brand-story-and-store-16; P-images-and-video-12. [official] Test listing content with Manage Your Experiments. [research] Seven Rules of Thumb. - **C9.** The words a change bets on are logged as SQP queries, so the verdict can read them and the queries the change did not bet on. — P-search-query-performance-9, 17; P-paid-keywords-7; P-bullet-points-19. [official] SQP schema; Search term reports [v3]. - **C10.** The log records fee estimate lines and their time, whether by SKU or ASIN, any coupon fees, and a fee estimate at any proposed price with the note that it is not guaranteed. — P-price-position-17, 18; P-inventory-and-fulfillment-22. [official] productFeesV0.json; Product Fees API; getMyFeesEstimateForSKU; getMyFeesEstimateForASIN. - **C11.** A proposed price that crosses a fulfillment-fee price band is logged as a fee change, with band edges read from Amazon's current rate card. — P-inventory-and-fulfillment-23. [official] 2026 US FBA fulfillment fee changes. - **C12.** A change that adds or removes a claim records the claim, the proof the seller named and the policy page that governs it. — P-listing-policies-16. [official] Misleading and prohibited claims; General Listing Restrictions. - **C13.** An A+ or Brand Story change records the content reference key, the full ASIN set, the submission time, each observed status and when it was first seen published, and publication, not submission, opens its window. — P-a-plus-content-4, 6, 7; P-brand-story-and-store-5. [official] A+ Content API; Create, edit, and publish A+ content; Content publishing status and content rejection; A+ Content Troubleshooting Guide. - **C14.** A description change records whether seller A+ was published at write time and at window end. — P-description-8. [official] A+ Premium Modules; aplusContent_2020-11-01.json. - **C15.** An image change is logged as applying to the ASIN in every store where it exists, notes any country-specific image set, and the seller hears this before deciding. — P-images-and-video-6, 21; P-marketplaces-and-languages-3. [official] Listings APIs FAQ. - **C16.** Every change, read and effect series names exactly one `marketplace_id`, and effects are never summed across stores. — P-marketplaces-and-languages-1. [official] Marketplace IDs; sellerSalesAndTrafficReport.json. - **C17.** Before a price, sale or offer change, cx asks whether a Build International Listings connection includes the store, and tells the seller Amazon does not say whether a hand edit in a target store survives later syncs. — P-marketplaces-and-languages-4, 6. [official] Build International Listings; Build International Listings price rules; Global Offer Manager. - **C18.** The ad bidding strategy for the window and the seller's decision on holding bids are recorded at baseline, and a title change on an advertised ASIN notes that ad creative shows the title. — P-paid-keywords-16, 18; P-title-27. [official] Sponsored Products 3.0 OpenAPI; Sponsored Products rule-based bidding overview. - **C19.** A family change records the whole family from both sides, and deleting or moving a parent or child or changing a theme gets the seller's explicit word even under a standing instruction. — P-variations-1, 3, 11. [official] Building Listings Management Workflows Guide; listingsItems_2021-08-01.json; catalogItems_2022-04-01.json. - **C20.** A delete is logged as irreversible with what it takes with it, and no inactive or unfamiliar SKU is deleted, closed or "cleaned up" without first reading why it is inactive, its FBA units and their stranded reason. — P-listing-writes-16, 17; P-inventory-and-fulfillment-5. [official] Delete a listing; Fix stranded inventory; Stranded inventory required removals. - **C21.** Several changes made together are logged as one combined change, and the verdict is about the combination. — P-effect-measurement-5. [official] Save time by running simultaneous experiments on brand listing content. [research] Controlled experiments on the web. - **C22.** The seller's own promotions, deals and coupons in either window are logged from the Promotions API, joined to the performance reports through `promotionsApiMappingId`, and creating or editing them is a Seller Central action because the API is read-only. — P-seasonality-and-events-6, 21, 22; P-price-position-8. [official] Promotions API; promotions_2025-12-01.json; Performance Reports. - **C23.** A pending AI title recommendation is logged as a scheduled Amazon-side change. — P-seasonality-and-events-19; P-seller-central-only-tools-8. [official] Updates to improve your product titles begin on July 27; Review Listing Changes Tool guide. - **C24.** With a deal in UPCOMING status on the ASIN or its parent, cx proposes no price edit, listing closure or new child, and records any active deal in the "before" of a family change. — P-seasonality-and-events-17; P-variations-20. [official] Amazon deals; Troubleshoot suppressed and active deals; Promotions API. - **C25.** Every change file records the listing's coupon, deal, Prime-exclusive, strike-through and Prime or shipping-speed state on the change date. — P-badges-and-placements-8. [official] Coupons; Amazon deals; Price discounts; productPricing_2022-05-01.json; Keepa Product Object. - **C26.** A Catalog `relationships` snapshot is kept per watched family, and a family change the seller did not write is explained from Amazon's documented causes in order. — P-variations-11, 13. [official] Notification Type Values; catalogItems_2022-04-01.json; Troubleshoot product variations. - **C27.** A+ relations are re-read after any variation-family change. — P-a-plus-content-21. [official] Create A+ content; Manage A+ content. ### effect-review #### Design - **E1.** Where the seller is a brand representative and the ASIN is eligible, a Manage Your Experiments test is proposed before any before/after reading, because randomisation is the stronger evidence. — P-effect-measurement-1, 3, 4; P-images-and-video-13; P-description-16, 27; P-bullet-points-14; P-brand-story-and-store-17; P-title-20; P-seller-central-only-tools-6. [official] Test listing content with Manage Your Experiments; Manage Your Experiments; Create a listing experiment. [research] Kohavi, Tang, Xu 2020. - **E2.** An MYE result is the content's effect on shoppers who reached it, not a forecast of exposure, its one-year projection is not restated, and a result spanning an event is reported as covering it. — P-effect-measurement-27; P-seasonality-and-events-20. [official] Manage Your Experiments; Experiment results; Tips for experiments. - **E3.** A running MYE test on the ASIN is a confounder that blocks a before/after verdict. — P-a-plus-content-13. [official] Create a listing experiment; Experiment results. - **E4.** The observation window opens when Amazon shows the change (`getListingsItem` holds it with no ERROR issue, then `getCatalogItem` shows it), A+ and Brand Story windows open at confirmed publication, and Amazon's stated search latency runs before any exposure reading. — P-effect-measurement-7, 8; P-title-9; P-images-and-video-19; P-bullet-points-17; P-search-terms-11; P-funnel-and-search-15; P-a-plus-content-6; P-brand-story-and-store-5. [official] Amazon search glossary; Listings APIs FAQ; Product title requirements and guidelines; Product image guide. - **E5.** Both windows are whole Sunday-to-Saturday weeks with matching weekdays, aligned to SQP's own periods, and requests are sent already aligned. — P-effect-measurement-9, 25; P-search-query-performance-5; P-seasonality-and-events-5. [official] Analytics Reports; SQP schema. [research] Controlled experiments on the web. - **E6.** No unfinalized day is read, a null is missing rather than zero, and the baseline is re-read when the verdict is written. — P-effect-measurement-10; P-own-traffic-funnel-9; P-triggers-9; P-report-lifecycle-16. [official] Seller Analytics Dataset Use Case Guide; Analytics Reports. - **E7.** The baseline spans the number of whole weeks effect-measurement.md sets where history allows, and a trend in it is carried into the expected path. — P-effect-measurement-33. [research] Penfold and Zhang; Lopez Bernal et al.; Pretest with Caution. - **E8.** A title change is judged on query-level impressions and clicks plus sessions and orders, a main-image change on click rate and click share against impression share, and bullet, description and A+ changes on unit session percentage and orders. — P-title-11; P-images-and-video-11; P-bullet-points-18; P-description-29; P-a-plus-content-12. [official] Search Catalog Performance schema; SQP schema; sellerSalesAndTrafficReport.json. - **E9.** A search-terms change by a seller without the Brand Representative role is told in advance that no per-query exposure report exists, so the review may end in cannot tell. — P-search-terms-14, 27. [official] Search Query Performance dashboard. #### Controls - **E10.