Measurement
Part of Programmatic advertising measurement without vanity metrics
Which programmatic advertising key metrics should an agency report?
The programmatic advertising key metrics agencies should report, from viewable impressions to cost per acquisition, with a decision table for common questions.
Impression : An ad serving recorded for delivery, not evidence that someone noticed it.
Viewable impression : An impression that meets a stated visibility test: 50% of pixels for one second for display, or two seconds for video.
Reach : Unique users or devices counted using the selected method.
Frequency : Impressions divided by reach.
Click-through rate (CTR) : Clicks in relation to impressions, used as a response indicator.
Completion rate : A video delivery measure, not a sales outcome.
Conversion : The recorded action chosen as a campaign outcome.
Cost per acquisition (CPA) : Stated spend divided by recorded conversions.
Read delivery and attention separately
An impression records that an ad was served, while a viewable impression indicates that it passed a specified visibility threshold. Neither measure proves a person noticed or acted on the ad. Report the two separately, and name the test used so readers can tell what the visibility figure means.
Market spending provides context, not a success test: IAB Europe's AdEx Benchmark 2023, published in 2024, put UK programmatic display spending at about £4.5bn. That figure is not a campaign target or proof of performance. For video, completion rate describes delivery; it does not establish a sale or another business result.
Frequency example
- 2,000,000 | Impressions delivered
- 250,000 | Unique users reached
- 8 | Impressions per user
Invalid activity can affect the totals shown. General invalid traffic is removed before billing, while sophisticated invalid traffic is addressed after review. Identify the filter behind the reported number instead of presenting an unqualified total. When vendors disagree, delivery teams should name the source and visibility threshold.
Pair audience size with repeated exposure
Reach and frequency are most useful when read together: reach estimates the unique users or devices counted, and frequency describes average exposure across that group. Explain how the audience total was calculated, because the pair cannot be interpreted properly without the method behind the count.
For example, two million impressions across 250,000 users produce an average frequency of eight. This may indicate repeated delivery to a relatively limited counted group and can prompt a review of frequency caps. It does not, on its own, demonstrate a creative problem or show that the audience definition should change.
Identity signals and counting approaches can produce different audience estimates. Explain the chosen basis alongside audience research, and keep compliance notes separate from the reach definition. Broader media audience research is context, not a replacement for explaining how a campaign's audience figure was counted.
| Reporting question | Measures to lead with | Do not use alone as the headline |
|---|---|---|
| What did the budget deliver? | Viewable impressions, reach and frequency | Raw impressions |
| Did activity work commercially? | Conversions, CPA and revenue | Click-through rate |
| Is there no direct sales route? | Brand lift and assisted conversions | Last-click return on ad spend |
Keep response signals distinct from outcomes
CTR relates clicks to impressions and can help investigate differences by creative or placement. Treat it as a response clue, not proof of business impact. Name the action definition and recording source beside the figure, so a click is not mistaken for evidence that the selected business outcome occurred.
A conversion is the chosen business action recorded for reporting. Assisted conversions show where programmatic activity featured in a journey that ended elsewhere; brand lift studies compare awareness change with a control group. These measures may support assessment when sales are not immediate, but they remain different from direct conversions. See measurement and reporting when agreeing what belongs in the headline set.
Make acquisition cost reproducible
CPA only makes sense when its inputs are clear. Divide the stated spend basis, whether media-only or media plus fees, by recorded conversions. Name the conversion definition and reporting period. Those details make the calculation repeatable and prevent one CPA label from disguising different inputs.
Assess return on investment using its stated inputs, and use attribution methods to explain how credit is assigned. Include the reporting window and conversion source too. Without these details, readers cannot determine whether two apparently similar cost figures use the same basis.
Cost per acquisition checklist
- Write down the conversion definition
- State the attribution window
- Divide total media and fee spend
- Label fully loaded or media-only
- Do not count last-click sales alone
Where a brief has no immediate sales route, assisted conversions and brand lift can contribute supporting evidence. Keep them apart from direct conversions and choose measures to fit the objective. A click, a completed video and a sale represent different kinds of evidence; they are not interchangeable outcomes.
Choose the lead measures from the decision
Let the client's decision determine which terms lead the report. State the reporting period, source, owner and refresh date, then distinguish lead measures from supporting ones. A concise headline set is easier to interpret when each measure has a definition and a clear place in the wider reporting view.
Use the table as a selection aid, not a fixed scorecard. Weekly delivery checks can support pacing, while monthly outcome reporting gives clients time to review results. Date-stamp each view, assign an update owner, and settle the intended measures before building a reporting dashboard.
