Overview

Amazon has introduced a machine‑learning‑driven “optimized targeting” option for Sponsored Display conversion campaigns. The feature automatically discovers high‑intent shoppers by analyzing real‑time shopping signals, and it is now live on 19 Amazon marketplaces. Sellers who adopt it can cut down on manual audience segmentation while still retaining the ability to fine‑tune targeting where needed.

Key Points

• Automated audience discovery — The algorithm surfaces shoppers who exhibit purchase intent, such as recent category browsing or similar‑product searches, without the seller having to create segments manually.
• Preserved manual controls — Advertisers keep full access to classic remarketing lists, including product‑view and purchase‑based audiences, as well as contextual placements.
• Hybrid ad‑group capability — Optimized and manual targeting can coexist in the same ad group, letting sellers test algorithmic recommendations against proven segments side by side.
• Streamlined launch process — Setting up optimized targeting only requires the basic campaign settings (budget, bid, and product list), dramatically shortening the time needed to go live.
• Global rollout — The feature is operational in North America, Europe, Asia‑Pacific and additional regions, covering a total of 19 marketplaces.
• API and console availability — Sellers can enable optimized targeting through the Amazon Advertising UI or programmatically via the Advertising API for large‑scale deployments.

How Optimized Targeting Works

1. Signal collection layer — Amazon continuously gathers shopper behavior data—such as recent searches, product‑page views, and purchase timelines—and feeds it into a real‑time model. Example: A buyer who searched for “wireless earbuds” and then viewed several audio accessories within the last 48 hours becomes a candidate for a campaign promoting a new Bluetooth speaker.
2. Conversion propensity modeling — The system evaluates multiple factors, including product affinity, category browsing depth, and cross‑shopping patterns, to assign a conversion likelihood score to each shopper.