AMC playbook: Lookalike audiences for promotional events
In short: The playbook shows how to build a seed audience of fast buyers from an earlier promotional event such as Prime Day and create an AMC lookalike audience from it. That reaches impulse buyers who are new to the brand. The event should be no more than 13 months back.
What it is about
Amazon writes that such events can account for 20 to 30 percent of annual revenue for some brands. Advertisers reach ad-exposed users through rule-based audiences; for non-exposed users only in-market and lifestyle audiences remain. Lookalike audiences close that gap by finding people who do not know the brand yet from the signals of existing customers.
The seed audience in this playbook is customers who converted quickly after an impression. The model builds an audience with similar signals from it that can be targeted in Amazon DSP.
Which questions the playbook answers
- Which ASINs and campaigns were used in earlier promotional events?
- When do you target audiences for promotional events (lead-in and lead-out phases)?
- How do you build an AMC lookalike audience for a promotional event?
- How do you target it: keywords, frequency and ad format?
Flow
- Identify event and ASINs: predetermined ASINs, or by analysing the campaigns and purchased ASINs of the last event.
- Create the seed audience: the time between first impression and purchase ("flight time analysis") decides who belongs in the seed group.
- Analyse the phases of the event.
- Evaluate the activation of the seed audience.
Prerequisites according to Amazon
- Completed onboarding for the AMC APIs and an Amazon DSP account.
- Historical data of a promotional event, preferably the one being planned; the event must have occurred within the last 13 months.
- Knowledge of AMC lookalike audiences, SQL and DSP campaigns.
Our take
Amazon's assumption: price promotions for the ASINs do not change. If the model was trained on buyers of a 50 percent promotion, it finds people who buy at big discounts. If you only offer 10 percent this year, read the result carefully.
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.