An ad impression is a single instance of an ad being displayed to a user. Under IAB and MRC standards, a display impression counts as viewable when at least 50% of the ad pixels are in view for at least 1 continuous second (2 seconds for video). Impressions are the base unit that CPM, CTR, reach, and viewability metrics are all derived from.
Every other paid media metric — CTR, CPC, CPM, reach, frequency — is built on top of the impression. Get the impression definition wrong (served vs. viewable) and every downstream KPI reports the wrong story.
What is an ad impression?
An ad impression is a single instance of an ad being displayed to a user. Every time an ad is served to a page or app slot, one impression is counted — regardless of whether the user actually saw it, clicked it, or engaged with it.
This is why the industry distinguishes between served impressions (any time an ad was placed on the page) and viewable impressions (met the viewability standard). Ad buyers evaluate performance using viewable impressions; ad networks historically reported served. The gap matters because CPM pricing is charged against served impressions, not the ones anyone actually saw.
The Media Rating Council (MRC), backed by the IAB, defines a viewable display impression as one where at least 50% of the ad pixels are in view for at least 1 continuous second. For video: 50% in view for at least 2 continuous seconds. Programmatic buyers filter for MRC-compliant viewable impressions on nearly every campaign.
Why ad impressions matter
Impressions drive every other paid metric. Four reasons every media buyer cares about counting them correctly:
- Pricing foundation. Almost all display and social inventory is bought on a CPM basis — cost per 1,000 impressions. Undercount impressions and you undercount cost efficiency.
- Reach + frequency math. Frequency = impressions / reach. You can't measure fatigue without accurate impression data.
- Viewability filtering. A cheap $0.50 CPM at 15% viewability costs more per real view than a $3 CPM at 80% viewability. Impressions without viewability context are misleading.
- Attribution baseline. Multi-touch and view-through attribution both start with the impression — was the user actually exposed?
How ad impressions are counted
The ad-serving handshake
When a page loads, the publisher's ad server calls out to available buyers. The winning bid renders the ad. At render, a tracking pixel fires and one impression is logged in both the publisher's and advertiser's systems.
Viewability measurement
Third-party verification tools (IAS, DoubleVerify, Moat) monitor whether the ad met viewability criteria — how many pixels are in view, for how long. Only impressions that pass become "viewable impressions" in the buyer's report.
Deduplication
An impression served twice to the same user in a short window (e.g. page refresh or SPA re-render) may be deduped. Platforms have different rules — Meta counts server-side; Google counts by pixel fire; DSPs use bid IDs.
Impression types compared
| Type | Definition | Use case |
|---|---|---|
| Viewable impression | Meets MRC 50%/1-sec rule | The gold standard for evaluating campaigns |
| Served impression | Ad was rendered, viewability unknown | Legacy reporting; ad-server billing baseline |
| Rendered impression | Ad markup loaded but may not have painted | Programmatic technical measurement |
| 100% viewable | Full ad in view for the full duration | Premium buys, brand campaigns |
| Completed view (video) | Full video watched to end | YouTube TrueView, CTV |
Real ad impression examples
Example 1 — Display campaign math
A brand spends $10,000 on a display campaign that generates 2 million served impressions at a $5 CPM. Viewability comes in at 62% — so 1.24 million viewable impressions. Real viewable CPM is $8.06. That's the number to use for benchmarking, not the reported $5.
Example 2 — Video campaign
A YouTube pre-roll campaign reports 500,000 impressions. Under video viewability rules (50% of player, 2 continuous seconds), 78% qualify — 390,000 viewable. Completed views (full 30-second view): 145,000. Each metric tells a different story about the same media buy.
Example 3 — Reach vs. impressions
A Meta campaign shows 300,000 impressions to 60,000 unique users. Frequency is 5.0. If the frequency cap were set at 3, the same $ would have reached ~100,000 unique users — a wider reach for the same impression count. See how frequency is calculated for the full reach-and-fatigue picture.
Impressions vs. clicks — what to optimize
Impressions and clicks answer different questions. Match the metric to the objective: impressions measure exposure, while click-through rate measures what that exposure earned you.
Optimize impressions when
- Brand awareness is the goal
- You're building recall and share of mind
- Reach + frequency drive the outcome
- CPM is the primary cost lever
- View-through attribution matters
Optimize clicks when
- Direct response is the goal
- Cost per action drives the outcome
- CTR is a leading indicator of quality
- You're diagnosing creative performance
- Landing page traffic is the KPI
6 best practices for measuring ad impressions
- Report viewable, not served. Make viewable impressions the default number your team looks at.
- Set an 80% viewability floor. On programmatic buys, block inventory below MRC-viewability of 70-80%. It filters out fraud and made-for-advertising sites.
- Track impression share. Google Ads reports the % of eligible impressions you actually got. Below 60%? You're leaving budget or bid on the table.
- Deduplicate across platforms. A user on Meta + Google + TikTok generates 3 impression counts for what's really one exposure. Use a unified reach tool.
- Split impressions by placement. Feed impressions on Meta ≠ Stories ≠ Reels. Segment before averaging.
- Verify with a third party. DoubleVerify, IAS, or Moat give the impression counts advertisers can defend in an audit. Platform-reported numbers alone aren't enough for large media buys.
Reporting "we served 5M impressions" as if it were reach massively inflates campaign performance. 5M impressions at frequency 8 is only ~625,000 people reached. Always report reach and frequency alongside impression volume.
Common ad impression mistakes to avoid
- Confusing impressions with reach. Impressions count every display; reach counts unique people.
- Ignoring viewability. An impression served below the fold on a page the user never scrolled to is worth $0.
- Buying cheap CPM without QC. $0.50 CPM inventory is usually cheap for a reason — fraud, ad stacking, or made-for-advertising sites.
- No cross-platform dedupe. Reporting cumulative impressions across Meta + Google + TikTok as unique reach is a big overcount.
- Optimizing to impression volume, not outcomes. More impressions ≠ more revenue. Always tie to a downstream metric.
Frequently asked questions
An ad impression is counted each time an ad is served to a page or app slot. Under IAB and MRC viewability standards, a display impression counts as viewable when at least 50% of the ad pixels are in view for at least 1 continuous second — 2 seconds for video.
Impressions count every time an ad is displayed. Reach counts the unique people who saw it. If one person sees the ad 5 times, that's 5 impressions but a reach of 1.
Impressions are priced on a CPM (cost per mille / cost per thousand) basis. Meta display averages $7-15 CPM in 2026. LinkedIn ranges $30-70. Programmatic display can run under $2 CPM on open exchanges. Costs vary by targeting, audience quality, and inventory.
Served impressions include ads that never became viewable — for example, an ad below the fold on a page the user never scrolled to. That's why buyers filter for viewable impressions using MRC-compliant measurement.
For video, MRC requires at least 50% of the video player in view for at least 2 continuous seconds. Platforms also report metrics like true-view (30-second view or complete video) and completed views for stronger engagement signals.
Related glossary terms
Sources
