Retail Media & Commerce

Amazon Gives Away 5-Year Purchase Data in Marketing Cloud

Amazon's five-year AMC purchase dataset, free since June 2026, replaces the 13-month lookback — unlocking product-level LTV, true new-to-brand rates and lapsed-customer winback segments.

Amazon Marketing Cloud’s 5-Year Dataset: 6 Use Cases Worth Building Now
Amazon Marketing Cloud’s 5-Year Dataset: 6 Use Cases Worth Building NowAI-generated

Amazon made its five-year Amazon Retail Purchases dataset free in AMC in June 2026, removing the monthly subscription fee advertisers had paid since the dataset launched in 2025. The move gives brands using Amazon Marketing Cloud up to 60 months of purchase history to work with, up from the 13-month rolling lookback that had constrained the clean room since Amazon launched it in 2021.

That 13-month limit mattered. New-to-brand rates, lifetime value and gateway-product analysis only registered whether a shopper had purchased within the prior 13 months, so earlier purchases and revenue dropped out of AMC queries entirely. For categories with long repurchase cycles — consumer electronics, luxury goods — the gap made several core metrics effectively unusable.

The five-year dataset changes what brands can measure and, just as importantly, what they can activate. Here are the use cases the extended lookback unlocks.

Lifetime value, down to the product. Vendors on Amazon previously had no way to calculate customer LTV at all, per the source. With 60 months of history, they can now model it at the product level and set customer acquisition cost targets accordingly. The source cites a concrete CPG example: brands may find that shoppers who buy a low-priced SKU later convert to higher-priced SKUs such as multi-packs. That insight can justify taking a loss on the first purchase and increasing ad spend, because the shopper brings incremental revenue over the longer term.

True new-to-brand rates. The 13-month window distorted NTB calculations for brands with repurchase cycles longer than a year, making it hard to know which products and tactics actually acquired net-new customers. The retail purchase dataset lets brands customize their NTB window up to the full five years — though the source notes several brands find five years too long and trim the lookback to fit their specifications.

Repeat purchase windows. Before this dataset, categories with extended customer lifecycles could not measure repeat purchase cycles that exceeded 13 months, and could not retarget past purchasers at all. Now they can identify their ideal repeat windows and build audiences targeting shoppers within a post-purchase time window, activated across Amazon's display network, Prime Video, sponsored ads and everything in between.

Gateway products. Once LTV and NTB are measurable by product, the next question is which products bring in customers who eventually buy higher-priced SKUs. For CPG brands that may mean bulk products or multipacks; for consumer electronics, an upgraded model. The extended lookback lets brands trace customer lifecycles across SKUs and decide which products to push to net-new shoppers, promote, or use for retargeting.

Winning back lapsed customers. Amazon DSP custom audiences cap lookback windows at 365 days, and AMC purchase segments were historically limited to 13 months. A brand that lost a customer for more than a year had no way to attempt reacquisition, regardless of that customer's prior lifetime value. Because the retail purchase dataset supports audience activation as well as queries, brands can now build winback segments with custom lookback windows, minimum LTV thresholds and product-level filters.

Joining first-party data. The source calls this one of AMC's most underleveraged capabilities. Brands with shopper-level purchase history from a direct-to-consumer site or cross-channel email list can hash that data, upload it to AMC and join it to Amazon datasets via privacy-safe shopper IDs. The result: a cross-platform view of shopper lifecycle, showing how Amazon affects DTC purchasing and vice versa.

The stakes extend beyond measurement. Amazon's ad network now spans Amazon.com, Prime Video — including live NBA and NFL sports — Twitch, Whole Foods and third-party supply such as Roku and Netflix. Precise, long-lookback audience segments activated across that footprint are worth more than the same segments were a year ago.

Amazon is also lowering the barrier to using AMC at all. Workflows that previously required SQL expertise now run through a library of pre-built queries and audiences, and Amazon is adding agentic functionality via Amazon Ads Agent. With the five-year dataset free through at least the end of 2026, the source's closing argument is direct: advertisers on Amazon who have not activated AMC data are leaving the network's most valuable dataset idle.

Original: advertising.amazon.com

Share this article:

Amara Osei

Show full bio

News editor covering industry trends and analytics at Marketing Herald.

More from Amara Osei