If your Amazon ad conversions look lower in 2026 than they did last year — but your actual orders held steady — you're not going crazy, and your campaigns probably aren't broken. On 1 January 2026, Amazon quietly changed how it counts a conversion. It retired the old blanket 14-day view-through window and replaced it with a machine-learning attribution model that reduced reported view-attributed conversions by an estimated 15–30% overnight. Same sales. Stricter scorekeeper.
This is one of the most misunderstood changes of the year, and it's quietly causing sellers to panic-cut budgets on campaigns that are actually working. Here's exactly what changed, why your numbers moved, and how to read your account correctly in the new era.
Until the end of 2025, Amazon used a simple, generous rule for view-through attribution: if a shopper merely saw a display ad and then bought anything from that brand within 14 days, the ad got credit for the sale. It didn't matter whether the ad actually influenced the purchase.
On 1 January 2026, Amazon retired that "blind" 14-day window and switched to a shopping-signal enhanced last-touch model — machine learning that evaluates whether an ad view genuinely influenced the purchase by analysing browsing behaviour, category-search patterns, and where the shopper was in their journey. The change applies across all Amazon Store ads: Amazon DSP, Sponsored Brands, and Sponsored Display.
The old model over-credited advertising. It handed ads the win for purchases they had little or nothing to do with — someone glancing at a display impression, then buying a week later for entirely unrelated reasons. The new model strips out that phantom credit. So the "drop" you see in your dashboard isn't lost revenue; it's the removal of conversions your ads were never really responsible for.
Here's the trap: a seller compares January 2026 to December 2025, sees "conversions down 20%," and slashes DSP or Sponsored Display budgets — killing campaigns that are genuinely driving sales. They cut spend based on a measurement artifact, not a performance problem. That's how a reporting change quietly becomes a revenue loss.
The right lens: check your total business. If total orders and revenue held steady while reported ad conversions fell, that's the attribution change, full stop. Your return on ad spend didn't really move — the ruler did.
Instead of a fixed time window, the model asks a smarter question at the moment of each ad view: did this actually nudge the shopper? It weighs real signals and applies a shorter, behaviour-based window.
Notably, the model rewards ad views that reach shoppers during early discovery — exploratory browsing and general category searches — rather than impressions served to people who were already going to buy. In other words, it's trying to measure genuine incrementality, not coincidence.
Your sales didn't drop. Your ruler got more honest. The sellers who panic are the ones measuring success against a number Amazon deliberately made more accurate.
The impact isn't uniform. Upper-funnel, awareness-style placements — the ones that leaned hardest on generous view-through credit — show the biggest reported drops. Here's how we're re-reading each:
The goal isn't to fight the change — it's more accurate, and that's good. The goal is to stop making decisions off a broken comparison. Here's what we do for the accounts we manage:
QWhy did my Amazon ad conversions drop in 2026?
On 1 January 2026 Amazon replaced its blanket 14-day view-through attribution with a machine-learning model that only credits ad views that genuinely influenced a purchase. That cut reported view-attributed conversions by an estimated 15–30% — while actual sales stayed the same.
QDoes this mean my ads are performing worse?
Usually not. If reported conversions fell but total orders held steady, that's the measurement change, not a performance drop. Rebaseline your targets to the new model and judge with total-business metrics like TaCoS and new-to-brand orders.
QWhich campaign types are most affected?
Amazon DSP and Sponsored Display, which relied most on view-through credit, show the biggest reported drops. Sponsored Products, being click-based, is least affected.
Sources & further reading: Amazon Ads (view-attribution updates for Store ads), Code3 (Amazon quietly tightened attribution), and Skai (2026 State of Amazon Ads). Percentage impacts are third-party estimates; validate against your own account data.