How does a YouTube Brand Lift study measure ad recall?
Part of YouTube channel planning
The short answer
A YouTube Brand Lift study is a Google survey test that compares people who saw a campaign's ads with similar people who were eligible but withheld. Short surveys shown on YouTube ask about ad recall, awareness, consideration or other perceptions, and the gap in positive answers is the lift. Access needs a Google account representative and a minimum budget.
Key takeaways
- Brand Lift compares survey answers from an exposed group and a withheld group; the difference in positive answers is the lift.
- Google lists 10 day minimums per number of questions, with the United States in Country B and Canada in Country C.
- A lift result measures perception, not sales or searches, which Google measures with separate Conversion Lift and Search Lift studies.
- Placements, keywords, retargeting, small geographies and mismatched survey languages all reduce survey responses, according to Google.
What is a YouTube Brand Lift study?
Google's About Brand Lift page (opens in a new tab) calls Brand Lift "a free tool for measuring the effectiveness of your video ads" in Video and Demand Gen campaigns. It measures ad recall, awareness and consideration rather than clicks or views. The same page says Brand Lift "isn't available for all Google Ads accounts" and that an advertiser needs a Google account representative to use it.
Definition
- Brand Lift study
- A controlled survey experiment in Google Ads or Display & Video 360. People who saw the ads and people who were eligible but did not see them answer the same short question, and the difference in positive answers is reported as lift.
A study belongs to a Product or Brand that groups the campaigns being measured. Google's setup page (opens in a new tab) lets the advertiser choose up to three metrics, pick the survey language and enter the brand as the preferred answer beside up to three competitor names. Surveys can take up to 48 hours to be approved and start when the campaigns start.
| Metric | Answer format Google lists | Example wording on Google's setup page |
|---|---|---|
| Ad recall | Multi-select | "Which of these vehicle brands have you seen online video advertising for recently?"; the verb adapts to audio, video or both |
| Awareness | Multi-select | "Which of these services have you heard of?" |
| Association | Multi-select | No example question given |
| Consideration | Multi-select | "Which of these laundry brands would you consider buying from?" |
| Favorability | Multi-select | "Which of these oil brands do you have a positive opinion of?" |
| Purchase intent | Single-select | "Next time you buy a vehicle, which one are you most likely to buy?" |
| Custom metric | Single or multi-select | Written by the advertiser within Google's policies |
How does Brand Lift measure ad recall?
Once campaigns run, Google shows Brand Lift surveys "on YouTube, before a video starts" to people who saw the ads and to people who were eligible but were not shown them. For ad recall, the verb adapts: hearing for audio, viewing for video, encountering for both. Google limits each person to one survey a day, three a week, and one response per product or brand in 14 days. With several metrics, each viewer sees one question, chosen at random.
| Figure | Google's calculation | Result in Google's example |
|---|---|---|
| Absolute lift | Exposed positive rate minus baseline positive rate | 20 percent |
| Relative lift | That difference divided by the baseline positive rate | 100 percent |
| Headroom lift | Absolute lift divided by 1 minus the baseline rate | 25 percent |
| Lifted users | Estimated people whose perception changed, extended to the campaign's reach | Not given; Google says it is not a unique user count |
| Cost per lifted user | Campaign cost divided by lifted users | Not given; Google suggests it as a primary success metric |
Response volume decides what can be detected. Google's metrics and statuses page (opens in a new tab) says high performing campaigns can expect detectable lift at about 2,000 responses per metric, campaigns at the recommended budget minimum at 4,100, and that a campaign with no lift after 16,800 responses "may not be able to detect" it. Detecting a 1 percent absolute lift needs 20,000 to 45,000 responses in Google's table.
Who is eligible to run a Brand Lift study?
Three conditions apply in Google Ads: a Google account representative, a Video, audio or Demand Gen campaign, and enough budget across the campaigns in the study within a 10 day window. Google notes that Demand Gen measurement captures video ads on YouTube, not Gmail or Discover. YouTube Reserve campaigns need the study set up during campaign creation; adding it later can raise an instant reserve error.
| Questions measured | United States (Country B) | Canada (Country C) |
|---|---|---|
| 1 question | $10,000 | $15,000 |
| 2 questions | $20,000 | $30,000 |
| 3 questions | $60,000 | $60,000 |
Google offers two study strengths. The default completes in 3 to 14 days depending on response speed. Enhanced Lift needs three times the default minimum and runs 9 to 14 days to collect about three times the surveys; Google says it does not guarantee positive results. In Display & Video 360 (opens in a new tab), YouTube & partners studies measure through the insertion order's flight, with surveys for 14 days or until enough responses arrive.
What can a Brand Lift result not tell you?