** The control's same-period movement is taken out first and the control named as imperfect, with query search volume as the cleaner control for demand. — P-effect-measurement-12; P-competitor-set-17, 26; P-funnel-and-search-12; P-seasonality-and-events-7. [research] Inferring causal impact (Brodersen et al. 2015); Multiple Randomization Designs. - **E11.** The control set is fixed before the window, and a competitor with its own change, deal or stock-out in the window is dropped from that window's controls. — P-competitor-set-6, 27; P-competitor-discovery-13; P-effect-measurement-37. [research] Brodersen et al.; Synthetic Control Methods for Comparative Case Studies. - **E12.** No subtraction is made across an Amazon change rolled out on different dates for different sellers, categories or stores, and ASINs in the changed ASIN's family are never controls. — P-amazon-change-calendar-3; P-effect-measurement-34. [official] Changes to review sharing across product variations; Updates to Featured Offer eligibility requirements. - **E13.** If the treated ASIN, the seller's untouched ASINs and the competitors all break in the same week, the verdict is cannot tell and a market or Amazon-side shock is named. — P-effect-measurement-37. [research] Comparative Politics and the Synthetic Control Method. #### Confounders - **E14.** For both windows the verdict reads and names own price, Featured Offer share, promotions and deals, stock, reviews and rating, ads, the competitors' price, deal and stock-out events, and major sales. — P-effect-measurement-14; P-price-position-13; P-offer-competition-and-hijackers-5. [official] Becoming the Featured Offer; Business reports glossary; productPricingV0.json. - **E15.** A window that overlaps a major sale, its run-up or its ramp-down gets cannot tell with the event named, where event dates are read at run time from each store's own announcements and "a major sale" is defined by evidence such as Keepa `deals`, not a remembered list. — P-effect-measurement-15; P-seasonality-and-events-1, 2, 3, 14, 28; P-funnel-and-search-32. [official] Prime Big Deal Days announcement pages; Keepa Product Object. [research] Controlled experiments on the web; SPADE. - **E16.** No post-event halo or dip is assumed: the seller's own orders around past events set the exclusion, Subscribe & Save orders are part of the event's tail, and an event rank or `monthlySold` jump is event demand. — P-seasonality-and-events-9, 12, 30. [official] Subscribe & Save; Keepa Product Object. [research] SPADE. [opinion] Perpetua. - **E17.** A year-over-year comparison sets an event-aligned reading beside the calendar-aligned one when an event moved, and short windows hold the same part of the month. — P-seasonality-and-events-23, 24. [official] Seller Analytics Dataset Use Case Guide. [research] Stephens 2003; Hastings & Washington 2010. - **E18.** Returns- and review-targeted verdicts near an event wait until Amazon's extended return date and review-request window have passed. — P-seasonality-and-events-13, 25. [official] Amazon's holiday return policy; Set up the Solicitations API. - **E19.** Before any judgment, the Amazon-side changes dated inside both windows are listed with the metric each touches. — P-amazon-change-calendar-1, 2, 5, 6, 7, 8, 10, 11, 12, 13. [official] SP-API Release Notes; SP-API Deprecation Schedule; Updates to improve your product titles begin on July 27; Guidelines for reviews shared across product variations. - **E20.** A content difference cx did not write, found in the dated Catalog copies or a `BRANDED_ITEM_CONTENT_CHANGE` event, makes the verdict cannot tell with the attribute and date named. — P-effect-measurement-26; P-brand-protection-and-catalog-contribution-9, 10; P-title-10; P-images-and-video-27; P-description-13. [official] Notification Type Values; catalogItems_2022-04-01.json; BrandedItemContentChangeNotification.json. - **E21.** An A+, Brand Story or Store change inside a window is a confounder for every ASIN it touches, including siblings used for comparison. — P-description-9; P-brand-story-and-store-7, 14; P-a-plus-content-18. [official] A+ Content (G202102930); A+ Brand Story Modules; aplusContent_2020-11-01.json. - **E22.** Ad spend for both windows comes from the Data Kiosk Economics `ads` charges by ASIN and day when `createQuery` for that dataset succeeds on the connection, and otherwise ad spend is named as unread and a material spend change makes an exposure verdict cannot tell. — P-effect-measurement-23; P-paid-keywords-9; P-seasonality-and-events-27. [official] analytics_economics_2024_03_15.graphql; Data Kiosk API; Correction: Data Kiosk role requirements. - **E23.** Only ad changes a person made are confounders, while dynamic or rule-based bid responses are consequences of the change and are not subtracted. — P-paid-keywords-17, 18, 25. [official] Sponsored Products 3.0 OpenAPI; Guide to dynamic bidding - up and down. [research] Cinelli, Forney & Pearl, A Crash Course in Good and Bad Controls. - **E24.** SQP and Search Catalog Performance include Sponsored Products traffic from the search results page, so an ads change on the same queries moves the "search" figures too. — P-effect-measurement-28; P-paid-keywords-11, 19; P-funnel-and-search-11, 19; P-own-traffic-funnel-22; P-search-terms-26. [official] Search Query Performance dashboard (G8J4CB5ZBF3NX7TP v5); Search Catalog Performance Dashboard. - **E25.** Sponsored prompts are auto-enrolled and Economics `ads` cannot separate prompt spend, and a change aimed at assistant answers is judged only on blended detail-page metrics and orders. — P-ai-shopping-surfaces-6, 7. [official] Sponsored Products prompts and Sponsored Brands prompts; Prompt Ad Extension reports; analytics_economics_2024_03_15.graphql. - **E26.** A stock-out, a lost Featured Offer, a new seller, a promise change (handling time, template, SSA, channel) or a lost Seller Fulfilled Prime badge in either window makes the verdict cannot tell unless the change is shown not to touch the targeted stage. — P-inventory-and-fulfillment-9, 17; P-funnel-and-search-9, 33; P-offer-competition-and-hijackers-5; P-triggers-20. [official] My ASIN is unsearchable; Becoming the Featured Offer; Modify handling time; Seller Fulfilled Prime program policy. [research] Measuring and Mitigating the Costs of Stockouts. - **E27.** The seller's daily stock for both windows is reconstructed from the Inventory Ledger, not today's snapshot, and a stock event is dated from a settled read, not the first signal. — P-inventory-and-fulfillment-8, 12. [official] Fulfillment by Amazon (FBA) Reports; Notification Type Values; getInventorySummaries. - **E28.** Before crediting a click-through or conversion change to content, cx checks whether the delivery-message mix moved between windows in Search Catalog Performance. — P-inventory-and-fulfillment-10; P-badges-and-placements-7. [official] sellingPartnerSearchCatalogPerformanceReport.json; sellingPartnerSearchQueryPerformanceReport.json. - **E29.** An account event in either window (offer deactivated, listing removed or suppressed, an account status change) gets cannot tell. — P-account-health-8. [official] Review Account Health updates for FBM offers; Notification Type Values; Account health rating. - **E30.** Variation-family events and review-sharing steps are Amazon-side changes to the rating and page, and a window holding a family event gets no parent-level verdict. — P-variations-8, 12; P-reviews-and-qa-9, 17; P-amazon-change-calendar-6. [official] Changes to review sharing across product variations; Troubleshoot product variations; Analytics Reports. - **E31.** A product-type switch, a browse-node change or a browse-tree change in the window is a confounder for exposure. — P-attributes-and-browse-nodes-11. [official] Listings APIs FAQ; Browse Tree Reports; Notification Type Values. - **E32.** A badge gained or lost in either window by the listing or a competitor in the set is a named confounder. — P-badges-and-placements-9. [official] Amazon deals; Coupons; Frequently returned item badge; Sales Trend FAQ. - **E33.** A change in Amazon retail's presence, stock or Featured Offer share is a confounder, a hybrid seller's units are not read alone as the conversion outcome, and vendor glance views are never compared with seller sessions. — P-vendor-and-hybrid-13, 14, 15. [official] Becoming the Featured Offer; Sales and Traffic schema; Analytics Reports; analytics_vendorAnalytics_2024_09_30.graphql. - **E34.** Off-listing campaigns (Store, Live, Attribution, Brand Referral Bonus, tailored promotions, external discounts) in a window are read as campaign-driven before listing-driven, and their sales are called "attributed", never "incremental". — P-off-listing-traffic-5, 7, 9, 11, 14; P-brand-story-and-store-13. [official] Amazon Attribution API how-to; Brand Tailored promotions; Reporting columns (version 3); Selling policies and seller code of conduct. [research] Gordon et al. 2019; Lewis & Rao 2015; Blake, Nosko & Tadelis 2015. - **E35.