Reconcile a seller sales decline by ASIN before testing sessions, conversion, stock history and separate Amazon Ads evidence.
A $150 decline needs a $150 reconciliation
Suppose an illustrative account moves from $1,000 to $850 in ordered product sales. ASIN A falls from $500 to $350, B from $300 to $270, and the remaining ASINs rise from $200 to $230. The deltas are -$150, -$30 and +$30. Their sum is -$150. A contributes the entire net decline, though B fell too. These numbers are synthetic, not customer results.
With 1stPage connected, ask Claude which ASINs explain the seller’s two-week change. Make their deltas add up, then inspect visits and inventory for the products that moved. If the answer blames campaigns, ask for a separate Ads result from the same account and period; seller sales rows cannot prove that cause.
A percentage-only ranking would overstate a tiny ASIN's importance; use dollar contributions and retain both old and new rows.
Compare the sum of ASIN changes with the report's account/date total. Do not quietly invent an 'other' bucket for a mismatch. Inspect filters, omitted rows, currency, newly introduced ASINs and the population represented by each total. Keep a source reference next to the table so a second analyst can repeat the sum. A small discrepancy from reporting grain is still a discrepancy to identify, not permission to round away a material gap.
Sessions fell too. That is not a cause yet
For the ASINs accounting for the decline, put sessions and the seller report's named conversion metric beside ordered sales for both periods. Amazon distinguishes order-item/session and unit/session percentages. Units divided by sessions cannot silently replace an order-item conversion field; page views are not unique shoppers. Preserve ASIN grain and check whether its session values can be reconciled with a date-level view before rolling them up.
Fewer sessions with a steady conversion measure raises a traffic question. Similar sessions with fewer orders per session raises an offer, price, listing or availability question. Both measures can deteriorate together. None identifies an ad campaign as the cause. Verify a featured-offer change, suppressed listing or price change against a dated, authorized source. When there is no dated evidence, write down the hypothesis and leave it there. Claude can organize contradictory signals without pretending that correlation is diagnosis.
For ASINs explaining the material sales delta, show sessions, units ordered and the report's named conversion metric in both completed periods. State the grain and denominator. Split observations from hypotheses and identify any missing listing, offer or inventory evidence.
Lock the comparison after locating the loss
The ledger is only credible when the two weeks are comparable. Use an authorized seller Sales and Traffic report, such as GET_SALES_AND_TRAFFIC_REPORT, for the same seller account, marketplace, currency and ASIN scope. Record the report name, retrieval time and reporting time zone. Choose two finished seven-day periods with matching weekday boundaries. Today's partial data and late report updates can manufacture a decline; date the finding as provisional until the source period is ready.
Define ordered product sales explicitly. They are not seller settlements, shipped revenue or Amazon Ads attributed sales. Check whether the export separates B2B fields and whether filters changed between periods. Amazon documents a two-year start-date lookback for this seller report; confirm the current request rules when going farther back. Keep the original downloaded report and account total, rather than copying only Claude's explanation into the ticket.
Identify the seller report, account, marketplace, currency, retrieved-at time, reporting time zone and exact complete-week boundaries. Define the sales field and confirm the ASIN scope in both periods. Stop the comparison if either week is unfinished or its report unavailable.
Today's inventory cannot explain last Tuesday
Current sellable units are a snapshot. They do not establish when an ASIN ran out during the sales week. Check dated availability, fulfillment mode and listing status for the affected ASINs. An inbound shipment estimate is not sellable stock, and merchant-fulfilled quantity is not FBA quantity. If the only evidence is today's low stock, say 'historical stockout unverified.' The stockout companion article uses conditional days of cover to frame future exposure, not to backfill a historical outage.
Amazon Ads evidence runs on another clock. An approved Ads report can show spend and clicks for relevant campaigns over comparable dates, but Ads requires its own authorization. Campaign-attributed sales follow attribution windows and cannot be subtracted from seller ordered product sales to manufacture an 'organic' residual. Falling ad clicks alongside falling total sales calls for a closer look, not a causal verdict. With no Ads extract, report Ads as unobserved rather than stable.
Leave the next analyst an executable question
A useful handoff says exactly what changed and what evidence would decide the next branch. 'A accounts for $150 of the $150 net sales loss; its sessions also fell' points to a reconciled seller report. 'Advertising caused it' needs compatible campaign data and checks on offer, availability and other traffic. Put period, currency, source and calculation beside each finding. This is more useful than a polished paragraph whose numbers cannot be traced. A 403 permission error, a report still processing, an absent ASIN row and a measured zero are not interchangeable. Preserve each status. If report completion lags, rerun the comparison when the source catches up rather than recycling the last answer. The next question should name the missing report rather than replay the same sales summary.
Sources and boundaries
Check the current permissions and report definitions before acting on an answer. Example prompts are questions, not live account results.