Brand Lift measures perception. Google's lift studies overview (opens in a new tab) separates it from Search Lift, which measures brand searches, and Conversion Lift, which measures sales or sign ups. A strong ad recall result is evidence that people remember the ad, not evidence that the campaign sold more product.
Limits to state when a Brand Lift result is shared
- The result describes survey answers on YouTube, not purchases, store visits or searches.
- Lifted users is an estimate extended to the campaign's reach and may count one person more than once, per Google.
- Google says re-engaging lifted users with the data is not supported, so a study builds no audience list.
- A segment with the highest certainty is not necessarily the best segment; Google warns against that reading.
- A no lift result does not prove the ads failed; Google recommends a new test when certainty is below 50 percent.
- Results can move while a study runs, and Google recommends waiting for the final report.
Reading results channel by channel is a weak use of a study. Google's certainty page explains that each segment holds only part of the data, so segments are harder to read with confidence, and that performance between segments "is often similar (high overlap of confidence intervals)." A study is not a way to rank YouTube channels against each other.
How should a YouTube channel plan be set up for a Brand Lift study?
Google's guidance for studies with too few responses, on the metrics page, reads like a checklist for channel planners. It says "more restrictive targeting types, such as placements, keywords, and retargeting, reduce the number of eligible viewers," that too small a geography limits unique viewers, and that studies ideally run at country level. A tight channel list and a measurable study pull in opposite directions, so the plan has to decide which matters more.
| Plan decision | Google's guidance | Planning response |
|---|---|---|
| Channel list size | Placements reduce eligible viewers and impressions | Use a broad, approved list or measure a less restricted campaign |
| Geography | Too small a geography limits unique viewers | Keep the measured campaign at a large region or country |
| Language | A survey serves in one language; match campaign language to it | Split English and French channels into separately measured products |
| Number of campaigns | Too many campaigns in one study can leave campaign level data too thin | Measure fewer campaigns, or one campaign per study |
| Frequency | A cap spreads ads to more people but may lower lift odds | Agree the tradeoff with the client before launch |
| Other media | Creative seen before or outside the study can contaminate the control group | Keep the same creative out of unmeasured YouTube campaigns |
Two setups break the control group outright. Google says a study using campaigns that target people who already watched the ad "does not build a control group," with a video ad sequence as its example. Changing a video in a live sequence also interrupts lift measurement, per Google's sequence setup page (opens in a new tab); see the YouTube ad sequencing guide.
What does a Brand Lift channel plan look like on a real brief?
Fictional example
A Greek yogurt brand in California
A fictional Greek yogurt brand launching in California wants to know whether its new 15 second ad is remembered. The agency's first plan is 40 cooking and breakfast recipe channels. The planner points out that a placement list plus a single state narrows the eligible audience, which Google says slows survey collection.
The plan changes shape. The measured campaign runs across California with a broad approved list of 300 food, cooking and family meal channels, one English survey and one ad recall question, sized to the one question minimum converted into the account's currency. A smaller Spanish language campaign on Spanish cooking channels gets its own product and its own Spanish survey rather than sharing the English one.
Before launch, the planner confirms the same ad is not running in an unmeasured campaign and writes down what the result can say: ad recall among Californians reached on YouTube, nothing about grocery sales.
How does PlacementGraph help plan the channels a study will measure?
Frequently asked questions
Can a Brand Lift survey ask about health or finances?
No. Google's setup page says Brand Lift surveys cannot collect viewer feedback on sensitive topics, listing health or medical information, financial status or situation, religion, political affiliation, sexual behavior or orientation, racial or ethnic information and status as a child under 13, among others. Surveys also cannot collect personally identifiable information.
Can a Brand Lift study be extended after it ends?
Google says a study generally cannot be extended beyond its set duration. To keep measuring, the advertiser adds a new flight or uses re-measurement, which starts a new study for the same product or brand. Google recommends re-measurement at key moments, such as new creatives or a major media change.
Does Brand Lift count people watching together on a TV?
Google reports a separate lifted users (co-viewed) metric that adds lifted users from co-viewed impressions on connected TV devices. Co-viewers have no profiles, so Google treats them as the same audience profile as survey respondents. The figure is modeled, not counted.
Why can a study end before its planned dates?
Google says a study can conclude once it reaches its target responses per metric, which it gives as approximately 4,100, even while the campaign still runs. Google states that ending early for that reason does not harm the validity of the results.
Sources
- About Brand Lift (opens in a new tab)Google Ads Help
- Set up Brand Lift (opens in a new tab)Google Ads Help
- Understand Lift measurement statuses and metrics in Google Ads (opens in a new tab)Google Ads Help
- About certainty of lift (opens in a new tab)Google Ads Help
- About lift studies (opens in a new tab)Google Ads Help
- Set up Brand Lift measurement (opens in a new tab)Display & Video 360 Help
- Create a video ad sequence campaign (opens in a new tab)Google Ads Help
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.