An audience definition starts with a behavior and an exclusion
An AMC audience is useful when its membership matches the decision you want to make. A cart abandoner added an item but did not purchase within the chosen window. A lapsed customer bought before but has passed the expected repeat window. A high-value seed selects buyers using a stated value rule. Those definitions need different lookbacks and exclusions; one generic purchaser query cannot stand in for all three.
| Your decision | Define before querying | Supporting method |
|---|---|---|
| Revisit an unfinished purchase | Selected ASINs, cart window, purchaser exclusion | Cart-abandoner audience |
| Re-engage a previous buyer | Expected repeat interval and recent activity to exclude | Lapsed-customer audience |
| Find more customers like valuable buyers | Value measure, seed strategy, expansion type | High-value segmentation |
| Prepare a seasonal lookalike | Seed behavior, activation timing, destination | Prime Day lookalike workflow |
The guides develop selected audience workflows, not an exhaustive catalog of every AMC audience type. For access and clean-room fundamentals, use What is Amazon Marketing Cloud?.
Why does sizing come before activation?
A behaviorally sensible seed may still be too small to activate. Count distinct users using the same membership logic, check the applicable audience requirements, and leave room for suppression and normal audience decay. Changing the window to reach a threshold also changes who belongs in the audience, so review the definition again rather than silently broadening it.
The activation approval guide puts the SQL, size, date window, refresh cadence, and destination in one review. Creation success in AMC and an active audience in DSP are separate states. Confirm the destination status before treating the audience as ready for a campaign.
How do you know whether a lookalike is worth keeping?
Compare it with the rule-based audience it is intended to replace, using aligned periods and definitions. The lookalike evaluation guide considers new-to-brand share, reach, and acquisition cost before a keep, rotate, or rebuild decision. A higher conversion rate alone does not settle whether the expansion helped; an observational comparison does not establish causal lift.
This creates a repeatable loop: define membership, size it, review the activation request, verify delivery, then evaluate the result. AMC measurement covers related interpretation questions, while AMC SQL and modeling covers query construction and validation. The AMC software overview explains how Kuudo supports these workflows under your approval policy.