How to measure streaming TV ads
Measure streaming TV ads by what they do, not by clicks, because almost nobody clicks a TV ad. Use three reads together: did people watch (completion on TV screens), did they remember (a brand lift study against a control group), and did sales move (a model that ties exposure to sales over the following weeks).
Key takeaways
- Click-based reports undercount streaming TV. People watch on the big screen and buy later, on another device or in a store.
- Start with delivery: the share of TV plays that ran to the end, by cut, compared with your own campaign's average.
- A brand lift study compares people who saw the ad with a similar group who didn't.
- A media mix or effectiveness model ties TV exposure to sales, including the lag between seeing an ad and buying.
- Compare like with like: the same metric, on the same screens, over the same period.
Why streaming TV is hard to measure
Streaming TV ads run inside apps on internet-connected TVs. You'll hear it called connected TV (CTV) or OTT. It can be targeted like digital, but it's watched like television: on the couch, on the big screen, often by more than one person.
That breaks most digital reporting. There's nothing to click. The person who saw the ad may buy days or weeks later, on a phone or a laptop, or in a store. Last-click attribution gives that sale to whatever they clicked last, usually a search ad, and the TV ad gets nothing. Partners' own reports fill the gap with their own numbers, which you can't check against anyone else's.
So measure it in three layers. Each one answers a different question.
1. Did people watch?
Start with delivery. Completion rate is the share of impressions that played to the end. Look at it by cut and by screen, because a 10-second spot and a 30-second spot behave differently, and so do phones and TVs.
Compare your cuts with each other rather than with an industry average from a different mix of screens. For a luxury resort launch, our best cut, a 10-second spot, played to the last frame on 89% of its TV plays. The average across all 12 cuts on TV screens was 40.7%. That's the campaign's own average, not an industry benchmark. Across every device, the same two numbers were 80.4% and 27.9%, which shows how much the screen mix changes the answer. See the numbers.
2. Did people remember?
A brand lift study surveys two groups: people who saw the ad and a similar control group who didn't. The difference in awareness or consideration is the lift. Because the control group shows what would have happened anyway, the lift is incremental by design. Read more in incrementality vs. attribution.
In the same resort launch, a YouTube Brand Lift study found that on TV screens, 9.01% of people who saw the spot were aware, against 7.54% of the control group. That's about 20% higher. Not every age group or device showed lift, and a good report says so.
3. Did sales move?
Awareness isn't revenue. To connect TV to sales, use a media mix model or a similar effectiveness model. It lines up ad impressions or spend by channel over time against the outcome you care about, controls for things like seasons and holidays, and learns the lag between seeing an ad and buying.
For the resort, our model lined up daily ad impressions by channel against room nights. It credits about $1.54M in room revenue to the program's ads across all of its channels (Social, Google, CTV, DOOH and OTA), showing up in stays 2 to 6 weeks after people saw them. It's a model estimate, not a holdout test, and the range around it is wide.
A media mix model can also read video as a group, YouTube and streaming TV together, which is often steadier than reading each partner on its own when some of them only run a few weeks a year.
Mistakes to avoid
- Judging TV on clicks. A low click-through rate on a TV screen tells you almost nothing.
- Taking a partner's report as the answer. Use it, but check it against a lift study or a model that sees all your sales.
- Mixing screens. A completion rate across phones, desktops and TVs isn't comparable with one from TV screens alone.
- Expecting next-day sales. Give the model a window that matches how long your customers take to buy.
How we measure streaming TV
We plan and buy streaming TV alongside search, social, YouTube and offline media, and we measure it in the same three layers: completion by cut and by screen, brand lift against a control group, and a model that ties exposure to sales. Every comparison stays like for like. See the resort launch on our Results page.
Common questions
What's a good completion rate for a streaming TV ad?
It depends on the ad's length, where it runs and whether it can be skipped, so a single industry number isn't very useful. Compare your cuts with each other and with your own campaign average, on the same screens.
How long after a streaming TV ad do sales show up?
It depends on what you sell. For the luxury resort launch on our Results page, our model found the ads showing up in hotel stays 2 to 6 weeks after people saw them, strongest around week 4. Measure over a window long enough to catch that lag.
Can you measure streaming TV for a business that sells in stores?
Yes. A media mix model measures advertising against all sales, in store and online, so it doesn't need a click or a tracked visit. See measuring marketing when most sales happen in stores.
