Impressions in DOOH: how audience measurement works

If a screen has no cookies, no logins, and nothing to click, where does its impression count come from? DOOH impressions are modeled estimates of the people who had a realistic opportunity to see an ad while it played. They're built from venue traffic and screen placement rather than device signals, and the quality of the count depends on the quality of the traffic data underneath it.

That's the short answer. The longer answer is worth knowing, because buyers who understand the model ask better questions than buyers who just compare totals.

Where do the numbers come from?

Digital out-of-home measurement starts with a simple question: how many people move through this place while the screen is playing? For roadside media, that means traffic counts. For retail, it means store visits. The count is then adjusted for placement, since a screen at the register has a different audience than one behind a stockroom door.

Standards bodies exist to keep this from becoming guesswork. Geopath, the industry organization for out-of-home audience measurement in the US, maintains methods built around the opportunity to see, and that concept is the backbone of most DOOH impression reporting.

A retail media network anchored to point-of-sale systems has an unusually direct traffic source. The NRS network records 298 million weekly visits and 1.9 billion transactions annually across 34,000+ independently-owned stores, so the visit data behind the impression model comes from registers, not from a panel extrapolation.

Why can one play count as more or less than one impression?

Because a play and an impression are different things. A play is the screen showing your creative once. An impression is one person's opportunity to see it. When an ad plays during a busy morning rush, several shoppers may be in view, so one play can produce several impressions. When it plays to a quiet aisle at 2 p.m., a play may produce none.

That's why impression counts move with dayparts and store traffic even when the play count stays flat, and why comparing two networks on plays alone tells you very little.

How do reach and frequency fit in?

Impressions are the raw material; reach and frequency are what planners actually buy. Reach is the count of distinct people exposed at least once. Frequency is how many times the average person was exposed.

Neighborhood retail has a distinctive shape here. Corner stores serve regulars, and a shopper who stops in most days walks past the same screen again and again. A flight in this channel tends to build frequency against a loyal local audience rather than skimming one exposure off a huge transient crowd. Neither pattern is better in the abstract. They suit different jobs, and a media plan should know which one it's buying.

What should a buyer ask a network?

Four questions separate solid impression reporting from soft numbers. What is the traffic source, and is it measured or assumed? How often is it refreshed? Does the model account for screen placement within the venue? And can results be reported at the level you plan to buy, whether that's a retail channel, a region, or a zip code?

On the NRS network, campaigns activate by audience index, retail channel, SKU, or geography down to zip-code level, and the measured screen count, 39,000+, is a stated part of the network rather than a moving claim. Because the same company operates the point-of-sale platform and the scan-data analytics arm, traffic and outcome data live in one system instead of being stitched together from vendors.