Your advertising reports have always shown you the customers your ads reached. They have never shown you the customers who bought anyway.
That gap sounds academic until you try to answer a budget question with it. How much of your branded search revenue would have happened without the spend? Which products acquire customers organically and are being propped up by ads they do not need? What actually happened to the thousands of first-time buyers you picked up during Prime Day?
Standard advertising reporting cannot answer any of those, because it is built from ad impressions and clicks and therefore only contains people your ads touched. The customers who found you some other way are invisible to it.
Amazon's clean room has been able to close that gap for a while, behind a subscription. Reporting indicates the first-party portion of that paywall came down in June, through the end of December.
What AMC Actually Is
Amazon Marketing Cloud is a clean room. That term does more obscuring than explaining, so here is the plain version.
It is a secure environment holding pseudonymized event-level signals from your Amazon advertising, alongside your own first-party data if you upload it. You write queries against that data and receive aggregated results. No personally identifiable information is shared, you do not get raw records out, and Amazon does not access your uploaded data for its own purposes.
Clean room. A controlled analytical environment where two parties can combine sensitive datasets and query them jointly without either party gaining access to the other's underlying records. Outputs are aggregated and anonymized. In advertising it lets a brand join its own customer data to a platform's event data to answer questions neither dataset could answer alone, without either side handing over identifiable information.
The Amazon Ads product page describes AMC as available to eligible advertisers, with access depending on your advertising setup and region. AMC itself has not generally carried a license fee. What carried a fee was the premium signal tier on top, and that is what changed.
What Opened, And What Did Not
Precision matters here, because "AMC is free now" is circulating and is broader than what actually happened.
Reporting indicates Amazon announced in June 2026 that Amazon first-party paid feature signals inside AMC are available at no additional cost, effective immediately through December 31, 2026, with advertisers subscribing through the paid features page to begin running measurement queries and building audiences with those signals.
| What | Status In This Window | Note |
|---|---|---|
| AMC base access | Available to eligible advertisers | Depends on your ads setup and region. |
| Amazon 1P paid features | No additional cost through Dec 31, 2026 | You still have to opt in via the paid features page. |
| Third-party datasets | Still paid | Reporting names Experian as an example. Unchanged. |
| Your analyst time | Still the actual cost | This is where the real spend has always been. |
The opt-in detail is the one that quietly loses brands the window. This is not applied automatically. Reporting indicates you subscribe through the paid features page, so a brand that reads a headline, thinks it is now included, and never enrols will reach December having captured nothing.
Before anything else in this article, check whether your AMC instance has the paid features enabled. That takes minutes and it is the step that makes the rest possible. Everything after it is analysis you can schedule.
The Half Of The Journey You Could Not See
This is the section that explains why anyone should care, and it is buried in most coverage under a features list.
Standard AMC reporting tracks ad-attributed shoppers. It is assembled from ad impressions and clicks, so by construction it contains only people your advertising reached. Everyone who bought your product without ad exposure is absent.
Every advertising report you have ever read describes the people your ads touched. It cannot tell you about the customers who would have bought anyway, which is exactly the group that determines whether your spend is working.
The premium first-party signals change that. They let you see organic purchasers who were never ad-exposed, which unlocks a genuinely different class of question:
- Which products acquire customers without advertising? If a SKU pulls organic first-time buyers steadily, the ad spend behind it may be buying customers you already had.
- What happened to your event cohorts? Prime Day generates large volumes of organic purchase, and historically those customers vanished from view after the event because they were never ad-attributed.
- Do paid and organic customers behave differently afterwards? Repeat rate, subscription uptake, and category expansion by acquisition source is a question you could not previously ask.
- Who has lapsed? Building lapsed-customer audiences requires seeing customers regardless of how they arrived.
That last one connects to lifetime value work directly. Our customer lifetime value guide covers the modeling; the constraint on Amazon has always been that you could only see the ad-attributed slice, and this window relaxes it.
The SQL Objection Is Out Of Date
The standard reason mid-size brands skipped AMC was that it required SQL and a data person. That was true and it is no longer the whole picture.
Amazon introduced a generator turning plain-English requests into AMC queries, removing the requirement to write SQL by hand to get started.
Self-service access opened for sponsored ads advertisers, removing the traditional requirement to go through a partner or Amazon representative.
No-code templates, curated analytic and audience templates, and an in-product assistant that helps build queries and troubleshoot errors.
Templates produce output. Deciding which question matters, whether the answer is trustworthy, and what to change remains human work.
The honest version is that the technical barrier fell and the analytical one did not. You can now produce an AMC result without writing SQL. Knowing whether that result should move budget is the harder skill and no template supplies it.
Which reframes the staffing question usefully. You do not need a data engineer. You need someone who understands your media and can think clearly about what a number means, and most brands already have that person.
The Lookback Window
A quieter change that matters for anything seasonal.
