A percentage from owned sales, SQP, or ads is not category market share unless its denominator covers the defined category.
Put the universe in the denominator
Category unit share needs units sold by the defined seller or brand in a specified category, Amazon store, and period, divided by all eligible category units sold in that same universe. Value share needs comparable sales value on both sides. Define whether bundles, replacement parts, and adjacent product types belong to that category. A supplier's estimate can serve as a denominator only with its coverage and method attached. Changing the unit, date range, or product set on one side breaks the fraction even when the spreadsheet formula still runs.
An owned sales file usually has no total for everyone else. It can show that one ASIN is 10% of the ASINs in the file, or that owned units grew 10% month over month. Those are portfolio mix and growth, respectively. Neither is category share. Put the numerator source, denominator source, store, dates, unit, inclusion rules, and exclusions in a single worksheet row. If one cell is unknown, mark category share unavailable. Claude is not a source for the missing category total.
A real Amazon percentage with the wrong name
Amazon Search Query Performance reports query-stage counts and shares for specified ASINs. Suppose an explicitly hypothetical ASIN accounts for 12 of 120 purchases on one reported query in one week. Under those stated counts, its purchase share on that query is 10%. It is not 10% of the US bottle category. The report says nothing there about purchases through other queries or browsing paths, and it does not turn one query into the category universe. Keep 'query purchase share' in the chart title and the ASIN, store, and week in the footnote.
A 1stPage MCP result from brand_analytics can provide query-specific ASIN purchase share. Seller_analytics can summarize the products in your own account. Neither one is a census of all category sales. When Claude presents a percentage, ask which 1stPage dataset produced its numerator and where the denominator came from; refuse the label ‘category market share’ if no category-wide total exists.
Nor should impression share and purchase share be treated as the same population. Each stage has its own denominator. Amazon's documented SQP access requirements include Brand Registry eligibility and the Brand Analytics role for the requested API report. If an account lacks that report, an owned order export cannot recreate its observed query totals. If multiple SQP rows are available, adding their shares will not manufacture a category total. A group of measured query fractions remains a group of measured query fractions.
The advertising number is yet another fraction
An Amazon Ads dashboard can report attributed sales or campaign reach under its own product definitions and attribution windows. That is not a census of all category buyers or all seller sales. An ad-attributed numerator divided by owned gross sales might answer a carefully defined internal question, but calling the result market share changes both the population and the meaning. State the attribution window, currency, campaign scope, and reporting dates before placing ads data next to seller data.
Amazon Marketing Cloud (AMC) does not hand an advertiser the category denominator. Its eligible instance can analyze permitted pseudonymized signals and return privacy-protected aggregate results, which cover the instance's allowed population and rules. A suppressed small group does not become zero sales in the category. Keep any AMC result under its own label; do not use it as a filler for transactions outside the advertiser's observable scope. Selling Partner API and Ads permissions are separate, so access to one does not imply access to the other.
Let the estimate show its seams
Here is a second invented calculation for checking language, not a claim about any seller: owned units move from 800 to 880, while an outside provider estimates total category units moving from 10,000 to 12,000. On those assumptions, estimated unit share moves from 8% to about 7.3% while owned units rise 10%. Before publishing either share number, inspect whether the provider's category, store, period, and treatment of bundles match the owned data. If its methodology cannot be inspected, publish only the verified owned-unit movement.
The best use of Claude here is as a hostile slide editor. Feed it the proposed claim and only authorized source definitions. Require a numerator, denominator, unit, time period, store, and source for each percentage. Have it rewrite 'market share' as 'owned portfolio mix' or 'SQP query purchase share' where appropriate, and return 'not calculable from these files' for a missing category total. Independently recalculate at least one row. The point is not to produce a prettier caveat; it is to prevent a wrong chart title from surviving review.
Sources and boundaries
Check the current permissions and report definitions before acting on an answer. Example prompts are questions, not live account results.