Overview

Amazon Marketing Cloud (AMC) has introduced a paid data set named Amazon Retail Purchases that expands the purchase‑history window from 13 months to a full five years. The change, now live on the AMC UI and API across several marketplaces, gives advertisers the ability to compute true lifetime value, refine audience segments, and verify whether ad spend is attracting genuinely new shoppers. Sellers who rely on long‑cycle products should pay close attention because the deeper data can reshape budgeting and targeting decisions.

Key Points

• Five‑year look‑back — The new data set replaces the former 13‑month limit, allowing brands to trace buying patterns and seasonal spikes across up to five calendar years.
• Order‑level granularity — Each record contains SKU, purchase timestamp, and transaction amount, enabling analysis of the complete customer journey from first click to final checkout.
• Extended audience creation — Marketers can now build segments such as “customers who bought a starter model three years ago and are due for an upgrade” or “high‑spending buyers who have been inactive for 18 months.”
• Broader attribution windows — Campaign performance can be linked to purchases that occurred up to five years after an ad impression, offering a more realistic view of long‑term ROI.
• Seamless AMC integration — The dataset plugs into existing AMC dashboards and API calls, so teams do not need to redesign their analytics pipelines.
• Pricing model — After a 60‑day free trial, access costs $500 per month; however, using the data solely for audience creation remains free for eligible advertisers.

How Amazon Marketing Cloud Works

1. Data ingestion — AMC pulls anonymized purchase records from Amazon’s retail systems, covering the past 60 months. For example, a kitchen‑appliance brand can retrieve every dishwasher sale from January 2019 through December 2024, complete with SKU and purchase date.
2. Normalization and enrichment — The platform aligns each transaction with the advertiser’s product catalog and adds derived fields such as “time since last purchase.” A consumer who bought a coffee maker in March 2020 and a replacement in July 2023 will appear with a 39‑month interval, useful for replacement‑cycle modeling.