AMC playbook: Programmatic audience framework
In short: The framework automates the tedious part of creating audiences at scale. Instead of one large audience it creates many similar AMC audiences whose line item performance can be measured precisely. Amazon splits it into before, during and after the campaign: create, refresh every 30 days, measure the result.
What it is about
Many small, clearly delimited audiences, such as "added to cart but did not check out" per ASIN, can be assessed more cleanly than one large audience with many ASINs. The playbook covers creation, measurement, optimisation and analysis with custom AMC audiences and Amazon DSP.
Which questions the playbook answers
- Before the campaign: how do I quickly create many custom AMC audiences and see which are ready for activation?
- During the campaign: how do I monitor the audiences and adjust spend based on performance and refresh cadence?
- After the campaign: which audiences were most effective, which do I continue?
Flow
- Pre campaign: create and monitor audiences programmatically through the custom audiences API.
- Mid campaign: refresh every 30 days, update audience sizes after successful refreshes, adjust spend.
- Post campaign: deploy dashboards for overall performance and ROAS.
Prerequisites according to Amazon
- AMC Insights on AWS is deployed and Workflow Manager (WFM) or a similar solution is configured.
- AMC SQL, Python, AWS basics (SageMaker, Athena) and basic Amazon DSP knowledge.
- The knowledge to build the data flow pipeline for the automated optimisation.
Our take
The effort lies in operation, not in the idea: the framework presumes a running pipeline. To get started the audience recipes and minimum sizes are enough; the framework pays off when you actually produce audiences in series.
Amazon documents the playbooks in technical English with SQL and Python. This page summarises what the playbook does; the queries and scripts are in the original.