** Sponsored Brands or display attributed conversions are not compared across Amazon's view-attribution change without naming it, and Amazon's Store or Live visitor figures are observational claims, never an expected effect. — P-off-listing-traffic-8, 13. [official] View Attribution Updates for Amazon Store ads; Brand Stores | Amazon Ads; Amazon Live | Amazon Ads. - **E36.** BIL-linked price moves are confounders in target stores, image changes are confounders in every store the ASIN is in, and EU returns are counted over at least the statutory withdrawal period. — P-marketplaces-and-languages-3, 5, 15. [official] Build International Listings price rules; Listings APIs FAQ; About Our Returns Policies (amazon.de). - **E37.** An EU discount and its effect are measured against the store's own lowest recent price over the reference period that store's page states. — P-marketplaces-and-languages-14. [official] Price discounts (DE layer); Reference Pricing and Savings (amazon.fr, amazon.co.uk). - **E38.** A shift in the mix of B2B, app or other segments is checked before a pooled ratio is read. — P-effect-measurement-42; P-own-traffic-funnel-8; P-description-29. [official] sellerSalesAndTrafficReport.json. [research] A Dirty Dozen. - **E39.** Amazon's assistant usage figures are never used as the expected effect of a listing change, and a price move in the baseline is read knowing the assistant shows lowest-price history. — P-ai-shopping-surfaces-15, 16. [official] How Amazon is using generative and agentic AI to transform the shopping experience; Alexa for shopping. #### Verdicts - **E40.** There are three verdicts: held (after the control is taken out, the metric moved the expected way by at least the logged smallest move, orders did not fall, nothing else explains it), not held (the window could have seen that move and did not, or it moved the wrong way), and cannot tell (everything else). — P-effect-measurement-18. [research] Controlled experiments on the web; Seven Rules of Thumb. - **E41.** Every verdict names the exact field it read, uses the same field from the same file in both windows, and states the counts behind it. — P-effect-measurement-17, 24, 44. [official] analytics_sellerAnalytics_2025_03_31.graphql. [research] Seven Rules of Thumb; NIST Poisson Distribution. - **E42.** Noise is not computed as if each session were an independent shopper: binomial and Poisson intervals are floors, and an overdispersed count model fitted to the ASIN's history sets the noise where history allows. — P-effect-measurement-29; P-triggers-30. [research] NIST/SEMATECH e-Handbook. [opinion] Minitab. - **E43.** One primary metric is fixed in advance, and a single "significant" query or segment among many is reported as chance unless it survives false-discovery-rate control or a pooled estimate. — P-effect-measurement-31. [research] Benjamini and Hochberg 1995. - **E44.** Every change has two guardrails, the ASIN's whole-catalog search funnel and its previous top queries the change did not bet on, and a material fall in either blocks "held". — P-effect-measurement-30. [official] Search Catalog Performance schema; analytics_sellerAnalytics_2025_03_31.graphql. - **E45.** Before the first verdict on an ASIN, a placebo check runs the same rule on past windows with no change, and if it often returns held or not held, verdicts there are cannot tell and MYE is proposed. — P-effect-measurement-32. [research] placebo and synthetic-control sources in effect-measurement.md. - **E46.** A large win from a small wording change is checked for a data or confounder explanation first, and a listing picked because it dropped is told that some recovery was expected anyway. — P-effect-measurement-16, 19. [research] Seven Rules of Thumb; Regression to the mean. - **E47.** The expected lifts of several changes are never added into one forecast. — P-effect-measurement-22. [research] Seven Rules of Thumb. - **E48.** Every held or not held is provisional and re-read at a logged durability window, and a changed reading is added as a dated note that never overwrites the first. — P-effect-measurement-35. [research] Focusing on the Long-term; The preregistration revolution. - **E49.** A revert is a new change with its own baseline, window and verdict. — P-effect-measurement-36. [research] Design and Analysis of Switchback Experiments. - **E50.** Once enough verdicts of one change type exist, a pooled estimate is shown beside each new one. — P-effect-measurement-40. [research] Bayesian Data Analysis, ch. 5. - **E51.** A week-by-week manual alternation counts as randomised only if its order was randomised in advance. — P-effect-measurement-41. [research] Design and Analysis of Switchback Experiments. - **E52.** Returns are a guardrail read only for days older than the return lag Amazon's dataset guide states, and a sooner verdict says returns are not yet in, with returns read alongside conversion for a gallery change. — P-effect-measurement-43; P-returns-21; P-images-and-video-28. [official] analytics_sellerAnalytics_2025_03_31.graphql; Seller Analytics Dataset Use Case Guide. - **E53.** Cannibalisation is read from the seller's other ASINs' sessions and orders, never from Market Basket data, and a Brand Story or comparison-chart change reads the brand's other ASINs too. — P-effect-measurement-45; P-a-plus-content-18. [official] sellingPartnerMarketBasketAnalysisReport.json; A+ Brand Story Modules. ### scripts/report.sh - **S1.** A report is fetched in Amazon's three asks, `createReport`, `getReport` on the `reportId` Amazon returned and `getReportDocument` on the `reportDocumentId` the record carries, and never on an id typed or remembered. — P-report-lifecycle-20, 23. [official] Request a report; Verify that report processing is complete; Retrieve a report; reports_2021-06-30.json. - **S2.** A report is finished only at `DONE`, `CANCELLED` or `FATAL`, and one still queued after the single wait is told to the seller as "still queued since ", never as a failure. — P-report-lifecycle-1, 26. [official] Verify that report processing is complete; reports_2021-06-30.json; Analytics Reports. - **S3.** `CANCELLED` means no report from this request (an explicit cancel, no data, or a repeat inside the type's period) and is never zero sales. — P-report-lifecycle-2. [official] Verify that report processing is complete; Reports API FAQ. - **S4.** `FATAL` with a `reportDocumentId` is read as Amazon's error explanation, quoted, fixed once at the named parameter, and never parsed as data, and a vendor `FATAL` for a period not yet available means "too early". — P-report-lifecycle-3; P-vendor-and-hybrid-18. [official] Verify that report processing is complete; Analytics Reports. - **S5.** Before any `createReport`, cx looks for its own landed receipt and then `getReports` for the same type, store and range, weighs what it finds by its `createdTime`, and reads an unfinished `reportId` again on a later turn rather than creating another. — P-report-lifecycle-6, 26. [official] reports_2021-06-30.json; Analytics Reports ("Cache results locally"); Optimize Rate Limits for Application Workloads. - **S6.** Report type names, `reportOptions` and period rules come from door.md's entry, and when door.md has none cx says so and asks rather than probing names. — P-report-lifecycle-7, 22. [official] Analytics Reports; Inventory Reports; Reports API FAQ; Acceptable Use Policy. - **S7.** One Brand Analytics report is requested per period, from its first day to its last, because a range over several periods or off its boundaries is accepted and then ends `FATAL`. — P-report-lifecycle-8; P-search-query-performance-7. [official] Analytics Reports; SQP schema. - **S8.** For Sales and Traffic at week or month granularity, the expanded range Amazon will use is recorded, and a period ending in the future is marked partial. — P-report-lifecycle-9. [official] Analytics Reports. - **S9.** A week's Brand Analytics report is not requested before the availability Amazon states, read at run time. — P-search-query-performance-22; P-triggers-8. [official] Analytics Reports. - **S10.** A report type or dataset is requested only if its Availability line includes the connection's account type. — P-vendor-and-hybrid-16. [official] Analytics Reports; SP-API Vendor Use Cases; Reports API FAQ. - **S11.** `GET_MERCHANT_LISTINGS_DEFECT_DATA` is never requested, store-wide suppression is `GET_MERCHANTS_LISTINGS_FYP_REPORT`, and only current FBA report types are used. — P-listing-content-and-health-21; P-inventory-and-fulfillment-25; P-triggers-26. [official] Report Type Values; Inventory Reports; SP-API Deprecation Schedule; Fulfillment by Amazon (FBA) Reports. - **S12.** A merchant listings report may be a cached copy, so the seller is told and one store is asked per report. — P-report-lifecycle-18. [official] Inventory Reports. - **S13.