Reporting indicates that on November 11, 2025, at Amazon's annual advertising conference, the AMC ad traffic lookback window expanded from 13 months to 25 months in the United States and Canada, launched in open beta. One source describes the new figure as 24 months rather than 25, so treat the exact number as approximate and confirm in your instance.
Either way the practical effect is the same and it is significant. At 13 months you could just about compare a period against the same period last year, with no margin. At two years or more you can compare two full annual cycles, which is the minimum for distinguishing a trend from a one-off.
For seasonal brands this is the difference between knowing what happened last Q4 and knowing whether last Q4 was normal. If you have ever argued internally about whether a category is declining or simply had a soft year, this is the data that settles it.
Reporting also indicates Prime Video viewership signals arrived in open beta in November 2025 across several markets, which matters mainly to brands running streaming media and is a smaller story for most sellers.
The Three Queries To Run First
A brand starting from nothing with fourteen weeks should not attempt a comprehensive program. Three questions, each capable of changing a budget.
Each of these has a decision attached, which is the test any AMC query should pass before you run it. If you cannot name what you would do differently depending on the answer, the query is entertainment.
Deliberately absent is anything resembling a dashboard. A recurring report that nobody acts on is the most common AMC failure mode, and it consumes the analyst hours that are the real cost of the program.
Why Q4 Is The Right Quarter
The timing of this window is either coincidence or good design, and either way it works in your favor.
Q4 produces your largest volume of new customers, your highest concentration of organic purchase driven by event traffic, and your most extreme mix of paid and unpaid demand. It is the richest cohort data your brand generates all year, and the window covers exactly the period in which it is created.
There is an asymmetry worth understanding. The data is being generated now regardless. Whether you can analyze it with premium signals depends on being enrolled while it happens. A brand that enrols in January can still query historical periods, but a brand enrolled through Q4 can watch the cohort form and act during the season rather than after it.
The specific Q4 question worth planning for is what happens to event-acquired customers afterwards. Large numbers of first-time buyers arrive during peak, many of them organically, and most brands never find out how many came back. That is a retention question that determines how you value the event, and it needs the organic signals to answer.
Where AMC Changes A Budget
Concrete cases, because abstract capability does not persuade anyone to allocate hours.
Branded search spend. The perennial argument. AMC lets you look at how many branded-term converters had no other ad exposure and were already customers, which moves the discussion from opinion toward evidence. It is not a controlled experiment and does not settle causation, but it is considerably better than the ROAS number that currently ends the argument.
Upper-funnel justification. Display and video campaigns look poor on last-touch reporting. Path overlap analysis shows how often they appear earlier in converting paths, which is the argument for funding them.
Portfolio allocation. If two products show similar ROAS but one acquires meaningfully more new-to-brand customers, they are not equivalent investments, and standard reporting shows them as equivalent.
Audience building. Beyond measurement, the signals support constructing audiences such as lapsed purchasers or organic buyers who have never been targeted, which is the point at which analysis turns into media.
Where this connects to existing spend decisions, our guide to monthly spend by revenue tier covers the allocation question and the ad type decision framework covers where upper-funnel sits.
The Ecom Profit Box
Our collection of ecommerce growth resources, including the measurement frameworks behind this analysis.
Get It FreeWorth It For Your Account?
We do not sell AMC dashboards. We can tell you whether you have a decision that this data would actually change.
Book A CallThe DSP Caveat
An important limitation that promotional coverage tends to leave out.
Reporting indicates the richest AMC analyzes join DSP impression data with conversions, which means a DSP entity is required for a meaningful portion of what AMC can do. A brand running only sponsored ads gets real value, particularly on the organic and new-to-brand questions, and it gets a narrower version of the platform.
The path overlap analysis in particular becomes considerably more interesting once display impressions are in the dataset, because that is where the last-touch distortion is largest.
That does not mean adding DSP to unlock AMC. Buying a media channel to improve your analytics is backwards. It means calibrating expectations: if you are sponsored-ads-only, focus on questions AMC can answer with the data you have rather than pursuing analyzes that assume a channel you do not run. Our overview of Amazon DSP for growing brands covers whether DSP makes sense on its own merits, which is the right way to decide it.
Where It Is A Vanity Dashboard
Reporting from agencies working in AMC suggests most mid-size brands using it extract a small fraction of what is available. That is worth taking seriously rather than treating as a sales line, because the failure pattern is consistent and recognisable.
- Reports nobody reads. A scheduled query producing a weekly file that is opened once and never acted on. This is the default outcome without a named owner and a decision attached.
- Analysis with no counterfactual. AMC describes what happened. It does not tell you what would have happened otherwise, and treating a path analysis as proof of causation overstates it.
- Precision mistaken for accuracy. Event-level data produces confident-looking numbers. Confidence is not the same as being right about why something occurred.
