Amazon Marketing Cloud (AMC) is a clean room: you send it questions, it runs them against your Amazon Ads signals inside Amazon’s own environment, and it returns aggregated, anonymous answers. It is not a dashboard, not a customer list, and not a data export. Amazon describes it as “a secure, privacy-safe, and cloud-based clean room solution in which advertisers can easily perform analytics and build audiences across pseudonymized signals,” and states that it is “available at no cost to eligible advertisers via web-based UI and API” (Amazon Ads, data checked 2026-08-12).
Two questions decide whether AMC is worth any of your time: can you actually get in, and what would you do with it if you did. This guide answers both from Amazon’s published pages, marks clearly where Amazon publishes no answer, and closes with five specific analysis questions worth taking to AMC — each paired with the campaign decision it feeds.
What “Clean Room” Means in Practice
A clean room is a place where two parties’ data can meet without either party seeing the other’s raw records. In AMC’s case, the two parties are Amazon and you.
Amazon’s stated rules for that environment, quoted from its product page (data checked 2026-08-12):
- Inputs are pseudonymized. “AMC only accepts pseudonymized inputs, and all information in your AMC instance is handled in strict accordance with Amazon’s privacy policies.”
- Outputs are aggregated. “You can only access aggregated and anonymous outputs from AMC.”
- Your uploads stay yours. “The information you choose to upload stays within your dedicated AMC instance, and cannot be exported or accessed by Amazon.”
One nuance that secondhand explanations get backwards: Amazon says “advertisers can query event-level and user-level signals across sources using custom metrics and attributes of their choice, with built-in aggregation thresholds ensuring user privacy.” The querying happens over user-level signals; the reading out does not. You can ask how many people who saw a display ad later clicked a Sponsored Products ad, and get a number. You cannot ask who they were.
Amazon confirms thresholds exist but does not publish the numeric floor on that page. Practitioner write-ups commonly cite roughly 100 unique users per result — third-party information, not an Amazon figure.
What Data You Actually Get
Two categories of signal, per Amazon’s product page (data checked 2026-08-12).
| What you get | Amazon’s wording | What it means for you |
|---|---|---|
| Your Amazon Ads campaign signals | “hundreds of fields about their Amazon Ads campaigns … including ad-attributed impressions, clicks, and conversions” | Event-level records of your own advertising, joinable across ad types rather than siloed per report |
| Your own uploaded signals | advertisers can “upload their own pseudonymous inputs … and join the inputs with Amazon Ads campaign signals for analysis” | Your CRM or subscriber list, matched against campaign exposure — without either side handing over raw records |
| Lookback | “Access 25 months of ad traffic signals” | Just over two years — long enough for year-over-year comparison, not long enough for anything older |
| Optional add-ons | Amazon offers subscription-based Paid Features inside AMC (announced on Amazon Ads’ What’s New pages) | Extra datasets exist beyond the free baseline; pricing is not published, so ask your Amazon Ads contact |
What AMC will not do
Being explicit about the ceiling saves more time than the feature list does:
- No personally identifiable output. Names, addresses, and individual purchase histories are outside what it returns, by design.
- No small-slice answers. Aggregation thresholds mean a narrow-audience query comes back blank rather than wrong. Low ad volume is a hard limit, not a tuning problem.
- Not a bidding console. AMC is where you analyse and build audiences; the action still happens in Amazon DSP or your sponsored ads campaigns.
- Not shopper-search data. Search term and market-share reporting lives in Amazon Brand Analytics inside Seller Central. AMC is ads-side: your campaigns, not the category’s search behaviour.
- Nothing older than 25 months.
Who Can Access It
This is the part most secondhand articles get wrong, usually by asserting a spend minimum that Amazon does not publish. Here is what Amazon’s product page actually says about access routes (data checked 2026-08-12):
| Your situation | Amazon’s stated route |
|---|---|
| Registered with Amazon DSP | “reach out to their Ad Tech Account Executive to request access” |
| Registered with sponsored ads | “can access AMC via self-service” |
| Working with an agency or tech partner | “Advertisers working with partners registered on the Amazon Ads Partner Network can also utilize AMC via their partners” |
Three things follow directly.
