Display advertising quality: what to check beyond cheap CPMs
Quick answer
A low CPM means you bought impressions cheaply. It doesn’t tell you whether the right people could see the ad, whether they noticed it or whether it changed anything. Check the audience, placements and frequency before celebrating the price.
Cheap inventory can be an expensive distraction
A campaign can beat its CPM target while reaching people who will never buy, appearing in unsuitable contexts or serving the same audience too often. The headline price becomes useful only after the inventory meets the purpose of the campaign.
The IAB/PwC report for 2025 puts US display advertising revenue at $81.6 billion. Market scale says little about the quality of a particular media plan. It does explain why buyers need a repeatable way to distinguish useful reach from readily available volume.
Write the campaign’s job before buying: introduce a brand, support a named-account programme, remind a relevant audience or promote a specific offer. Each requires a different mix of creative, reach, frequency and measurement.
Understand what viewability establishes
Google’s Active View documentation describes a viewable display impression as at least 50% of the ad’s area visible for at least one second. This is an opportunity-to-see threshold. It is not evidence that a person noticed, understood or remembered the message.
Keep served impressions, measurable impressions and viewable impressions distinct in reports. A viewability rate based on measurable impressions does not necessarily describe all delivered inventory. Ask the platform or supplier which denominator it uses.
Creative still has to do its work inside that opportunity. A tiny logo and an unreadable paragraph can technically appear on screen without communicating anything. Review the actual rendered sizes and mobile placements, not only a large design-board preview.
| Dimension | Inspect | Interpretation |
|---|---|---|
| Placement | Sites, apps and unexpected concentrations | Context and inventory suitability |
| Viewability | What counts as viewable and how the rate is calculated | Opportunity to see, not proven attention |
| Frequency | Distribution across the audience | Average can conceal overexposure |
| Outcome | Qualified response and feasible lift tests | Attribution is not causation |
Inspect placement and audience evidence
Review site and app reports, excluded categories, geography, device mix and unexpected traffic concentrations. Investigate sudden cheap volume rather than automatically rewarding it. Ask how invalid traffic is filtered and which measurements are independently supported.
For an account-based campaign, check how the platform maps users to organisations and how coverage varies. A targeting label is not proof that every impression reached the intended buying group.
Keep frequency visible across the time window that matters. An average can hide a heavily exposed minority. Where the platform allows it, inspect the distribution and exclusions. Avoid declaring a universal ideal frequency without accounting for message, format, audience and campaign duration.
Compare cost after quality filters
Say campaign A buys a million impressions for £2,000, while campaign B buys 600,000 for £1,800. A looks cheaper: a £2 CPM against B’s £3. Now apply the quality checks you agreed before buying. If only 300,000 of A’s impressions qualify, against 420,000 of B’s, the cost per thousand qualifying impressions is about £6.67 for A and £4.29 for B. The apparently cheaper campaign is costing you more for the inventory you want.
Use your own figures for that comparison, and keep the quality criteria consistent. Don’t multiply together unrelated scores just to produce a reassuring number. You should be able to explain what passed the checks and why.
Use a report that keeps cost, reach, viewability and outcomes visible. Our marketing chart guide helps prevent a favourable headline metric from concealing an unfavourable denominator.
Check delivery before blaming the creative
First ask whether the ad reached a plausible buyer in a place where they could see it. Then ask what the message changed. You won’t fix poor targeting with another headline, and better placements won’t rescue a message nobody understands.
Clicks are useful for some objectives, but display may also influence later behaviour. View-through attribution can report that sequence without proving the impression caused the conversion. Use incrementality testing where feasible, and keep observational contribution claims qualified.
For a small programme, the next review should be concrete: inspect the largest placements by spend, the cheapest unexpected placements, the frequency distribution and the landing experience. Make exclusions or creative changes based on that evidence. The aim is a defensible media decision, not a dashboard where every number is green.
Related reading
Marketing incrementality testing: find out what the campaign changed · Marketing attribution vs MMM vs incrementality: which question are you asking? · Marketing budget allocation: fund the next useful decision
Photo: Pawel Czerwinski / Unsplash.
