MeasurementMay 3, 20268 min read

What “verified impressions”actually means

Two campaigns can report the same impression number and mean entirely different things by it. The difference is whether anyone independent checked.

AAdvertising WheelsMeasurement team

Where the big numbers come from

A modeled impression estimate multiplies a traffic count by an assumed visibility factor by a duration. Every one of those inputs is an assumption, and small changes to any of them move the headline number enormously.

That is not inherently dishonest — modelling is how most outdoor has always been sold. It is simply not a measurement, and it should not be compared against one.

What verification actually adds

A verified impression starts from a logged fact: this vehicle was on this road segment at this time, at this speed, for this duration. Audience is then applied to a route that is known to have happened rather than one that was planned.

The count is usually lower than a modeled equivalent. That is the point — it is defensible, and it can be reconciled line by line.

A smaller number you can defend is worth more than a larger one you cannot.

Four questions that separate the two

Ask whether impressions are logged or modeled. Ask at what interval position is recorded. Ask who audits the data and whether you can see the raw route file. Ask how shortfall against the planned route is handled.

Vendors working from verified data answer all four immediately. The answers themselves matter less than the speed with which they arrive.

The short version

Four things to take away.

  1. 01

    Modeled impressions are assumptions stacked on a traffic count.

  2. 02

    Verified impressions begin from logged position, speed and duration.

  3. 03

    Expect verified counts to be lower — and defensible.

  4. 04

    Ask: logged or modeled, at what interval, audited by whom, shortfall handled how.