View-Through Attribution Windows Compared

Why this comparison is worth doing surface by surface

Treating "Amazon's attribution window" as one number is the single most common inaccuracy in this category, and it's an easy one to fall into because most explainers quote a single figure — usually 14 days — without naming which ad product or billing type it applies to. The real picture is a table, not a number, and it changed materially on January 1, 2026, which means content written before that date is now describing a system that no longer exists in the form it describes.

Sponsored Products: click only

Sponsored Products has no formal view-through attribution — it's a click-attributed model. The click window itself runs 7 days for sellers and 14 days for vendors. If a shopper sees a Sponsored Products ad and doesn't click, and buys later, that purchase isn't credited to the ad through a view-through mechanism the way it can be for Sponsored Brands, Sponsored Display or DSP.

Sponsored Brands and Sponsored Display

Click attribution for both runs a 14-day window, unchanged by the January 2026 update. View attribution is where the update matters: for campaigns billed on a viewable-impression (vCPM) basis specifically, Amazon replaced the blanket 14-day view window with a shopping-signal enhanced last-touch model — shorter, and judged on whether the view plausibly influenced the purchase rather than simply preceding it. Campaigns on other billing structures within these ad types aren't described as affected by Amazon's own update page.

Amazon DSP: two different stories depending on placement

This is the split most explainers miss entirely. Amazon's own update names "Amazon DSP showing in-store ads" as affected by the January 2026 change — meaning DSP inventory that serves on Amazon's own retail surfaces moved to the same shorter, ML-judged view model as Sponsored Brands and Sponsored Display vCPM. DSP inventory served offsite — the open web, third-party apps, exchanges — isn't named in that update, and by omission continues to operate under the attribution structure DSP has used historically, generally described as a 14-day click and view lookback. If you're comparing DSP performance before and after January 2026, check which supply source you're looking at before assuming the comparison is apples to apples.

Amazon Attribution: a separate system entirely

Amazon Attribution — the tool that measures off-Amazon marketing channels' impact on on-Amazon activity — uses its own 14-day last-click model, distinct from the sponsored ads and DSP windows above. It's worth not confusing the three: Amazon Attribution measures external channels driving to Amazon, sponsored ads and DSP measure Amazon's own ad products, and none of them share one universal window across the board.

This matters specifically for brands running Google, Meta or email campaigns pointed at an Amazon listing alongside their own sponsored ads and DSP spend. A shopper's path might genuinely include a Google ad, a DSP display impression and a branded search click, and each of those three touches is measured by a different tool, on a different window, with different rules about what counts. Reconciling that full picture into one honest read of the path is a job for Amazon Marketing Cloud, not for stacking three separate dashboards on top of each other.

The common mistake, including ours

The mistake is quoting one attribution window figure for "Amazon" as a whole, without naming the ad product, the billing type, or whether the click date was before or after January 1, 2026. We've had to correct our own client materials that said "Amazon uses a 14-day view window" as a blanket statement — true for some surfaces, materially wrong for others post-update, and it's exactly the kind of imprecise claim that reads as authoritative right up until someone checks it against Amazon's own current documentation.

The safest habit going forward is to name the specific ad product and billing type every time an attribution window appears in a report or a client conversation, rather than referring to "Amazon's attribution window" as a single fact. It's a small amount of extra precision in the sentence and it's the difference between a claim that survives scrutiny and one that doesn't.

Why this matters more than it looks like it should

Attribution window mismatches quietly break more than one kind of analysis. A multi-touch or path-to-conversion query that spans channels with different windows will systematically undercount touches from the shorter-window channel relative to the longer-window one, making it look like it contributed less to converting paths than it actually did — not because it performed worse, but because its reporting window closed sooner. Any cross-channel comparison — deciding whether DSP or Sponsored Products deserves more budget, for instance — needs to account for this before treating the two numbers as directly comparable.

What to do when your reporting doesn't match this table

If a number you're seeing doesn't match what's described here, check the report date first — a dashboard pulled before January 1, 2026 and one pulled after are describing genuinely different attribution logic for the affected surfaces, not the same metric with different results. If the dates are consistent and the numbers still don't reconcile, confirm the billing type on the specific campaign, since the update applies to vCPM-billed Sponsored Brands and Sponsored Display specifically, not every campaign under those ad type names.