An ad impression is a single instance of an ad being rendered to a user — the atomic unit underneath CPM, eCPM, and most digital ad pricing. But "impression" is a surprisingly fuzzy word. The same campaign measured by three different counting standards can produce three substantially different impression counts and three different effective prices.
Three impression types that matter
- Served impressions — the ad request was sent to the device. The ad may or may not have rendered, may or may not have been visible. Highest count, lowest standard.
- Viewable impressions — by the Media Rating Council (MRC) standard, ≥50% of the ad's pixels were on screen for ≥1 second (≥2 seconds for video). This is what "an impression" generally means in modern brand-side conversations.
- Verified impressions — third-party verified (IAS, Moat, DoubleVerify, Pixalate). Premium advertisers buy only verified inventory because they don't trust publisher-reported impression counts.
Why this matters for pricing: a CPM that looks low may be priced against served impressions but include 30% non-viewable inventory. A CPM that looks high may be priced against verified-viewable impressions only. Always know which standard your inventory is being priced and reported against — comparing them at face value misleads you about actual unit economics.
For publishers: eCPM is normally reported against served impressions in mediation dashboards (AppLovin MAX, IronSource, Admob), but the advertiser-facing pricing may be against viewable impressions. The gap is sometimes substantial.
Impression frequency caps: a separate but related concept. Most ad platforms let you cap how many impressions of the same campaign one user can see in a defined window (e.g., 3 impressions per user per 24 hours). Frequency caps prevent ad fatigue but also reduce reach. Tune them based on creative count and audience size — high creative count + small audience = aggressive cap; low creative count + large audience = looser cap.