AdSense and Ad Manager change impression counting: begin-to-render starts February 17, 2027
Per Google's official AdSense help page, impression measurement for display inventory moves from counting the moment an ad starts to download to counting the moment it starts to render. Google writes that no action is required from publishers, but that total impressions may decrease.
The measurement method changes on February 17, 2027
In a document published in the AdSense help center, Google announced that impression measurement for display inventory in Google Ad Manager and AdSense will shift from count-on-download to begin-to-render beginning February 17, 2027.
The document gives the scope as well: the change affects display inventory on web, mobile web and CTV banner.
Google states that no action is required from publishers for this transition.
Search Engine Land covered the change in a story dated September 1, 2026 and wrote that publishers should expect a decrease in reported impression counts.
The difference between the two methods and Google's reasoning
Google defines both methods explicitly in the document. Count-on-download counts an impression when an ad starts to download to a user's device. Begin-to-render counts it after the ad creative has successfully loaded and has started to render on the device.
Per Google, begin-to-render marks a point where the user is more likely to be able to view the ad. The company describes it as a more rigorous measurement standard than count-on-download.
The reasoning behind the change is consistency. Google notes that Ad Manager and AdSense already use, or are compliant with, begin-to-render for native, app and video inventory. Moving display to the same basis unifies the impression counting methodologies.
There is a clear warning about the impact too. Google writes that total display impressions may decrease after the shift. The reason is the method itself: begin-to-render counts an impression later in the ad serving process.
The gap comes from a concrete behavior. Count-on-download could also count ads that had begun loading on a page but were abandoned before fully rendering, for example when a user navigates away quickly. Begin-to-render does not count those cases.
Per Search Engine Land, Google recommends that publishers run reports comparing current count-on-download and begin-to-render metrics to estimate their own impact, and review creative optimization practices so assets render efficiently.
The document makes no commitment about revenue. What Google writes concerns the impression count.
When the counter moves, comparison with the past breaks
The assessment in this section is ours. When the definition of a metric changes, comparing it against its own history loses meaning. The decline visible after February 17, 2027 will not be a loss of performance, it will be the counter moving.
Knowing this in advance matters, because a drop in a report is easy to misread. A decline that looks like lost traffic, a broken ad placement or a seasonal swing may in fact come from the change in method.
The second point is more lasting. Begin-to-render ties the technical performance of a page directly to the impression count. If a creative is slow to start rendering, ads that had not begun rendering before the user left are no longer counted.
That moves page speed out of the user experience column and into the measured inventory itself. A slow page does not only lose the user, it also fails to record that user's impression.
Third, Google's reasoning points in a direction. Pulling display onto the same basis after native, app and video inventory shows that the industry's move toward tying an impression to a more viewable moment is continuing.
The timeline is an advantage as well. There are months before the change, and Google recommends running comparative reports. The size of the impact can be estimated, so it does not have to arrive as a surprise.
What it means for businesses in Türkiye
The assessment below is not in the source, it is our reading. The document holds no Türkiye specific information.
The directly affected group is publishers serving ads through AdSense or Ad Manager. In Türkiye that includes news sites, blog publishers and content driven ecommerce sites.
For those publishers the task is clear and the timing works: record today's level before the change takes effect. If the estimated impact is measured with a comparative report, the picture after February reads as an expected deviation rather than a surprise.
There is something to read on the advertiser side too. When the definition of an impression tightens, the reported count may fall while the likelihood that each counted impression actually reached a user rises. The number gets smaller as its meaning gets stronger.
The technical side matters particularly here. In a market where mobile connection quality varies, anything that delays a creative from starting to render now affects measured inventory as well. Heavy pages, late loading scripts and disorganized ad slots become more visible in that picture.
In short, this looks like a reporting footnote but it ties page performance directly to the measurement on the revenue side.
The UNALSOFT angle
On the web design side we have long treated page performance not as a user experience heading but as a component of measurable outcomes. This change makes that link concrete: if a creative is late to start rendering, the impression is no longer counted. Our recommendation to publisher clients has two steps. First, record the current impression level before February 2027 arrives and derive the estimated deviation with a comparative report. Second, review the placement and loading order of ad slots. When the counter moves later in the process, anything that arrives late stops appearing in the report.
Is your page abandoned before the ad renders?
Let us review your page performance and the loading order of your ad slots, and take a baseline measurement before February.