DSP is not a prerequisite. The sponsored-ads self-service route is stated on the same page as the DSP route. If you have only ever run Sponsored Products, the door is not automatically closed.
Cost is not the barrier. AMC is “available at no cost to eligible advertisers.” Optional Paid Features are a separate subscription layer on top.
“Eligible” is doing the work — and Amazon does not define it publicly. No minimum spend figure appears on the product page. Anyone quoting you a specific dollar threshold for AMC access is quoting an unpublished number. The only reliable way to find out where you stand is to confirm eligibility with your Amazon Ads representative or your agency, and ask what “eligible” means for your account, in your marketplace, this month.
Gate 1 — Do You Have Enough Ad Volume?
Aggregation thresholds are the real entry test, and they bite before eligibility does. If your monthly ad-attributed conversions number in the dozens, most interesting slices — by campaign, by audience, by exposure sequence — will fall below the floor and return nothing usable. AMC rewards volume. Below it, your existing campaign reports already contain everything the data can support.
Gate 2 — Do You Have a Question Standard Reports Can’t Answer?
AMC returns answers to questions you pose. It does not proactively tell you anything. If your open questions are “what’s my ACoS on this campaign” or “which keyword converts,” the advertising console answers those already — see ACoS and TACoS vs ACoS for what standard reporting already covers. AMC earns its place only on questions that require joining signals across ad types or across time: overlap, sequence, lag, incrementality.
Gate 3 — Which Access Route Is Actually Open to You?
Before planning any analysis, establish your route: self-service via sponsored ads, request via an Ad Tech Account Executive, or through a partner on the Amazon Ads Partner Network. This determines your timeline more than anything else. A self-service path may be a login away; a managed request runs on Amazon’s calendar; a partner route means the analysis capability is rented, not built.
Gate 4 — Who Will Write and Read the Queries?
AMC has historically been a SQL environment, and Amazon now also advertises a no-code path — “Generate, visualize, and act on insights with a few clicks and no SQL needed” (data checked 2026-08-12). Templates lower the floor for getting a result. They do not lower the floor for interpreting one. Someone has to decide whether a difference between two cohorts is a real effect or a selection artefact. If nobody on your side owns that judgement, the queries will run and the findings will sit.
Gate 5 — Would the Answer Change What You Do?
Write down the decision before you write the query. “If overlap between display and search exposure is above X, I will cut display frequency.” If you cannot finish that sentence, the query is entertainment. This is the gate that disqualifies most AMC projects, and it costs nothing to apply.
Five Questions Worth Taking to AMC
Each of these is a question standard reporting cannot answer, stated with the signals it needs and the decision it feeds. Field availability varies by instance and by marketplace — confirm the exact fields in your own AMC schema before scoping the work.
1. Which campaigns actually bring in first-time buyers? Signals: ad-attributed conversions joined across campaign types, split by whether the purchaser had prior brand purchases in the window. Decision: stop judging upper-funnel campaigns on blended efficiency. If display or video is your main source of new buyers, cost per acquisition is the number to manage, not ACoS.
2. How much do your ad types overlap on the same people? Signals: exposure records across Sponsored Products, Sponsored Brands, and display, for converters and non-converters. Decision: if most converters were reached by two or three ad types, part of your spend is buying the same person twice. Trim the most duplicative line — not the least efficient-looking one.
3. At what exposure frequency does response stop improving? Signals: conversion rate by number of exposures per user, aggregated. Decision: set DSP frequency caps where the curve flattens. One of the few AMC outputs that maps to a single adjustable setting.
4. How long is the lag from first exposure to purchase? Signals: time between first ad-attributed impression and conversion, distributed rather than averaged. Decision: match your evaluation windows to the real lag. If purchases typically close three weeks out, judging a campaign at day seven guarantees you kill working campaigns.
5. Which campaigns reach existing customers rather than new ones? Signals: your uploaded pseudonymous customer list joined against campaign exposure. Decision: suppress or deliberately target. Retargeting buyers of a consumable is sound; doing it accidentally on a one-time purchase is waste. AMC is where you find out which one you are doing.
