Advertising

Amazon Marketing Cloud: what you can analyze (and what you can’t)

Burst team · Updated 2 November 2026 · 7 min read

In short
  • AMC is a clean room: it joins Amazon advertising data with your own, but only ever returns aggregated results.
  • The most useful analyses cover conversion paths, campaign overlap, frequency and new-to-brand customers.
  • Queries can be turned into audiences for DSP and Sponsored Ads.
  • It pays off when ad spend runs across several formats and someone can actually read the output.

Amazon Marketing Cloud (AMC) answers a question that standard Amazon Ads reports cannot. Those reports tell you what each campaign sold, one by one. They do not tell you how campaigns work together. Here is what you can really analyze with Amazon Marketing Cloud, which audiences you can build, what the new AMC Expert Tool changes and which limits you should know before you start.

What Amazon Marketing Cloud is

AMC is a clean room: a secure environment where Amazon exposes ad events (impressions, clicks) and purchase events at pseudonymized user level. You never see an individual person. You write a query, AMC runs it on the raw events and hands back only the aggregated result.

Queries are written in SQL. If you do not write SQL, Amazon provides sample queries (the so-called instructional queries) and, in recent releases, no-code templates and an assistant that helps you build analyses. Since 2025 Sponsored Ads advertisers can turn AMC on themselves from the ads console in most major markets, without going through an Amazon contact.

AMC holds data from Sponsored Products, Sponsored Brands, Sponsored Display and Amazon DSP, plus conversion events on Amazon. You can also upload your own first-party data (from your DTC store, for example), always in protected form.

The analyses worth running

AMC can do a lot. In our projects, these are the questions that most often lead to real decisions.

Conversion paths

Which mix of touchpoints comes before a purchase? For example: how many customers saw a DSP ad, then searched for the brand, then clicked a Sponsored Brands ad before buying. This shows the role of upper-funnel formats, which tend to look weak in standard reports because they rarely get the last click.

Channel overlap

How many users saw both Sponsored Products and DSP? Do people exposed to both convert differently from those who saw only one? If combined exposure converts much better, that is a case for coordinating budgets and audiences.

New-to-brand customers

New-to-brand metrics already sit in the reports, but AMC lets you cut them by timing, format and product. You can see which campaigns truly bring new customers and which mostly catch people who would have bought anyway.

Frequency

How many exposures does it take before purchase likelihood stops rising? This lets you set smarter frequency caps in DSP and stop paying for users who are already saturated.

Time to conversion

How many days pass between the first ad touch and the purchase? For premium or repeat-purchase products, the answer changes how you read a campaign’s first week.

Audiences for DSP and Sponsored Ads

AMC is not only for reading data. The same query logic can become audiences you activate. A few practical examples:

  • users who added to cart in the last 30 days without buying;
  • customers who bought a consumable and are near their repurchase window;
  • users who saw a video and never visited the product page;
  • lookalike audiences based on repeat buyers.

These audiences become targets in DSP, targeting in Sponsored Display and, for Sponsored Products and Sponsored Brands, a basis for raising bids on the people inside them. This is the point where AMC stops being an analysis exercise and starts moving budget.

What AMC does not do

The limits are clear, and knowing them keeps expectations realistic.

ExpectationWhat actually happens
Seeing individual customersNot possible. Results are always aggregated, and rows under a minimum user threshold are dropped.
Analyzing very small products or segmentsWith few users per row, much of the data vanishes under aggregation thresholds. Group things together.
Unlimited historyLookback has a window set by Amazon: check it in your own instance before planning long-range work.
Real-time dataThere is a lag between the event and the data being available. AMC does not replace daily reports.
Seeing competitor dataNo. AMC only holds events tied to your campaigns and your products.
Answers without questionsAMC does not generate insight on its own. Without a clear hypothesis you get tables, not decisions.

