The Maze: Amazon plans to change when it records a display-ad impression on its shopping pages from January 1, 2027. Instead of counting an ad when it starts downloading to a shopper’s device, it will wait until the ad appears on screen. According to an advertiser notice reproduced by PPC Land, Amazon expects about 63% fewer recorded impressions on the affected placements, on average. That is a change in measurement. It does not establish that shoppers will receive 63% fewer ads or that advertisers will spend 63% less.
Source: Amazon Ads. Amazon’s DVA+ campaign illustration shows objective selection. It depicts the existing buying workflow, not the January impression-counting change.
The Details:
The rule follows the ad onto the screen. The notice covers display placements bought through DVA, Amazon’s Display, Video & Audio advertising product, on the Amazon Store homepage, product pages, search results and order pages. An ad can start downloading before a shopper scrolls down to its position. The old trigger can therefore record an impression even if the shopper leaves before reaching it. Amazon says placements lower on a page will see larger decreases under the new rule. The 63% estimate is an average, not a prediction for every campaign. The reproduced notice supplies no country list and does not identify sponsored search ads, video, audio or third-party websites as covered. Amazon’s broader DVA+ campaign product can buy across Amazon properties and the open internet; that broader capability does not expand this notice’s scope.
A smaller count can make the same campaign look different. Consider a hypothetical campaign with $500 spending and 100,000 recorded impressions. Its reported cost per thousand impressions, or CPM, is $5. If only the count falls by 63%, it becomes 37,000 impressions and the reported CPM becomes about $13.51. No extra spending is required for that increase. Click-through rate, the share of recorded impressions producing clicks, would also rise if clicks stayed unchanged. Return on ad spend, or ROAS, measures attributed sales per advertising dollar. Cost per action divides spending by attributed actions, such as purchases. Changing the impression count alone improves neither of those measures. Amazon says its tests produced higher ROAS and lower cost per action, but the notice provides no effect size, sample, test period or explanation of the cause. Changes in billing, campaign optimization or attribution could matter; this notice does not isolate them.
On-screen counting is only one measurement boundary. The notice does not define how much of an ad must appear or for how long. The Media Rating Council’s desktop display standard generally requires at least half the ad’s pixels in view for one continuous second to count as viewable. Even that establishes an opportunity to see, rather than proof someone paid attention. Google is making a separate change: its AdSense and Ad Manager banner ads on desktop, mobile web and connected TV will move from download-based counting to “begin-to-render” on February 17, 2027. That means the ad has loaded and started drawing on the device; it is distinct from a viewable impression. The figures cannot be compared as if both platforms were adopting the same rule. For scale, Amazon reported $19.637 billion in advertising-services revenue in Q3 2026, against $17.703 billion a year earlier. That total includes sponsored, display and video ads; it cannot reveal how much revenue these particular Store placements generate.
MarketMaze Take: Amazon is changing the unit brands use to judge its own advertising inventory. A stricter count can make a placement’s reported audience more meaningful. But its commercial value depends on the invoice as well as the dashboard. A thousand downloaded ads and a thousand ads that appear on screen describe different exposure. The notice leaves open whether the billed unit and auction prices change alongside the reported count.
That makes January a break in the comparison, not an automatic verdict on campaign quality. Brands have a reason to preserve placement-level spending and outcome baselines before moving budgets in response to higher CPM or click-through rates. If the same money produces the same sales, a cleaner count improves reporting without improving the economics. Amazon’s claimed test gains may reflect something more, but the missing test and billing details are what would establish it.
