Observations, not impressions: how Adsup.ai counts ads

Most ad intelligence tools show a big number next to each ad: impressions, reach, estimated spend. Very few explain where that number comes from. Adsup.ai takes a narrower and more honest position. Every count you see is an observation: one of our browsers saw that ad, once, on one page.

What an observation is

Our workers load public pages in real Chrome, with desktop and mobile profiles, from the country being measured. When an ad renders, the slot detector takes an element screenshot and records the request chain behind it: the ad call, the click URL, each redirect and the landing page. That record is one observation.

Repeat observations of the same creative are folded into a single ad, which carries:

• First and last seen dates, so you can tell a test from a long-runner.
• Days seen, the number of distinct days we observed it.
• Publishers, the sites it ran on.
• Networks, countries and devices, from the evidence in each observation.

What it is not

An observation is not an impression. A campaign that serves a million impressions a day might appear in a handful of our observations, and a small campaign on a site we crawl often might appear in many. Observations tell you that an ad ran, where and when. They are a sample, and we never present them as spend or reach.

Check our coverage yourself

Because the numbers are samples, it matters what we sampled. The coverage ledger lists, for every network, format and country, how many observations we recorded in the last 24 hours and 7 days, and when we last saw each one. It is public and needs no account.

If a network or country you care about is thin in the ledger, you know before you rely on it. The methodology page documents the rest: how sites are chosen, how consent banners are handled and the known limits of the data.