** A role refusal (a 403 on a report type, a Data Kiosk authorization error) is read as "this connection lacks the role for this data", said to the seller with the role the refusal implies, and not retried under other names. — P-report-lifecycle-19; P-sp-api-access-6, 7, 8, 13; P-returns-22. [official] Reports API FAQ; Role Mappings for SP-API Operations; Selling Partner API Roles; Data Kiosk API. - **S14.** The script adds no rate limiter or retry of its own: a 429 lands in the receipt with Amazon's body, and cx stops, tells the seller and waits with back-off before one retry, never looping `createReport`. — P-report-lifecycle-10, 21; P-sp-api-access-12. [official] Usage Plans and Rate Limits; Optimize Rate Limits for Application Workloads; Resolve Common HTTP and Authorization Error Codes. - **S15.** Published rates are defaults only; the `x-amzn-RateLimit-Limit` header is the plan in force, and no fixed rate is quoted. — P-sp-api-access-11; P-competitor-set-29; P-report-lifecycle-10. [official] Usage Plans and Rate Limits; Listings Items API Rate Limits; Catalog Items API Rate Limits. - **S16.** A document link is fetched in the step it is issued and never printed, stored or reused, a `GZIP` document is decompressed, and a Data Kiosk document relies on `Content-Encoding`. — P-report-lifecycle-4, 5. [official] reports_2021-06-30.json; dataKiosk_2023-11-15.json; Building Data Kiosk workflows guide. - **S17.** Every landed document is kept as a file in the seller's tree (never `/tmp`) with its report or query id, type, store, data range and processing times, so a change log can cite it, and readers key on column names and strip a byte order mark. — P-report-lifecycle-17, 24, 25; P-triggers-6; P-competitor-set-7. [official] Report Type Values; Retrieve a report; Data Protection Policy; Reports API. - **S18.** A report schedule is created or replaced only on the seller's word after `getReportSchedules` is read, because a new schedule silently replaces an existing one. — P-report-lifecycle-11. [official] reports_2021-06-30.json; Schedule and retrieve reports. - **S19.** cx cancels only its own report while `IN_QUEUE`, and its own Data Kiosk query while `IN_QUEUE` or `IN_PROGRESS`. — P-report-lifecycle-12. [official] reports_2021-06-30.json; dataKiosk_2023-11-15.json. - **S20.** A Data Kiosk query is read with `getQuery` to a terminal status, `dataDocumentId` is data, `errorDocumentId` Amazon's explanation, and `DONE` with neither means no data for that range. — P-report-lifecycle-13. [official] Verify that query processing is complete; Data Kiosk API. - **S21.** A Data Kiosk result is complete only when no `pagination.nextToken` remains, each page being a new `createQuery` with that token, and an identical pending query is never resubmitted. — P-report-lifecycle-14, 15. [official] dataKiosk_2023-11-15.json; Data Kiosk API; Building Data Kiosk workflows guide. - **S22.** The dates Data Kiosk actually returned are recorded, and recent days are marked partial or null as the source states. — P-report-lifecycle-16; P-own-traffic-funnel-9; P-paid-keywords-6; P-triggers-9. [official] Seller Analytics Dataset Use Case Guide; Sponsored ads reporting FAQ (version 3). - **S23.** Weekly buckets from different sources are never lined up by label, so the Monday-start Replenishment weeks are read by day and other requests are sent aligned to Sunday–Saturday. — P-price-position-26; P-effect-measurement-25. [official] replenishment-2022-11-07.json; analytics_economics_2024_03_15.graphql; Analytics Reports. - **S24.** A bulk change goes as a `JSON_LISTINGS_FEED` through `createFeedDocument`, `createFeed`, `getFeed` to a terminal status and `getFeedDocument`, one store per feed. — P-listing-writes-19, 20, 21; P-seller-central-only-tools-11. [official] Submit a feed; listings-feed-schema-v2.json; Deprecation of Feeds API support for XML and Flat File Listings Feeds. - **S25.** The FBA Fee Preview report is read within the request limit the FBA Reports page states. — P-inventory-and-fulfillment-20. [official] Fulfillment by Amazon (FBA) Reports. ### The templates and the standing instruction /listing-watch is `@comchatx/cx-listing-watch`'s template from 0.3.0; T1–T16 stay here as the research it and the playbook's listing-review rest on. - **T1.** /listing-watch writes a standing instruction that states what to watch (the ASINs, the competitor set, the trigger families), how often, the seller's own thresholds, when to speak and when to stay quiet, and where to report, with no default threshold supplied. — P-triggers-13. [opinion] Jungle Scout, Amazon Product Alerts; Helium 10, Alerts; Google SRE ch. 6. - **T2.** Between summaries cx speaks only for a new signal the seller can act on (a suppression, a lost Featured Offer, a stock-out, a content change nobody on the seller's side made), and everything else goes into the scheduled summary. — P-triggers-14. [official] Manage Listings Issues. [opinion] Google SRE ch. 6. - **T3.** The standing instruction states a false-alarm budget for the whole watch, and each metric is charted against that ASIN's own history. — P-triggers-29. [research] NIST/SEMATECH e-Handbook §6.3; Benjamini and Hochberg 1995. - **T4.** Rates over many sessions are not charted with plain binomial limits, because those limits are too narrow when the rate itself drifts. — P-triggers-30. [research] NIST/SEMATECH e-Handbook §6.3.3.2. - **T5.** Each look's cadence matches its source's refresh: weekly Sunday to Saturday for Brand Analytics, and the readiness the dataset guide states for daily analytics. — P-triggers-8; P-seller-central-only-tools-3. [official] Schedule Data Kiosk Queries; Seller Analytics Dataset Use Case Guide; Analytics Reports. - **T6.** Amazon's push signals reach only an AWS destination, so a watch from the box pulls state on its cadence, a notification may arrive more than once and out of order, and `ANY_OFFER_CHANGED` "External" is not read as a new seller. — P-triggers-12, 21; P-offer-competition-and-hijackers-13; P-account-health-11; P-brand-protection-and-catalog-contribution-9. [official] Notification Type Values; notifications.json; Set up notifications using the Amazon EventBridge workflow; Set up notifications using the Amazon Simple Queue Service workflow. - **T7.** An account watch reads the seller-performance report and `getMarketplaceParticipations` on the seller's cadence. — P-account-health-11. [official] Performance Reports; sellers.json. - **T8.** A stock watch reads stock-out precursors (restock alerts, days of supply, below-minimum levels, capacity blocks) and stranded units as "stock risk", never as a listing problem, and warns before a stock-out lands in a planned test window. — P-inventory-and-fulfillment-6, 7. [official] Fulfillment by Amazon (FBA) Reports; Low-inventory-level fee; Inventory performance; FBA capacity limits. - **T9.** The competitor set is the seller's named rivals plus Amazon's `similarItems`, each confirmed by the seller before tracking. — P-triggers-15; P-competitor-set-3. [official] getCompetitiveSummary reference; SP-API Release Notes. - **T10.** The set is frozen per window, and a re-pick is a new, dated version of the set. — P-competitor-set-6; P-competitor-discovery-13. [research] Brodersen et al. - **T11.** Competitor data drawn from one seller's connection is used only for that seller. — P-competitor-set-20. [official] Acceptable Use Policy 4.4 and 4.6. - **T12.** Seller Central-only tools (MYE, RLC, View Change History, Enhance, Seller Assistant, VOC, A+ performance, Store insights) are not read by cx: they arrive as files the seller puts in the tree, and cx never claims to have read them itself. — P-seller-central-only-tools-1, 5, 6, 7; P-a-plus-content-17; P-brand-story-and-store-12. [official] Analytics Reports; Create a listing experiment; Seller Assistant help page. - **T13.** Each handoff is stored with the tool's name, the marketplace, the date range or as-of time shown, and when the seller captured it, and an Amazon-generated file is preferred to a screenshot and a screenshot to typed figures. — P-seller-central-only-tools-3, 4. [official] Review Listing Changes Tool guide; Business Reports FAQ; Inventory Reports. - **T14.** /listing-watch keeps content tests out of event windows and their run-ups, proposes a change before an event only when its whole observation window can close first, and tells the seller when the next clean window opens. — P-triggers-10, 27; P-seasonality-and-events-4, 15, 29. [official] Prime Big Deal Days announcement pages; Prime Day readiness playbook; Understand deal fees. - **T15.** An A+ change is not submitted where Amazon's stated review and publish time could land the publish inside an event. — P-seasonality-and-events-16. [official] Complete Guide to A+ Content for Brand Owners. [opinion] AMALYZE. - **T16.