Read the seller SQP report at query, ASIN, store and period grain with separate denominators for each funnel stage.
Read the two-week table before naming a cause
Synthetic data for one query, ASIN and store: Week A has query totals of 10,000 impressions, 1,000 clicks, 200 cart adds and 100 purchases. The selected ASIN has 1,000 impressions, 100 clicks, 20 cart adds and 8 purchases. Its impression share is 1,000/10,000 = 10%; click share is 100/1,000 = 10%; cart-add share is 20/200 = 10%; purchase share is 8/100 = 8%. Each percentage has its own query-stage denominator.
In synthetic Week B, query totals rise to 12,000 impressions, 1,200 clicks, 240 cart adds and 120 purchases. The ASIN records 1,200 impressions, 120 clicks, 24 cart adds and 6 purchases. Its first three shares remain 10%, while purchase share drops to 6/120 = 5%. The query registers 20 more purchases overall, yet this ASIN registers two fewer. The numbers justify checking an offer or listing; they do not establish that a price change, a competitor or an ad placement caused it. All ASIN counts here are within the corresponding query totals.
Confirm this is a seller SQP report
The documented SP-API report is GET_BRAND_ANALYTICS_SEARCH_QUERY_PERFORMANCE_REPORT. Amazon describes this report for Brand Registry sellers with the Brand Analytics SP-API role. A broad Brand Analytics role description that also mentions vendors does not confer access to this particular seller SQP report on a vendor account. Verify the authorized application, seller account, eligible brand, ASIN and store. A prompt cannot supply permission the integration lacks.
With eligible Brand Analytics data in 1stPage, Claude can request brand_analytics with asins_ranking_for_keyword for the exact query, seller account, marketplace and report period. Keep the reported ASIN and query counts attached to each share. An account without the SQP capability returns no usable query result; the connector cannot grant Brand Registry eligibility on the seller’s behalf.
The report is requested, not scheduled, and returns JSON. Its request uses a required reportPeriod and a space-separated ASIN parameter subject to a documented 200-character limit. Check that the requested ASINs fit, and keep the marketplace ID attached to the result. Amazon documents supported stores in NA, specified EU stores and FE stores; it does not follow that every account can query every store or combine them into a single row. A missing report may mean entitlement, date, processing or eligible-scope trouble. It does not mean search demand was zero.
A week is not seven daily observations
SQP period options are WEEK, MONTH and QUARTER, and one request covers one valid period. A WEEK runs Sunday through Saturday; month and quarter requests likewise need valid first and last boundaries. Compare complete periods of the same type in the same store. Dividing a weekly count by seven invents daily SQP observations the report did not provide. Treat an incomplete current period as unavailable for an equal-window comparison rather than dressing it up as a full week.
Preserve the exact report start and end dates, ASIN and query in the output. If a requested period falls outside available retention or misses the required boundary, say so and request a valid period. Do not silently shift a date and keep the user’s original label. Use searchQueryVolume and searchQueryScore only as defined in the actual report, not as a made-up monthly keyword-planner volume.
Shares, rates and market size answer different questions
For each stage, ASIN share divides that ASIN’s count by the query total at the same stage: impressions by impressions, clicks by clicks, cart adds by cart adds, purchases by purchases. Click-through rate instead divides clicks by impressions for a specified population. A purchase rate needs its own chosen denominator and population. Neither rate can be substituted for SQP purchase share. These events are aggregated stage counts, not proof that the same shopper moved through every stage in a single tracked journey.
A 5% SQP purchase share for one query and one ASIN is not 5% of a product category, all Amazon searches or a brand’s market. Organic position, competitor bids and total category sales do not emerge from these rows. If the real question is broader market share, state that this report cannot answer it. No 1stPage SQP connection or live seller result is implied by this synthetic example.
Set one measurable follow-up
For the two-week example, verify availability, price and offer condition for the selected store before attributing the purchase-share decline to listing content. Name one proposed change, the owner, the eligible next reporting period and the count and share you will compare. Keep the unchanged 10% impression, click and cart-add shares in the brief; that contrast is the reason the purchase stage deserves scrutiny. Publish the denominator next to every percentage so a later reader can test the same claim.
For one query, ASIN and marketplace, compare two complete equal SQP periods. Print the ASIN and query total at each stage beside its share, and separate observed differences from causal hypotheses.
Sources and boundaries
Check the current permissions and report definitions before acting on an answer. Example prompts are questions, not live account results.