There is a practical limit too: badly written queries return numbers that look plausible and are wrong. One mistake in the time window or the attribution changes the conclusion. You need someone who can sanity-check results against common sense and the standard reports.

Who should use it

AMC makes sense when there is enough data and enough formats to compare. It is generally worth it if:

  1. you invest in several formats (at least Sponsored Products and Sponsored Brands, ideally Display or DSP too);
  2. your volumes are high enough that aggregation thresholds do not empty the results;
  3. you have concrete questions about budget and the role of each channel, not just curiosity;
  4. someone in-house or a partner can write and check queries (this is the work we do in our Amazon Marketing Cloud analysis projects).

If you only spend on Sponsored Products across a handful of SKUs, the best returns are still in campaign structure and basic reporting. AMC comes later, once the picture gets bigger.

AMC Expert Tool: Amazon Marketing Cloud in plain language

At Amazon unBoxed 2026, Amazon announced the AMC Expert Tool. It takes a question written in plain language, for example “which campaigns did new-to-brand customers see before their first purchase”, and turns it into Amazon Marketing Cloud SQL. Amazon has not stated when it will be available or in which markets, so check your own instance.

It lowers the biggest barrier to AMC, which has always been SQL. It does not remove the work around the query:

  • The question still has to be the right one. A clean query on a vague question gives a precise answer to the wrong thing.
  • Time windows and attribution still need checking. The tool writes the SQL, you decide whether its logic fits your decision.
  • Aggregation thresholds still apply. Small products and narrow segments still come back empty.
  • Totals still need reconciling with the console reports before anyone moves budget.

Our view: it makes AMC faster for people who already know what to ask, and it makes confident wrong answers easier for people who do not. Use it to draft queries, then review them like any other.

How to start without getting lost

The approach that works best starts from one business question, not a list of queries. A process we use:

  1. Pick a decision you need to make, for example whether to grow the DSP budget or move it to Sponsored Brands.
  2. Turn it into a measurable question: do users exposed to DSP convert more when they also see Sponsored Brands?
  3. Start from the sample queries and adapt them instead of writing from scratch.
  4. Reconcile the totals with the console reports. If they are far apart, there is an error to find.
  5. Turn the result into an action: an audience, a frequency cap, a budget shift.
  6. Rerun the same analysis a few weeks later to see whether the decision worked.

Example: a brand running DSP and Sponsored Products finds that users exposed to both convert far more often than those who see only one. A sensible decision is to build an audience of DSP-exposed users and raise Sponsored Products bids on them. The outcome then has to be checked, not assumed.

What to do now

Check whether AMC is already active in your ads console and look at the sample queries on offer. Write down the two or three questions about your advertising that standard reports cannot answer today. If at least one is about how formats interact or about customer value over time, AMC is the right tool for the job.

Frequently asked questions

What is Amazon Marketing Cloud used for?

Amazon Marketing Cloud is a clean room where advertisers analyze how their Amazon campaigns work together: conversion paths, overlap between formats, frequency, new-to-brand customers and time to conversion. The same queries can then become audiences for DSP and Sponsored Ads.

Does Amazon Marketing Cloud cost anything?

There is no base fee for eligible advertisers: you switch AMC on from the ads console. Some extra data sets or features may come with different terms, so check in your own console. The real cost is the time of whoever writes and interprets the queries.

Do you need SQL to work with AMC?

Not for basic analyses: sample queries and no-code templates cover them. For custom work, and to verify results, SQL makes a big difference. Many brands lean on a partner for that part.

What is the AMC Expert Tool?

It is a tool Amazon announced at unBoxed 2026 that turns natural-language questions into Amazon Marketing Cloud SQL. Amazon has not stated its availability yet. It speeds up query writing, but the results still need to be checked against the console reports.

Does AMC work for brands selling as a Vendor?

Yes. AMC runs on advertising data and purchase events on Amazon, whether the brand sells as a Seller or a Vendor. What counts is the volume of campaign data, not the selling model.