Amazon Marketing Cloud (AMC) queries are SQL questions you run against your own Amazon Ads signals to see how shoppers move between ads, product pages and purchases. Access is easy now. The hard part is asking the right question, writing it in AMC’s restricted SQL, and trusting the answer. For a mid-market or enterprise brand, the best results come when a team that knows both SQL and Amazon advertising owns the queries and ties them to budget decisions. That is the work our team does inside Amazon Growth & Optimization.
Standard Amazon Ads reports tell you what a campaign did. AMC queries tell you what shoppers did across campaigns. You write a request in SQL (the standard language for asking questions of a database), and AMC runs it against ad exposure and shopping events tied to your account, then returns an aggregated table.
Amazon describes the Instructional Query Library as a set of SQL queries available in the AMC web interface. Each comes with business context, usage instructions and guidance on reading results. They range from path-to-conversion measurement to technical examples such as filtering text fields with regular expressions.
The questions brand teams bring to AMC tend to fall into four groups:
Nancy McLaughlin, a marketplace executive, told Ecommerce Fastlane that AMC gives brands detailed data on audience journeys and the links between events, beyond what traditional reports show. That is the practical value: you stop optimizing each campaign in isolation.
For years AMC was gated. According to MediaPost, since mid-September 2025 any advertiser running Sponsored Products, Sponsored Display, Sponsored Brands or Sponsored TV campaigns gets AMC access directly in the Amazon Ads console. Earlier access required a partner or registration and was effectively limited to larger DSP advertisers.
The path is short:
You may see older guides that list DSP spend minimums and a required SQL-literate technical resource. Ecommerce Fastlane’s 2023 update quotes those original requirements, which is useful context for why so much AMC advice assumes an analyst on staff. The skill requirement still matters even though the gate is gone. Opening the editor is free. Getting a decision-grade answer from it is not automatic.
The home page gives you three levels, and most brands should use all three.
These run pre-built analyses with a few settings. They suit a marketing manager who wants a first look at conversion paths or new-to-brand performance without writing anything.
The library gives you working SQL with explanations. You change dates, ASINs (Amazon’s product identifiers), campaign filters or lookback windows. This is the best place to learn how AMC tables behave, because each query documents what its output means.
The editor is for questions the templates do not cover. In January 2025 Amazon announced a generative AI SQL generator that turns a plain-language request into SQL. Its example asks for customers who viewed product pages and saw streaming TV ads but did not purchase in the last 30 days. Amazon said the tool reduces query development time from hours to minutes. It is a faster draft, not a replacement for review, because a query that runs is not necessarily a query that answers your question.
AMC uses a restricted dialect. PPC Land’s explainer notes that SELECT * is rejected, so every column must be named, and that rows describing too few people are removed before results come back. Analysts who write SQL daily still trip over these rules in their first weeks.
A working AMC query usually has the same anatomy:
A typical cross-channel question works like this. Take shoppers exposed to a DSP ad and a sponsored ad, join them to purchase events inside a conversion window, group by which combination they saw, and count distinct purchasers. The output shows whether shoppers who saw both ad types converted at a different rate than shoppers who saw only one. That is the kind of insight no single campaign report can give you.
For a brand with established Amazon spend, these four give the fastest return.
Shows the sequence of touchpoints before a purchase. Use it to find which ad type opens the journey and which closes it, then check whether your budget split reflects that. Amazon lists path-to-conversion as a core use case in its query library.
Separates first-time brand buyers from repeat buyers by campaign and by product. A campaign with a high return on ad spend but a low new-to-brand share may be harvesting existing demand rather than growing the customer base. That distinction changes how you set targets.
Build a segment of shoppers who viewed your detail pages, or who saw video ads, but did not purchase in a set period, then activate it for sponsored ads or DSP. AMC audiences built from Amazon signals can only be bought through Amazon’s own channels, according to PPC Land, so plan your media accordingly.
Shows how many days pass between first exposure and purchase. If most buyers convert within a week, a long retargeting window wastes spend. If the lag is long, short attribution windows undercount your upper-funnel ads.
This is where most AMC mistakes happen, and where a clean-looking table can mislead a budget meeting.