Common Mistakes
- Treating AMC as a better dashboard. It answers questions; it does not surface them. Arrive without a question, leave without a finding.
- Assuming DSP is required. Amazon’s own page lists a sponsored-ads self-service route alongside the DSP route.
- Repeating unpublished spend minimums. Amazon publishes no AMC access threshold. Circulating figures are inference; ask your rep.
- Expecting user-level output. You query user-level signals; you read out aggregates. Any plan that depends on identifying individuals is dead on arrival.
- Slicing too thin. Narrow segments return blanks. That is the privacy floor doing its job, not a bug.
- Confusing AMC with Brand Analytics. Different consoles, different data: one is your advertising, the other is category search behaviour.
- Buying an agency AMC retainer first. Establish the question and the decision it feeds, then buy the capability to answer it.
The AMC Readiness Checklist
Work down this list before requesting access:
- Monthly ad-attributed conversions are high enough that segment-level slices will clear aggregation thresholds.
- You have at least one written question that standard advertising reports cannot answer.
- For that question, you have written the decision rule it feeds, including the threshold that would change your action.
- You know your access route: sponsored-ads self-service, DSP via an Ad Tech Account Executive, or a partner on the Amazon Ads Partner Network.
- You have confirmed current eligibility with an Amazon Ads representative rather than with a published figure.
- A named person owns both running the query and interpreting it.
- If you plan to upload first-party signals, you have confirmed the pseudonymisation requirements before building the file.
Sources
- Amazon Ads, Amazon Marketing Cloud product page — all quoted definitions, access routes, cost statement, signal descriptions, and the 25-month lookback. Data checked 2026-08-12.
- Aggregation threshold figures (~100 unique users) are third-party practitioner reporting, not an Amazon-published number.
Frequently Asked Questions
What is Amazon Marketing Cloud in plain terms?
It is a clean room where your Amazon Ads signals can be queried in ways standard reports do not allow, returning only aggregated, anonymous results. Amazon calls it “a secure, privacy-safe, and cloud-based clean room solution” (data checked 2026-08-12). You bring questions; it returns numbers, never people.
Do I need Amazon DSP to use AMC?
No. Amazon’s product page states that advertisers registered with sponsored ads “can access AMC via self-service,” alongside a separate route for DSP advertisers via an Ad Tech Account Executive (data checked 2026-08-12). If you are weighing DSP separately, see the Amazon DSP guide.
How much does Amazon Marketing Cloud cost?
Amazon states AMC “is available at no cost to eligible advertisers via web-based UI and API” (data checked 2026-08-12). Optional Paid Features are offered as a separate subscription; Amazon does not publish their pricing, so ask your Amazon Ads contact.
Is there a minimum ad spend to qualify for AMC?
Amazon’s product page does not publish one. The qualifying word is “eligible,” which Amazon does not define publicly on that page. Any specific dollar threshold you see quoted is unpublished third-party inference — confirm your own eligibility with your Amazon Ads representative or agency.
Can I see individual customer data in AMC?
No. Inputs are pseudonymized and, in Amazon’s words, “you can only access aggregated and anonymous outputs from AMC” (data checked 2026-08-12). Queries can run over user-level signals; what you read out is aggregated, with built-in thresholds enforcing that.
How far back does AMC data go?
Amazon states you can “Access 25 months of ad traffic signals” (data checked 2026-08-12) — just over two years, which supports year-over-year comparison but not longer-horizon history.
Conclusion
AMC’s barrier is not price and, for many sellers, not access either — the sponsored-ads self-service route is stated on Amazon’s own page, and the product carries no published cost. The real barriers are volume and intent: not enough ad-attributed events and the privacy thresholds return blanks; no written decision rule and the answers change nothing.
So the sequence is: write the question, write the decision it feeds, confirm your eligibility and route with Amazon rather than with a blog post, then request access. If you are still deciding how much of your ad budget belongs in upper-funnel channels at all, work through Amazon PPC strategy and TACoS vs ACoS first — those decisions do not need a clean room, and making them well is what gives AMC something worth measuring.