- Analysis instead of decisions. The most common outcome. Brands generate insight, present it, and change nothing, because no budget decision was ever attached.
The discipline that prevents all four is the same: name the decision before running the query. If nothing changes based on the answer, the analysis is a hobby funded by hours you could spend elsewhere.
A 90-Day Plan
Roughly the time remaining. Deliberately modest, because the failure mode is over-scoping.
- Week 1. Enrol. Confirm AMC access and opt into the paid features. Nothing else matters if this does not happen.
- Week 1. Name the owner. One person accountable for the output. Not a team, not the agency by default.
- Weeks 2 to 3. Write the decisions down first. Three budget questions you would act on. Then find the queries that answer them.
- Weeks 3 to 5. Run query one. Organic purchasers with no ad exposure, by ASIN. Start with templates rather than custom SQL.
- Weeks 5 to 7. Run query two. Path overlap across ad types.
- Weeks 7 to 10. Run query three on new-to-brand and repeat behavior by acquisition source, ideally spanning your peak period.
- Weeks 10 to 12. Act. Move budget based on what you found, or document explicitly why you are not.
- Before December 31. Decide about next year. If the paid features return to a subscription, you will have evidence about whether they earned it.
Step eight is the underrated one. This window is also a free trial of a paid product. Running it properly gives you a documented basis for a renewal decision rather than a guess, and that is worth something independent of the analysis itself.
Who Should Skip This
Free is not the same as worth doing, and the hours are real even when the signals are not billed.
Skip if your fundamentals are not fixed. If your campaign structure is disorganised, your listings underconvert, or you cannot state your contribution margin per SKU, AMC will produce sophisticated analysis of a situation that needs simpler work. Fix the structure first. Our Amazon PPC strategy guide and the contribution margin playbook cover that groundwork.
Skip if no decision is waiting. The test is whether you can name a budget question whose answer would change what you do. If not, you will produce a report and file it.
Skip if nobody owns it. Analysis assigned to everyone is assigned to nobody. Without a named person with hours protected, this does not happen, and the free window is not a substitute for capacity.
Skip if you are below meaningful spend. The value scales with media budget. At small spend the decisions AMC informs are small decisions, and your attention pays better elsewhere.
Who Should Definitely Do It
Brands spending enough that a few percentage points of allocation efficiency is real money, running more than one ad type so path overlap means something, with someone able to own it and at least one budget argument currently being settled by opinion rather than evidence. If that describes you, the branded search question alone probably justifies the window, and it closes on December 31.
One more honest note. Nothing here needs an agency. The templates exist, the natural-language tooling exists, and the three queries in section six are a starting point you can work through yourself. We do this for clients, and we would rather you knew it was optional.
What To Remember
- Reporting indicates Amazon 1P paid feature signals in AMC are free through December 31, 2026, announced in June, and you must opt in through the paid features page rather than receiving it automatically.
- Third-party datasets such as Experian still require a paid subscription. Only the Amazon first-party signals are covered, so "AMC is free now" overstates it.
- Standard AMC reporting only covers ad-attributed shoppers. The premium signals open the organic side, letting you see purchasers who were never ad-exposed.
- The SQL objection is out of date. Natural-language query generation arrived in January 2025 and direct self-service access for sponsored ads advertisers followed in September 2025.
- The ad traffic lookback expanded from 13 months to a reported 25 months in the US and Canada in November 2025, enough to compare two full annual cycles rather than one.
- The richest analyzes require a DSP entity, so sponsored-ads-only advertisers get real but narrower value and should not add DSP purely to unlock analytics.
- Name the decision before running the query. The most common failure is producing insight nobody acts on, which consumes the analyst hours that are the program's real cost.
Where This Came From
- Amazon Ads, Amazon Marketing Cloud product page, for the description of clean room functionality, the expanded lookback window, Ads Agent, and no-SQL templates.
- Industry reporting on the June 2026 announcement that Amazon first-party paid feature signals are available at no additional cost through December 31, 2026, that advertisers subscribe through the paid features page, and that third-party signals such as Experian continue to require a paid subscription.
- Industry reporting on the January 2025 introduction of natural-language SQL generation, the September 16, 2025 opening of direct self-service access to AMC for sponsored ads advertisers, and the November 11, 2025 expansion of the ad traffic lookback window and launch of Prime Video viewership signals in open beta.
- Agency reporting that standard AMC reporting covers only ad-attributed shoppers built from impressions and clicks, and that the premium first-party datasets open analysis of organically acquired and previously invisible customers.
- Agency reporting that a DSP entity is required for the richest AMC analyzes joining display impression data with conversions, and that most mid-size brands using AMC extract only a fraction of its available value. Sources disagree on the lookback window figure, with most reporting 25 months and one reporting 24. None of these figures are confirmed on an Amazon-owned page beyond the product page cited above.