Measurement
Part of Why connected TV advertising measurement needs a plan before launch
Before you report connected TV advertising key metrics to a board
Choose connected TV advertising key metrics by decision, not by dashboard: delivery, attention and outcome definitions, plus a worked pound example.
What to take away
- The usual mistake is to open the platform dashboard and report whatever it shows. Choose the decision first, then pick connected TV advertising key metrics that inform it.
- Delivery, attention and outcome measure different things, so one blended number hides which is failing.
- Write down definitions before launchcompleted view, device universe, attribution window. Most arguments later are definition arguments.
- Review delivery weekly against plan and outcomes on a longer cycle, so a pacing wobble is not read as a performance problem.
Set definitions before you set targets
Connected TV buying runs across several platforms, and none of them counts the same way. One may register a view after two seconds, another after a full screen of playback. Fix your own definitions before you compare anything.
Platform view definitions differ
Platform A
- View trigger
- 2 seconds
- Impressions
- Direct
- Reach
- Panel estimate
- Co-viewing
- Counted once
Platform B
- View trigger
- Full screen
- Impressions
- Direct
- Reach
- Modelled panel
- Co-viewing
- Counted once
The connected TV advertising measurement and reporting guide maps how the pieces fit together, which is worth reading before you inherit somebody else's dashboard.
Impressions, reach and frequency
Impressions show how much delivery you bought. Reach estimates how many distinct viewers saw it. Frequency divides impressions by reach. Platforms report impressions directly, then estimate reach from panel data or modelled panels. The Nielsen explanation of panel data explains why those estimates rest on a sample rather than a census.
Completion rate and co-viewing
Completion rate is the share of started plays that reach your definition of a view. A fall usually points to creative or placement, not the media plan.
Co-viewing matters because one living-room screen can hold several people, yet most platforms count them once. If a spot's compliance is uncertain, the ASA Copy Advice service offers pre-publication guidance on the advertising codes.
Link delivery to commercial numbers
Delivery figures only earn their place once they connect to a budget decision.
Cost per completed view and cost per reach point
Cost per completed view divides spend by completed views. It rewards creative that holds attention and penalises cheap inventory nobody watches. Cost per reach point pulls the other way, towards broad inventory. Use both, and be explicit about which one decides where the next pound goes.
Incremental reach and attribution windows
Incremental reach counts viewers your other channels missed, which is the number that justifies CTV inside a mixed plan. Attribution windows decide how much credit CTV receives for action taken later. A seven-day window and a 30-day window can tell very different stories about one campaign, so agree the window in advance.
The guide to connected TV advertising attribution methods sets out how matching choices and window length change reported performance.
Build a review rhythm you can defend
A worked example with labelled pound figures
Take an illustrative six-week campaign with £60,000 of media spend. It delivers 3,000,000 impressions, a £20 cost per thousand. If 85% of plays complete, that leaves 2,550,000 completed views, so about 2.4p per completed view.
The same numbers can look strong or weak depending on the benchmark. Without an incremental reach figure, the buy looks like an expensive way to buy impressions. Add 600,000 incremental viewers and the cost is £100 per thousand, which you can set against other channels.
What to review weekly, monthly and quarterly
Weekly reporting should cover pacing, impressions and completion rate. Anything dragging on those numbers gets a conversation, not another column. Move reach, frequency and incremental reach to a monthly pack, where audience estimates have time to settle. Keep the quarterly conversation for budget share and for retiring metrics nobody uses.
Thinkbox's step on measuring results and reviewing TV campaigns is a practical prompt for that quarterly review.
Common questions
How many metrics should a CTV report hold?
Three to six per audience. A board pack needs cost and outcome, while a media team needs delivery and pacing. Extra columns dilute attention.
Should I trust platform-reported reach?
Treat it as an estimate. Ask whether the figure comes from a panel or a model, and check which devices sit in the universe.
What counts as a good completion rate?
It varies by format and placement, so build your own baseline from the first campaign and watch the trend. A sudden fall tells you more than a static number.
Can CTV metrics sit beside linear TV metrics?
Only after harmonising definitions. Impressions, reach and completion are counted differently across the two, so align them before they share a chart.