AMC returns only aggregated analytics, and Ecommerce Fastlane reports that every aggregation must contain a minimum of 100 users. Rows below the floor are dropped. PPC Land points out that this suppression is silent: a missing small segment looks the same as a segment where nobody behaved that way. If your result seems to be missing a niche product, a small geography or a new campaign, the cause may be the threshold, not zero activity.
The practical fixes are to widen the date range, group at a higher level (category instead of single ASIN) or combine segments before you conclude anything.
Before a new query drives a decision, run a simple control: total impressions, clicks, sales and units from your query for a given period should reconcile with what the Ads console shows for the same period. If they do not, find out why before acting. Differences in attribution window, ad type coverage or filters are common causes.
At unBoxed in November 2025, Amazon extended AMC’s ad-traffic lookback from 13 to 25 months, starting in open beta in the US and Canada. Existing SQL editor queries can reach the longer data without endpoint changes. For brands with seasonal products, this finally allows two full holiday comparisons in one query.
AMC shows who was exposed and who bought. It does not prove the ads caused the purchase. To know that, you need a designed test with a holdout group. Our AMC and incrementality testing work pairs AMC data with that kind of design so that spend decisions rest on lift, not on exposure counts.
Once a query is validated, you can save it, schedule it and send results where they are useful. PPC Land explains that AMC outputs go two ways: reporting workflows that run on demand or on a schedule and return aggregated files, and audience workflows that build rule-based and lookalike audiences from the same signals, activated across sponsored ads, video and display.
Two habits keep this manageable at scale:
Amazon also introduced Ads Agent at unBoxed 2025, which, per PPC Land’s report, translates business questions into SQL and automates campaign management tasks across AMC and Amazon DSP using natural language. Faster query drafting makes the review step more important, not less, since more people can now generate queries than can judge them.
Opening the editor takes a minute. Running it well is a staffing question. Brand teams typically face three gaps:
For a brand running Amazon, Walmart and its own D2C site, there is a fourth gap: AMC covers Amazon signals, while the budget decision spans every channel. That is why a unified view matters. Our iDerive analytics platform brings marketplace data into one place, and our write-ups on AMC use cases for brands and using AI for Amazon analytics show how that looks in practice. Our piece on AMC audience targeting and search bidding covers the activation side.
If you are starting now, this sequence avoids the usual false starts.
The discipline is the same at every scale: one question, one validated query, one decision.
We see brands treat AMC as a reporting upgrade, and that undersells it. The value is not a prettier dashboard. It is being able to ask why a shopper bought, and which combination of ads helped, then move money accordingly.
That takes three skills in the same room: SQL that respects AMC’s rules, a working knowledge of how Amazon Ads, DSP and retail media interact, and a commercial view of what the business needs, which is profit, not just attributed sales. Our approach is built around profit-first advertising, with iDerive giving one unified view across marketplaces so an AMC finding becomes a budget decision rather than a slide.
Our advice to enterprise and mid-market brands is simple. Use the free access, learn the library, and be strict about validation and thresholds. Then put the insights to work through the retail media, creative and operations teams that can act on them together.
Dan Katona
If your team wants AMC answers tied to real spend decisions, our fully managed Amazon Growth & Optimization service combines retail media management, marketplace operations and analytics powered by iDerive. For the advertising side, see our Amazon sponsored ads management.

Bring us the question you want AMC to answer, and we will show you how we would structure the analysis and the follow-up test.
No. Since September 2025, advertisers running Sponsored Products, Sponsored Brands, Sponsored Display or Sponsored TV campaigns can access AMC in the Amazon Ads console, according to MediaPost. Look under Measurement & Reporting. DSP data adds more tables to query, but it is no longer the entry requirement.
AMC drops any output row that describes too few users, and it does not flag the omission. The floor is 100 users for aggregations, per Ecommerce Fastlane. Widen your date range or group at a higher level to bring thin segments back above the threshold.
Yes, in part. Amazon announced a generative AI SQL generator that turns plain-language requests into queries. Treat the output as a draft. Check the joins, the filters and the date windows, then reconcile totals with the Ads console before using the result.
Amazon extended the ad-traffic lookback window from 13 to 25 months, starting in open beta in the US and Canada in November 2025, per PPC Land. Check the Amazon Ads documentation for the current availability in your account and region.