** During announced peak inbound periods, the published inbound cut-off dates are read from the current season's playbook as the last dates for event-ready FBA stock. — P-inventory-and-fulfillment-26. [official] Q4 2026 peak readiness playbook. - **T17.** Deal and coupon eligibility floors and event deadlines are stated only from Amazon's page by name and recorded version, or from the seller's Seller Central copy, with both floors (seller and product) given. — P-seasonality-and-events-11, 26; P-badges-and-placements-14. [official] Amazon deals; Coupon eligibility criteria; Price discounts; Prime Day | Sell on Amazon. - **T18.** A help page is cited by its title, canonical id, version and read date, read first from its capture under the office's capture folder (outside this repository), and cx does not depend on the undocumented content call at run time. — P-help-page-access-1, 2, 3, 7. [official] the captured pages under the office's capture folder (outside this repository) - **T19.** /listing-review starts from the trigger or the seller's question, runs the gates (L2 to L10), names the stage (L1), and only then reads the elements for that stage (L13). — P-funnel-and-search-1, 3; P-triggers-1, 2. [official] My ASIN is unsearchable; Manage Listings Issues. ### listing-create - **R1.** The stores a creation goes to are the ones the seller names, each checked in `getMarketplaceParticipations`, which gives that store's language and currency. — P-listing-create-1; P-sp-api-access-1. [official] sellers.json; Get the market participation of a seller. - **R2.** A catalog search by identifier sorts the product into an existing ASIN (offer-only listing, `PRODUCT` with `LISTING_OFFER_ONLY`, after its restrictions read) or a new item with product facts. — P-listing-create-2, 3, 22. [official] Building Listings Management Workflows Guide; Listings APIs FAQ. - **R3.** A new item's product type comes from `searchDefinitionsProductTypes`, and its schema is read separately for each store, with `requirements`, `sellerId` and `parentageLevel`. That schema's keywords decide what each value must look like. — P-listing-create-4, 5, 6. [official] definitionsProductTypes_2020-09-01.json; Retrieve a Product Type Definition; Meta-Schema (v1). - **R4.** Identity is a GTIN the seller owns or is licensed to use, or an approved exemption with the brand spelled exactly as approved. A Brand Registry brand takes the brand's own account or one the brand authorised. — P-listing-create-9, 10, 11. [official] Product detail page rules (G200390640 v21); Listings APIs FAQ; How to create Amazon product listings. - **R5.** A creation is one put per store, in that store's language and currency. It carries images as locations Amazon fetches (a new object key for new content in private S3), the fulfilment record for MFN or FBA, and the compliance attributes the store's schema names, all filled from the seller's own data. — P-listing-create-7, 8, 12, 13, 14, 15. [official] listingsItems_2021-08-01.json; Submit Listings data; Submit media attributes; Manage purchasable offer; Manage Amazon Haul … Multiple-Fulfillment Use Cases. - **R6.** A preview is the same put with `mode=VALIDATION_PREVIEW`. It persists nothing and answers `VALID` or `INVALID`, and its issues name any required attribute the data left empty. — P-listing-create-17; P-listing-writes-5. [official] Preview errors before creating or fully updating a listing; Listings Items API Issues Troubleshooting. - **R7.** A family is written parent first, then each child with the relationship and the parent's theme. — P-listing-create-16. [official] Building Listings Management Workflows Guide; Submit Listings data. - **R8.** `ACCEPTED` starts processing. The creation is done when a read-back shows an ASIN, `BUYABLE` and `DISCOVERABLE`, and no `ERROR` issue; an empty summary is, in the FAQ's words, a submission that failed and did not match an ASIN, and Amazon states no time for any of this. — P-listing-create-18, 19. [official] Listings APIs FAQ; listingsItems_2021-08-01.json. - **R9.** A static sandbox answer proves only the request's path, signing and shape. Build International Listings is a Seller Central tool, so it is not part of this path. A creation is logged as a change with an empty before. — P-listing-create-20, 21, 23. [official] Selling Partner API Sandbox; Build International Listings (G202121570 v14). - **R10.** Each conflict issue is answered as its own text says: `4005011` change the value or contact Amazon, `4005033` revert and resubmit, `101165` make the identifiers unique and resubmit, with support where the catalog itself is wrong. — P-listing-create-24. [official] Listings Items API Issues Troubleshooting. ## Where the files disagreed, and what won - **Help pages called unreachable.** Files that called a page unreachable or behind sign-in were stale wherever a capture exists under the office's capture folder (outside this repository); the capture won (P-help-page-access-3). - **Latency figures.** Files attributed the title display window and the search latency figure to forums or SellerEngine; they are on Amazon's own pages (Product title requirements and guidelines; Amazon search glossary; My ASIN is unsearchable). Each page scopes its own figure, so read-back won over any fixed wait (W6). - **Search-terms page.** G23501 is the live US page; G201047930 serves no topic. The 2022 UK "spaces are counted" wording is contradicted by the current UK page. - **Is the description indexed?** description.md's "no Amazon page says so" lost to Optimize your product discoverability (G10471) and Classify your products (G1661), which name it as matched content (L11). - **SQP and Sponsored Products.** The Search Query Performance dashboard page says SQP includes Sponsored Products from the search results page and excludes widgets; older lines in paid-keywords.md and search-rank-and-volume.md were corrected (E24). - **SQP "MONTH" retention.** Analytics Reports gives a figure under "Maximum reporting periods" that its own text does not explain; the first real request settles it. - **Featured Offer eligibility.** triggers.md's eligibility test and the Amazon Ads guide's "eligible" wording were stale against Featured Offer eligibility (G200418100 v23), which won (L15). - **Featured Offer denominators.** Business Reports uses page views and Data Kiosk `featureOfferPct` glance views; neither won, and the two are never mixed (L85). - **"Conversion rate".** The SCP report's `conversionRate` (purchases over clicks) and Data Kiosk's `search.catalog.conversionRate` (purchases over impressions) share a name across two schemas, `conversionRatePct` is search clicks only, and `conversationRatePct` is deprecated; the file and field won over the name (L81). - **Data Kiosk roles.** The Data Kiosk API page lists Brand Analytics for every operation, the correction note gives Economics to Selling Partner Insights, and the seller analytics schema gates by a PAAM right neither names; the pages disagree, and the `createQuery` reply on the connection binds (E22, S13). - **Ads in effect review.** effect-measurement.md subtracted all ad change and paid-keywords.md only manual edits; paid-keywords.md won, because automatic bid responses are mediators (E23). - **429 handling.** sp-api-access.md had growing, jittered retries; report-lifecycle.md, which owns the script, has no retry inside the script and one back-off retry decided by cx; report-lifecycle.md won (S14). - **Keepa fields.** `csv[17]` is "rating count" on the web page and "review count" in Java, and description and variation limits differ between Keepa pages; the Java source and a run-time check won. - **AI title rewrites.** competitor-set.md said Amazon does not rewrite untouched titles; the developer guide says nothing changes unless you submit, while the seller announcements say rewrites roll out gradually without a submission; the announcements won for what shoppers see (L24). - **`merge` operation.** Amazon documents `merge` for `purchasable_offer` and `fulfillment_availability` only; every content write uses `replace` (W8). - **Deal and coupon floors.** The deal floors are the seller's overall rating and the product's stars; the coupon's seller floor is the feedback rating; reviews-and-qa.md's practitioner floor lost. The coupon product floor differs between Coupon eligibility criteria and Understanding coupon errors and stays unresolved; the creation flow binds (T17). - **Image count.** The upload tab, the variant codes, and two SP-API pages give different counts; the schema binds (W1). - **Title length.** Product title requirements and guidelines and Product detail page rules agree with each other; Suppressed listings and the API troubleshooting pages give older figures; some store layers carry two figures; the schema binds and the title page is cited for the policy (W1, W3). - **Search-terms figures.** Help pages, an SP-API page and third-party schema copies give different byte figures; the schema binds for a write and the store's live page for indexing (W14). - **MYE uplift.