A served impression is not the same claim as a viewable one

A served impression is not the same claim as a viewable one

An impression total is the easiest large number for a case study to print and the hardest to read. You want to know how much of it had a chance to land in front of a person. Usually it counts deliveries.

What a platform counts as an impression

The base impression metric records a delivery event: requested, returned, rendered. That is a fact about the ad server, not about anyone’s eyes. The Google Ad Manager documentation says as much, noting that people may never see ads that were counted as impressions.

Delivery without a chance to be seen is ordinary. The ad sits three screens down and the reader stops after one. The page waits in a background tab. The creative renders after the reader swiped past.

Read the definition page of the system that produced your number before the figure itself, which is also how you catch one metric name covering three different calculations.

Viewability is a threshold, and thresholds have parameters

Viewable is not a common sense word here but a measured condition with two parameters: how much of the ad is on screen, and for how long. Ad Manager gives criteria it attributes to the Interactive Advertising Bureau: half a display ad’s pixels for one continuous second, 30 percent for ads above 242,000 pixels, and half an in stream video ad’s pixels for two seconds. The Media Rating Council’s standards index lists Viewable Ad Impression Measurement Guidelines and the Bureau’s index carries the mobile version, so this is published guidance, not platform preference.

Display and video are different tests, so one rate across a mixed campaign averages two. The report also has three buckets, not two: measured and viewable, measured and not viewable, and not measurable at all. The rate is calculated against measurable impressions, so an unmeasured impression got no verdict, and folding it into either bucket is a choice.

A worked example, invented on purpose

Every figure below is hypothetical, invented to show the mechanism. Picture 20,000 dollars of hypothetical spend, four million served impressions, 85 percent measurable.

Hypothetical figures 62 percent viewable (hypothetical) 38 percent viewable (hypothetical)
Served 4,000,000 4,000,000
Measurable 3,400,000 3,400,000
Viewable 2,108,000 1,292,000
Cost per 1,000 served 5.00 dollars 5.00 dollars
Cost per 1,000 viewable 9.49 dollars 15.48 dollars

The viewable count differs by more than 800,000 between the two hypothetical rates, and 600,000 impressions sit outside the calculation entirely. Cost is where it bites: five dollars per thousand served, nine or fifteen per thousand viewable. Only the higher figure prices an opportunity to be seen, and Ad Manager reports that basis as average viewable cost per thousand impressions. Neither rate is a benchmark.

What the case study would have to disclose

  • Served or viewable, in the same sentence as the number, not a footnote.
  • Which standard or product produced the viewable figure: an in house measure and an accredited third party are different claims about who measured the number.
  • The measurable share, separately: an unmeasured remainder is neither viewable nor non viewable.
  • The placement and format mix, since in feed, in stream and display placements behave differently on screen.
  • Whether any cost figure beside the total uses served or viewable impressions as its denominator.

What this is not: the neighbouring mistakes

This is not fraud or non human traffic. The mechanism is real people on real pages who never had the ad in front of them, so a clean invalid traffic report settles nothing. Nor is it the reach question: reach counts distinct people, viewability asks whether a delivery had a chance to land, and a claim can be wrong about both independently.

A low rate is also not evidence the agency did anything wrong. Inventory, placement and format drive much of it, so the useful reaction is to ask for the disclosure, not to accuse. It helps to know why the biggest available number ends up in the headline.

Reading the figure in front of you

Most case studies disclose none of this, so you have to ask. Two questions, short enough to paste into an email: is the impression total served or viewability qualified, and what measured it?

A good answer is specific and slightly boring. It names the reporting system, the threshold, and the measurable share, sometimes admitting measurement was never running. An evasive answer restates the total, or quotes a viewability percentage with no system named.

Labelled correctly, it is still usable. Treat a served total as an upper bound on the opportunity to be seen. It is a ceiling, not an audience.

Check the claim against the library

Browse the case study library to see how much disclosure a published write up carries, or submit your own. For the rest of the checks, there is the general reading method.

FAQ

Does an impression mean somebody saw my ad?

No. The base metric records that the ad was delivered, and the Ad Manager documentation says ads can be counted as impressions when nobody saw them. A served total is a ceiling on opportunity to be seen, not a count of people.

What makes an impression viewable?

Enough of the ad has to be on screen for long enough, under a published rule. Ad Manager’s criteria, attributed to the Interactive Advertising Bureau, are half a display ad’s pixels for one continuous second and half an in stream video ad’s pixels for two.

Is a viewable impression the same as a view?

No. Viewable means the delivery met a measurement rule about pixels on screen and time on screen. It says nothing about attention, recall, or whether the person looked at the ad. It records an opportunity, and stops there.

If a case study only reports served impressions, is it dishonest?

Not necessarily. Plenty of reporting setups never had viewability measurement running, so the figure may not exist. The problem is delivery language dressed as audience. The honest version reads: the campaign served four million impressions, viewability was not measured.

Why would some impressions be unmeasurable?

Measurement depends on the environment cooperating with the measuring tag. Ad Manager lists creatives rendering in a cross domain iframe, truncated ad request URLs, a browser closed before the tag finished reporting, and out of page creatives, which are not measured at all.

Sources

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *