Brands that advertise on Amazon have long lacked visibility into customer history beyond a 13-month window—so campaigns that drive longer-term value looked weaker than they really were. Amazon’s Retail Purchases dataset, introduced in 2025 and made free through the end of 2026, expands Amazon Marketing Cloud (AMC) shop-history to as much as five years and changes how marketers measure acquisition, LTV and re‑engagement.
That matters because extending the time grain changes who counts as “new,” reveals multi-year purchase patterns, and creates actionable audiences for winback and upsell strategies. The dataset is both an analytics upgrade and an activation enabler, but it requires measurement discipline and validation of activation limits before you scale budget based on the new signals.
What changed technically — and what it unlocks
AMC is a privacy-safe clean room for Amazon advertising events; it was designed to move attribution beyond last-touch. Until recently, AMC limited shopping-event queries to a 13-month rolling lookback, which compressed metrics such as new-to-brand (NTB) and lifetime value (LTV) into short windows.
The Retail Purchases dataset extends the available shopper-history window to as much as five years. That longer time grain improves accuracy for multi-year behaviors and supports audience activation: segments built from longer purchase histories can be exported and used across Amazon-owned inventory and supported third-party supply.
Six practical use cases marketers should prioritize
1. Product-level lifetime value (LTV)
Five years of purchase history lets advertisers calculate LTV at SKU level instead of relying on short-term revenue spikes. That matters when low-priced or introductory SKUs act as gateways to higher-margin purchases later. With clearer LTV benchmarks, teams can set acquisition cost targets and allocate budget by SKU rather than by imperfect short-term signals.
2. True new-to-brand (NTB) measurement
For categories with long repurchase cycles—consumer electronics, premium goods, some CPG subcategories—the previous 13-month window undercounted net-new customers. With up to 60 months of data, brands can choose an NTB lookback that fits their product lifecycle (for example 12, 24 or 36 months) and compare how acquisition rates shift under the longer definition.
3. Identifying repeat-purchase windows
Extended histories expose typical repurchase cadences that exceed a year. Those insights let teams schedule retargeting and replenishment windows to match real customer behavior rather than an artful guess, and create audiences timed to the most effective post-purchase moments.
4. Mapping gateway products
SKU-level timelines show which entry products tend to precede later purchases of premium SKUs. That informs which SKUs you should push to new shoppers, which to discount to accelerate trial, and which purchasers to prioritize for upsell campaigns.
5. Reacquiring lapsed customers
Previously, brands could not reliably target purchasers whose last buy was beyond 13 months inside AMC segments. The five-year dataset enables winback cohorts filtered by recency, minimum LTV, or purchased SKUs and lets advertisers activate those audiences across Amazon DSP and other eligible inventory.
6. Joining first-party data for cross-channel insight
AMC supports secure hashing and upload of first‑party shopper lists to join with Amazon purchase histories. Combining multi-year Amazon data with DTC purchase records or CRM entries lets brands measure how Amazon influences off‑Amazon behavior—and vice versa—while preserving privacy through hashed identifiers.
Activation, inventory reach and practical constraints
The value of longer histories depends on your ability to act on them. AMC-derived segments can be activated across Amazon-owned inventory (Amazon.com, Prime Video including live sports, Twitch, Whole Foods) and some third‑party placements Amazon supports. Amazon Ads also exposes DSP targeting and sponsored formats that can incorporate AMC audiences. To lower the analytics barrier, Amazon offers some pre-built query templates, but most implementations still require SQL or analytics support.
The public write‑ups leave several operational details unclear: exact activation pathways, audience-size minimums, refresh cadence, and privacy controls for joined first-party data. Teams should confirm these constraints and any current access costs with their Amazon Ads account team before redesigning measurement or activation strategies around the five-year dataset.
Note: the Retail Purchases dataset was introduced in 2025 and, as of June 2026, was made available without cost through the end of 2026. Confirm current pricing and access windows with Amazon Ads when you plan experiments.
What marketers should do next
1) Revisit NTB and LTV definitions. Recalculate those metrics using the extended lookback and run A/B comparisons to see how acquisition rates and cost targets shift.
2) Audit SKU funnels. Use multi-year SKU histories to identify gateway products and set acquisition priorities by expected downstream revenue.
3) Build and test winback cohorts. Create segments for lapsed customers beyond 13 months and pilot DSP or display campaigns with tailored creative and time-limited offers.
4) Test first‑party joins cautiously. If you plan to upload hashed DTC data, verify matching rules, update cadence and privacy guardrails to ensure reliable joins and compliance.
5) Confirm activation limits. Before reallocating budget, verify audience activation paths, minimum sizes, refresh frequency and any dataset costs with your Amazon Ads representative.
Extended purchase histories are a measurement-first lever: they tell you which acquisition plays are genuinely net-new, which SKUs seed long-term value, and which lapsed customers are worth pursuing. Treat the five-year dataset as an opportunity to reframe experiments—not to replace disciplined testing with optimism—and prioritize proofs of value that drive activation-ready audiences.