** The MYE page and the Brand Registry page give different "up to" figures; neither is used as an expected effect (C8). - **Healthy listing.** "Catalog equals submitted" (P-listing-content-and-health-3) lost to the explainable-difference rule (L57). - **Returns lag.** The Seller Analytics Dataset Use Case Guide's figure won over the shorter practitioner figure (E52). - **Promotion report roles.** Amazon's pages give different roles for the promotion and coupon performance reports; the refusal on the connection settles it (S13). - **Rate limits.** Files quoting other figures for Catalog Items, and the A+ page against the A+ model, lost to the response header (S15). - **Variation display time.** Amazon's variation pages give different times; read-back won (W6). - **Deprecated variation theme error.** A file read it as a seller report; it is Amazon's own post on theme removal. - **Premium A+ eligibility.** A+ Content (G202102930 v17) sets criteria that the A+ Content guide (GLG4RQK2Y2RJADU4 v3) says do not exist; unresolved, so eligibility is read from the console or the seller. - **A+ module caps.** Pages disagree; validation binds (W23). - **Non-compliant titles in search.** A staff post says such titles stay searchable; the title page says they "might not appear"; the page is quoted. - **door.md report types.** Files assumed door.md lists report types; it names none (S6, open question). - **Archive copies.** The refusal of web.archive.org was the fetch tool's, not the archive's; curl worked where it was not rate-limited. - **Assistant name and Sponsored prompts.** search-rank-and-volume.md's "unlinked to any Amazon source" was wrong; both changes are on Amazon pages (L18, E25). ## The seat's own (not for a skill) These are [seat] reasoning and run lessons. They guide design; no skill states them as fact. - **Healthy-listing test.** "Catalog equals submitted" fails for many listings once Amazon's title rewrites and other contributors are counted, which is why L57 asks for an explainable difference; on hybrid ASINs the test should name whose value is displayed. - **Required orders per window.** Poisson arithmetic says a small relative move in orders needs many orders per window, and overdispersion raises that, so most single-ASIN before/after windows cannot see a small change; cannot tell will be common and MYE is preferred. - **Competitor-control bias.** Competitors lose what the seller wins, so subtracting their fall overstates the effect, while a rival that reacts hides it; a contaminated control withholds "held" rather than correcting by a guess. - **Stock-outs and the control set.** Competitors' ranks improve while the seller is out of stock, so the set can look like it "rose" during the seller's own outage. - **Accepted, not displayed.** The ACCEPTED-but-not-shown case should be its own outcome in the change log, distinct from rejected and displayed, with escalation as its next step. - **Hidden offer causes.** The most common hidden cause of a "failed" change on a brand's own ASIN is an offer event (a stock-out, a lower-priced new seller), so the offer read belongs in the baseline file. - **FBM offer deactivation.** Of the account-layer causes, the FBM offer-level deactivation is the most likely to be misread as a content problem: the account stays green while one listing goes dark. - **Hybrid brands.** A hybrid brand is best reviewed per ASIN (which account sells it, which contributed each attribute), its cleanest effect reviews are on single-channel ASINs, and content is best submitted from one account per ASIN. - **Thin stock is a choice.** The fees push in opposite directions, so a seller running thin is often choosing it, and cx's tone should reflect that. - **Peak-fee boundary.** A window straddling the start of peak fees also holds Black Friday Week preparations and the start of holiday demand, so a content verdict across it is unlikely to be clean. - **Report waits.** SQP reports appear to be processed one at a time and can take from minutes to hours, so SQP is a background job collected on a later turn; a `reportDocumentId` already seen on another receipt is the sign of a cached copy; an unexplained "A client error occurred" `FATAL` may be Amazon's own fault. - **Report quotas.** Report-type limits appear to be metered per seller account and shared with the seller's other applications, so a 429 or `FATAL` may be caused by another tool. - **False-alarm worked example.** Many watched series at a long average run length still give a steady trickle of false alarms, which is why T3 budgets alarms across the whole watch. - **Chart design for /listing-watch.** EWMA or CUSUM on each ASIN's own history, with overdispersion-adjusted limits; hard constraints (suppression, a lost Featured Offer, a stock-out) alert on occurrence, not on statistics. - **Reading order and handoff tree.** A reading follows the funnel (gates, then the stage that fell, then elements), and Seller Central handoffs go in a tree by tool, marketplace and capture date, with metadata beside each file. - **Competitor tiers.** A small confirmed control set and a wider watch list read daily. - **Origin-claim check.** A machine check can flag US-origin wording in copy or images against the country-of-origin attribute. - **Content for the assistant.** Copy written for the assistant should answer the questions shoppers ask (use, audience, compatibility, care, dimensions), following Amazon's own examples. - **Keepa coverage.** Several stores have no Keepa domain, so effect reviews there may have to answer cannot tell for the competitor subtraction. - **Retail seller id.** Amazon retail's seller id is widely reported to equal the marketplace id; no Amazon page read says so, and Keepa's `isAmazon` is the documented test. - **Run lessons (2026-10-13/14, no seller data).** An earlier run showed the sandbox origin as if it were the live store, wrote without reading the schema, used non-schema attribute names and a hyphenated language tag, put an `` where `` belongs, and offered a DELETE as a cheap fix, and fell into deleting inactive SKUs without reading why; W1, W5, W10, L58, C20 answer these. ## Facts for human readers only Every figure below was read on 2026-10-14 on the page named. No skill states them; skills read the schema, the page or the seller's console at run time. | fact | page (version or date) | |---|---| | Title: 75 characters including spaces; Item Highlights add 125; display "can take 24 to 48 hours" | Product title requirements and guidelines (GYTR6SYGFA5E3EQC v21); Product detail page rules (G200390640 v21) | | Older title figures: under 80 characters (apparel, shoes, luggage) and 130 (other); `item_name` 200 on SP-API pages | Suppressed listings (G200898440 v13); Listings Items API Issues Troubleshooting; Manage Listings Issues | | "Reducing product title length to under 100 characters" (an idea list, not a rule) | Tips for experiments (G2BNMJ8NFCGZQVFH v5) | | Title rule history: 200-character rule (January 2025); 75 + 125 rule (July 2026); Item Highlights shown from 2026-08-10; AI rewrites from July 27, 2026 | Updates to improve your product titles begin on July 27; amazon-change-calendar.md | | Search terms: "less than 250 bytes"; table limit 249; over the limit "completely ignored"; spaces and punctuation not counted | Keyword attributes explained (GF2C2L6RCFZGWBXC v4); Optimize your product discoverability (G10471 v13); Use search terms effectively (G23501 v14 US, v9 en-GB) | | Generic keyword required for 23 product types from December 6, 2023 | Use search terms effectively (G23501 v14) | | `generic_keyword` 500 bytes; third-party schema copies 250 or 2000 bytes | Listings Items API Issues Troubleshooting; as cited in elements/search-terms.md | | `generic_keyword` dropped from Catalog Items responses on 2023-08-16 | SP-API Release Notes | | Search latency "up to 72 hours"; bullets together under 1,000 characters and not always indexed | Amazon search glossary (GF6B8ZPNY2AP72VV v2); My ASIN is unsearchable (GHGHDMU4X5K9QGCV v6) | | Description over 1000 characters is an error example | Error code explanations (G17781 v13) | | `getListingsItem` reflects a write in 1–5 minutes; notifications 15–20 minutes; a 15-minute backend cycle | Listings APIs FAQ | | A verified contribution shows "within 48 hours" | Suggest changes to your product detail page (G200335450 v23) | | Images: 500 px minimum, 10,000 px maximum, 1,000+ px for zoom, 85% fill, RGB 255 background, display up to 24 hours, six additional images and one video recommended | Product image guide (G1881 v28) | | "Maximum 9 images are allowed" in the upload tab; video for sellers selling three months or more | Add product images and video (G200216080 v15) | | Variations: families over 4,000 children not displayed; merges "in under 15 minutes", wait 24 hours before contacting support; other pages give 24–28 h, 24–48 h and 96 h for reviews | Variation relationship FAQ (G201951410 v16); Create and manage product variations; Troubleshoot product variations | | Review sharing changes from Feb 12, 2026, with 30 days' email notice | Changes to review sharing across product variations; Guidelines for reviews shared across product variations (G74DLBUEHT5GY88C v3) | | Irrelevant variation themes removed from September 2 to November 30, 2025 | Removal of irrelevant variation themes will start September 2 | | Featured Offer: "As of September 2026" no seller eligibility requirement | Featured Offer eligibility (G200418100 v23, removed page) | | Deals: seller 3.5 overall rating, product 4 stars; 65% of variations for some categories; seller-fulfilled late shipment ≤ 4%, ODR ≤ 1%, cancellation ≤ 0.5%; Lightning Deals 4 to 12 hours | Amazon deals (G202111490 v25) | | Deal fees: $70 per day plus 1.0% capped at $2,000; 2026 peak events $100 plus 1.5% capped at $5,000; early-bird saving $50 | Understand deal fees (G202111590 v27) | | Event deals: lowest price in the past 30 and 60 days; date revealed at least one week before | Deals for events (G202183600 v10) | | Prime Day 2026: submissions opened March 24, 2026; scheduling closed June 9, 2026; Prime Day in June 2026, four days | Prime Day readiness playbook (GM2ADTH3XH3A5N57 v4); seasonality-and-events.md | | Prime Big Deal Days: 2025 October 7–8; 2026 October 6–7. Black Friday Week: 2024 from November 21; 2025 from November 20 | aboutamazon.com event pages, as cited in seasonality-and-events.md | | Coupons: seller 3.5 feedback rating; product 2.5 stars (1–4 reviews) or 3 stars (5+); another page says 1; discount at least 5% | Coupon eligibility criteria (GJHCPEFJ5JJQD52D v14); Understanding coupon errors | | Holiday peak FBA fees 2026-10-15 to 2027-01-14 | 2026 US FBA fulfillment fee changes, as cited in inventory-and-fulfillment.md | | FBM offer-level deactivation from 2026-08-31 | Review Account Health updates for FBM offers | | Commingling ended March 31, 2026 | FBA Barcode Choice and Labeling FAQ | | MYE: 8 to 10 weeks typical, 8 recommended; "To significance" in as little as four weeks; auto-publish at "at least a 66% probability"; weekly results; Bayesian with a 95% interval; secondary-image tests for four or fewer children | Create a listing experiment (GYZYA77YZTWTF7YZ v13); Experiment results (GY6DCFA53HYDH79H v4) | | MYE "up to 20%" against "up to 25%" | Test listing content with Manage Your Experiments; Amazon Brand Registry (sell.amazon.com) | | A+: up to 7 business days review plus up to 24 hours to publish; awards no older than 2 years; at most four quotes; images under 2 MB | Complete Guide to A+ Content for Brand Owners (US, 2023); A+ content guidelines (GGW8U76SSNTRTBX7 v8) | | A+ rate figures: the page gives 5 to 20 per operation and an aggregate 30; the model gives 10/10 | A+ Content API; aplusContent_2020-11-01.json | | Catalog Items rate 2/2 | Catalog Items API Rate Limits | | Premium A+: Brand Story plus 5 approved projects in 12 months, against "no additional eligibility criteria" | A+ Content (G202102930 v17); A+ Content guide (GLG4RQK2Y2RJADU4 v3) | | Business Reports: week Sunday 00:00 to Saturday 23:59 PT; up to two years of data; 24-hour lag; mobile included since March 29, 2022 | Business reports glossary (G27691 v7); Business reports (G200633570 v8); Business Reports FAQ (G28761 v7) | | SQP: brand view top 1,000 queries; query detail top 10 ASINs; launched 2022-04-14 | Search Query Performance dashboard (G8J4CB5ZBF3NX7TP v5); search-query-performance.md | | SQP "MONTH: 17 months" under "Maximum reporting periods" (page updated 2026-10-06) | Analytics Reports | | Returns arrive about 45 days after ship day | Seller Analytics Dataset Use Case Guide | | Rufus renamed Alexa for Shopping on 2026-05-13 | 'Amazon Rufus' AI experience comes to the Amazon Shopping app | | Sponsored Products and Sponsored Brands prompts generally available in the US on 2026-03-25 | Sponsored Products prompts and Sponsored Brands prompts (Amazon Ads) | | Promotions API (2025-12-01) launched 2026-08-26 | SP-API Release Notes | | Posts ended July 2025; Manage Your Customer Engagement ended 2024-02-16 | Amazon Posts Discontinued; News_Amazon | | View-attribution change for Store ads from 2026-01-01 | View Attribution Updates for Amazon Store ads | | Ads v3 search-term report retention 65 days; deprecation June 30, 2027 | Sponsored ads reporting FAQ (version 3); as cited in paid-keywords.md | | Vendor report clocks: hourly, weekly, and 48–72 hours by type | Analytics Reports, as cited in vendor-and-hybrid.md | | Account Health Assurance window: 72 hours on one page, 13 days on another | Account Health Assurance FAQ; as cited in account-health.md | | Store moderation up to 24 hours | stores-best-practices, as cited in brand-story-and-store.md | | Keepa: variation-specific rating count frozen since 2025-04-09; price slots changed definition on 2026-02-23 | Keepa Product Object | | New listings: up to 24 hours for an image to appear; `getListingsItem` reflects state in 1–5 minutes and notifications in 15–20; `putListingsItem` 5 a second (burst 5 on the page, 10 in the model); validation previews 20 a second per application (patch row only); definitions 5 a second, burst 10; `getMarketplaceParticipations` 0.016 a second, burst 15; schema links valid seven days; no stated time for ASIN assignment | Product image guide (G1881 v28); Add product images and video (G200216080 v15); Listings APIs FAQ; Listings Items API Rate Limits; listingsItems_2021-08-01.json; definitionsProductTypes_2020-09-01.json; sellers.json; Retrieve a Product Type Definition (read 2026-10-27, as cited in method/listing-create.md) | | Seller Central's Add a Product form: up to five bullets and nine photos | How to create Amazon product listings (sell.amazon.com, Apr 29, 2026) | ## Open questions ### For Jack's word - **The content call.** May the help hub's undocumented content call have any standing use, at research or at run time? Until ruled, no further pages are fetched through it. (help-page-access OQ1; the office rules.) - **Is a preview a write?** Ruled by brief 2026-10-26-87: a preview (`putListingsItem` or `patchListingsItem` with `mode=VALIDATION_PREVIEW`) persists nothing (P-listing-writes-5), so it is sent as any read is; a write without the mode waits for the seller's word (R6, R8). - **Acceptable Use Policy.** How does AUP 4.3 bear on Keepa use and on recommending third-party tools? - **Brand Analytics terms.** May Brand Analytics figures be kept in change files and reports under the Brand Analytics Terms and Conditions (P-own-traffic-funnel-23)? (Jack, or counsel.) - **Selling Partner plugin.** Is the Selling Partner plugin in scope for the box? - **Amazon Ads source.** Will there be an Ads source (an MCP server or a gated door), or the seller's word only? Spend is covered by E22 where Economics is open. - **Notification destination.** Should the box ever get an SQS or EventBridge destination? Until then every watch pulls (T6). (A platform ruling.) - **Encryption at rest.** Does the seller's tree meet Amazon's encryption-at-rest rule for landed reports (P-report-lifecycle-24)? (The office.) - **cancelReport.** Should cx ever cancel a report it no longer needs, to free the store's slot? (report-lifecycle S11; the office.) - **door.md report types.** door.md names no report types; should it, so that scripts/report.sh has a list (S6)? - **Account type.** How does cx learn the connection's account type (seller, vendor, hybrid): from the door, the seller, or a probe call? (The office and engineering.) - **Hybrid brands.** Will a second, vendor authorization be sought for hybrid brands? (The office.) - **Competitor review topics.** May `getItemReviewTopics` be called on competitor ASINs? - **"ASIN + keyword" check.** May cx test whether an ASIN is found for a keyword on the public site, given the Conditions of Use and robots.txt (P-search-rank-and-volume-7, search-terms OQ6)? - **Economics query.** May a Data Kiosk Economics query be run under Jack's word to settle ad dating and role? - **Images.** Can the box model view images, and may it fetch the image CDN under PRODUCT.md §7? (The office.) - **Controlled tests on a live ASIN.** A description-token or search-terms-tail test runs only with the seller's word; is that the rule? - **Category style guides and the calendar.** Should dated category style guides and a machine-readable event calendar be kept in the tree? - **Operator routine and help-page re-reads.** Who owns the operator routine, including a periodic re-read of the help-page captures? ### For the first run on a real store - **Schema limits.** The real `item_name` `maxLength`, `generic_keyword` byte limit, bullet count and length, and image slots. *Settles:* `getDefinitionsProductType` for the store's product types. - **SQP details.** Lag after week close, whether "17 months" is a span or a lookback, the ASIN-view cap, whether a parent ASIN is accepted. *Settles:* the first SQP requests. - **Data Kiosk.** Which role opens `search.query` and Economics, and how `conversionRate` reads in the live response. *Settles:* the first `createQuery` replies. - **Restatement and time zone.** How much recent days are restated and where a day is cut. *Settles:* two reads of one week a week apart. - **Keepa.** How fresh `csv[17]` is for the competitor set. *Settles:* the first dated Keepa reads. - **`similarItems`.** What it returns for the seller's ASINs. *Settles:* the first `getCompetitiveSummary`. - **Title split.** How an over-limit title is split or rewritten. *Settles:* the first title write and its Catalog read-back. - **Issue codes.** The issue codes and enforcements the store actually carries. *Settles:* the first `getListingsItem` with `issues`. - **Economics `ads` dating.** Charge day or ad day. *Settles:* the first Economics query against the ad console. - **Overdispersion.** How overdispersed daily orders are per ASIN. *Settles:* the first Data Kiosk series. - **MYE export.** What an MYE result looks like as a handoff. *Settles:* the seller's first MYE handoff. - **Seller identity.** Whether `selling_partner_id` and `sellerId` are the same. *Settles:* `getMarketplaceParticipations` and the first Listings call. - **Variation timing.** Which variation timing figure holds. *Settles:* the first family write timed against Catalog `relationships`. - **Coupon product floor and deal eligibility.** *Settles:* the seller's coupon creation flow and Deals dashboard. - **Premium A+ access.** *Settles:* the A+ Content Manager banner as the seller reports it. - **Report processing times and caching.** *Settles:* landed receipts from the first reports. ### Pages that stayed unreachable The pages below were not read in this session by any route. *Settles:* a signed-in read by the seller or Jack, an archive copy the fetch tool can open, or the page's real address. | group | pages | |---|---| | Seller Central help not served (`HelpArticleNotFound`, `topic: null`, HTTP 400 or empty body) | About parent-child relationships (G8841); Add a Product to Create Variations (G40731); Variation Wizard (G202034620); G201047930; G202121590; G201749990; GF85G8KLPFJZZV5K; Enhance your listings (G200403880); Registering products to your brand (G6DU75NSM86VXKZC); Brand Representative / Selling role (GJ84K745AL3R5N3Q); Product safety and compliance (GUH6FA4XSJ2LZFLY); FBA inventory overview (GTMXYZN64UJL7TT6); the Seller Fulfilled Prime trial page; Amazon Business bulk order fee discount (GQBCQZZKZQ94LL55); Returns processing fee (G64LS955WNFT6EDP); a group of G2002857x0 / G2021131x0 ids; four DE-layer ids (G202102350, G201723400, G201819090, G201434010); the Voice of the Customer page | | Seller Central behind sign-in or a shell | Categories and products that require approval (G200333160); Customer product reviews policies (GYRKB5RU3FS5TURN); Amazon Vine (GX3CEV63Y4ZZAP37); Seller Assistant for support resolution (GCTMXALNDLGAR637); G202204770; G1881 on the .co.uk, .in and /gp routes; Fix Your Products; the Account Health dashboard; Climate Pledge Friendly seller page; Brand Analytics Metric Glossary; several Seller News articles; Seller Central EU G1661 | | Vendor Central and Brand Registry | Vendor Central title requirements; Vendor Central help and contact; Brand Registry help; Brand Registry Manage selling accounts | | Amazon Ads pages rendered by script or missing | several Ads help pages (Creator Connections and others); Ads Guidelines and Acceptance Policies; "What's new" index; Ads API overview, eligibility, reporting v3 columns, search-term report page, product targeting guide, theme-based bid suggestions, Stores OpenAPI; Sponsored Brands and Sponsored Display ad specs | | SP-API docs answering "Page Not Found" | Data Kiosk economics dataset guide; Data Kiosk schema explorer guide; Customer Feedback v2024-06-01 reference; brand-analytics-role and selling-partner-insights-role; report-type-values-listings and -seller-performance; old Product Type Definitions slugs; listings-feed-schema-v2 docs page; Notifications API v1 use case guide; SP-API vendor FAQs | | Amazon storefront and other Amazon pages | Best Sellers Rank customer-service page; Strike-Through pages on amazon.de, .co.uk, .fr; Climate Pledge Friendly storefront; Amazon Q4 2025 results; aboutamazon "trustworthy reviews"; guessed sell.amazon.com and sustainability addresses; Amazon Science MAVE page; Amazon Influencer sign-up | | Research behind publisher walls | ACM, INFORMS, SAGE, Springer, Wiley, SSRN, JSTOR, Taylor & Francis, AEA papers listed in each file's *Pages read* (open versions read where they exist) | | Regulators | Directive (EU) 2019/2161 on EUR-Lex; FTC case pages and the 2025 fake-review warning template | | Keepa | the product-object discussion page | | Practitioners | guessed Helium 10 and Jungle Scout addresses; Helium 10 KB (Cloudflare); Kapoq; SellerMetrics; LandingCube; Listing Forge; Search Engine Land; GeekWire 2018; Openbridge; ChannelEngine and MindgruveMacarta vendor guides; Acadia; eStore Factory; Gorilla ROI; Riverbend; Code3; Teikametrics; Perpetua and Pacvue guessed pages; EcommerceBytes; CNN Underscored | | Archive copies | web.archive.org copies of G1881, G202111490, GTRP6DXEA983DPD6, G23501, GF2C2L6RCFZGWBXC and G200390640 (fetch tool refused; curl got 429 or empty shells) | ## Pages read Counts are rows of each file's *Pages read* table, so a page read by several files counts in each. Rows were classed by a script, approximately: 1. An explicit [tag] in a row wins. 2. A row whose notes mark it unreachable (and that holds no capture or "read" note) counts only as unreachable. 3. Otherwise the host decides: academic, DOI, preprint and Amazon Science hosts count as research; Amazon's domains, Amazon's and Keepa's documentation and GitHub, regulators and Google Merchant Center help count as official; everything else counts as opinion. 4. 29 rows pointing to internal documents (door.md, the office mailbox, SKILL.md files, other research files) are left out, and 65 rows without a URL were classed by their text. | file | official | opinion | research | unreachable | |---|---|---|---|---| | angles/account-health.md | 49 | 6 | 1 | 3 | | angles/ai-shopping-surfaces.md | 48 | 8 | 6 | 5 | | angles/brand-protection-and-catalog-contribution.md | 33 | 6 | 2 | 5 | | angles/competitor-discovery.md | 39 | 11 | 6 | 3 | | angles/competitor-set.md | 67 | 8 | 11 | 16 | | angles/inventory-and-fulfillment.md | 67 | 6 | 1 | 4 | | angles/listing-content-and-health.md | 120 | 7 | 3 | 5 | | angles/marketplaces-and-languages.md | 52 | 5 | 2 | 4 | | angles/off-listing-traffic.md | 44 | 9 | 4 | 8 | | angles/offer-competition-and-hijackers.md | 46 | 11 | 1 | 4 | | angles/own-traffic-funnel.md | 57 | 14 | 7 | 2 | | angles/paid-keywords.md | 74 | 21 | 9 | 4 | | angles/price-position.md | 111 | 11 | 8 | 8 | | angles/returns.md | 60 | 15 | 3 | 7 | | angles/reviews-and-qa.md | 71 | 20 | 10 | 16 | | angles/search-query-performance.md | 45 | 16 | 5 | 8 | | angles/search-rank-and-volume.md | 49 | 17 | 3 | 3 | | angles/seasonality-and-events.md | 91 | 8 | 8 | 6 | | angles/vendor-and-hybrid.md | 55 | 19 | 4 | 8 | | elements/a-plus-content.md | 61 | 11 | 7 | 3 | | elements/attributes-and-browse-nodes.md | 84 | 7 | 11 | 7 | | elements/badges-and-placements.md | 53 | 8 | 2 | 7 | | elements/brand-story-and-store.md | 38 | 11 | 1 | 4 | | elements/bullet-points.md | 79 | 15 | 11 | 8 | | elements/description.md | 83 | 8 | 2 | 4 | | elements/images-and-video.md | 91 | 12 | 2 | 9 | | elements/search-terms.md | 61 | 12 | 6 | 7 | | elements/title.md | 93 | 12 | 5 | 6 | | elements/variations.md | 73 | 12 | 4 | 10 | | method/amazon-change-calendar.md | 30 | 3 | 2 | 2 | | method/effect-measurement.md | 50 | 12 | 26 | 6 | | method/featured-offer.md | 8 | 0 | 0 | 0 | | method/funnel-and-search.md | 89 | 12 | 12 | 9 | | method/help-page-access.md | 41 | 0 | 0 | 3 | | method/listing-create.md | 30 | 0 | 0 | 2 | | method/listing-policies.md | 93 | 15 | 7 | 11 | | method/listing-writes.md | 79 | 8 | 3 | 2 | | method/report-lifecycle.md | 62 | 4 | 1 | 5 | | method/seller-central-only-tools.md | 51 | 7 | 2 | 3 | | method/sp-api-access.md | 40 | 2 | 0 | 6 | | method/triggers.md | 84 | 10 | 6 | 4 | | **total** | **2543** | **399** | **204** | **237** | Distinct URLs on rows that were read (first URL per row): 1,663. The unreachable rows come to 166 distinct addresses not read by any other file. The office's capture folder (outside this repository) holds 54 dated captures of Seller Central help pages. The files hold about 921 position IDs across 19 angle, 10 element and 11 method files, merged into the 280 positions above (method/listing-create.md's 24 into R1–R10, its 32 page rows counted in the table but not in the distinct-URL figure).