Brand Lift research in digital advertising
Why CTR and ROAS do not replace Brand Lift
Clickability and conversions answer the question 'did the creative work right now'. Brand Lift answers a different question: has the audience's attitude towards the brand changed after contacting the advertisement, regardless of whether they clicked or not. For performance campaigns with a clear CTA, this is not critical — here they adjust conversions and ROAS. But for image launches, where the goal is product recognition of a new product, rebranding, or entering a new segment, conversion metrics simply do not show the main effect.
The measurement mechanism is simple: one part of the audience is shown an advertisement (test group), the other is not or is shown neutral content (control group). The difference in responses to the same questions about the brand between these groups is called the lift. This approach excludes distortion of the result due to the general increase in interest in the category on the market — you see the pure effect of your campaign itself.
| Metric | What it measures | Horizon |
|---|---|---|
| CTR | Click on the ad right now | Instantly |
| ROAS / conversions | Direct sales from advertising | Short-term |
| Brand Lift | Change in brand perception | Medium and long-term |
| Search Lift | Increase in brand-related search queries | Mid-term |
| Conversion Lift | Direct impact on sales through test/control | Short-term |
How test and control groups are formed
The credibility of the study rests on one condition: the groups must be statistically identical in everything except for the fact of being shown the advertisement. Platforms do this automatically through user randomization, but it's useful to understand the logic so as not to dispute the results afterward.
- A pool of users matching the campaign targeting is formed—usually 10 000 people in each group, otherwise the platform won't achieve statistical significance.
- The pool is randomly divided into a test group (will see the ad) and a control group (will not see it, will see a placebo creative).
- Both groups are asked the same questions via surveys or interactive widgets — about logo recognition, intent to purchase, and attitude toward the brand.
- A sample is taken from each group for analysis of responses — usually about half of the total number of invited participants, the rest do not respond or are filtered out due to response quality.
- The difference in the proportion of positive responses between the groups is the lift — in percentage points.
In Google Ads, the minimum budget for one question on 10 days is from 5 000 to 15 000 dollars depending on the country and its classification into group A, B, or C. In Yandex.Direct, the threshold is different: reach of at least 3 million users, total budget of at least 2 million rubles and at least 140 thousand rubles per day. This is not bureaucratic formalism — without such a sample size, the study simply will not yield a statistically significant difference, and the result will be noise.
What lift to consider a good result
There is no universal threshold of ‘good/bad’, and any benchmark without context of the category is deceitful. But reference points from real cases give an idea of scale:
- Bank «Point» — a rise in recognition by 21,4 percentage points after the campaign
- «PIK» — a rise in memorability of the key message by 82%
- «Petrovich» — brand awareness in the test group was higher by 2,5 times
How to read these figures in practice: a lift of 2-5 points in recognition for a large, already known brand — is quite a normal result, with a low growth ceiling there. And for a new player on the market, growth of 15-20 points is a realistic and achievable goal precisely because the starting base of brand knowledge is close to zero. You should look not at the absolute number, but at the direction in which the indicator moves relative to the brand's starting point.
How to interpret the results and what to do next
First, what to look at — not the overall lift, but the breakdown by audience segments and by creatives. If the lift by recognition is high, but by purchase intent almost zero — the message is remembered, but does not motivate to action, and here it makes sense to change not the media planning, but the offer itself in the creative. If on the contrary, recognition is low, but intent is high — the advertisement probably reaches an already warm audience, and there is a sense to expand the reach to cold segments.
Results do not come immediately: standard Brand Lift in Google gives data in 3-14 days, Enhanced Lift with a triple budget — in 9-14 days, Amazon Ads study by request takes 10 working days. Optimizing a campaign based on interim figures before accumulating a significant sample — a typical mistake that leads to premature and incorrect conclusions.
Cross-channel measurement and linking with other metrics
A separate issue that is rarely examined: a brand today rarely lives in a single channel. A campaign runs simultaneously in Google, VK and on reach platforms, and isolated Brand Lift for each channel gives a distorted picture — channel effects intersect and overlap with each other for the same audience. The correct approach is to combine Brand Lift, Search Lift and Conversion Lift data into a single attribution model, rather than evaluate each channel on its own scale.
Here it is important to distinguish between tasks: Sales Uplift measures short-term effect on sales and is suitable for promo campaigns and performance launches, whereas Brand Lift works over the long term and shows cumulative effect on perception. Using them as interchangeable metrics is a crude methodological error. A proper linkage looks like this: Brand Lift shows whether awareness and loyalty are growing, Search Lift indicates whether this awareness translates into active search interest, and Conversion Lift and ROAS capture whether this interest reaches the cash register.
When Brand Lift is not working and what to do with small business
The method is meaningless with small budgets and narrow audiences — if you don't have those 10 000+ users per group and the required daily budget, the platform will either refuse to launch the study or will deliver a statistically insignificant result that cannot be relied upon. Brand Lift also performs poorly in narrow B2B-niche markets with extremely small target audiences — there simply isn't enough volume for randomization.
For small businesses, real alternatives are as follows:
- Conducting surveys before and after the campaign via forms, email databases, or social networks — without strict randomization, but with a basic comparison of dynamics
- Tracking branded search traffic in Yandex.Wordstat and Google Trends as an indirect indicator of brand awareness growth
- Monitoring direct visits to the website and brand mentions in social networks before/after launching advertising
- UTM tags and comparing organic traffic by branded queries over time
This won't provide the same accuracy as native Brand Lift in the advertising platform dashboard, but it will at least allow you to roughly assess whether brand perception is moving in the right direction, without spending hundreds of thousands of rubles just to measure it.
Why CTR and ROAS do not replace Brand Lift
Clickability and conversions answer the question 'did the creative work right now'. Brand Lift answers a different question: has the audience's attitude towards the brand changed after contacting the advertisement, regardless of whether they clicked or not. For performance campaigns with a clear CTA, this is not critical — here they adjust conversions and ROAS. But for image launches, where the goal is product recognition of a new product, rebranding, or entering a new segment, conversion metrics simply do not show the main effect.
The measurement mechanism is simple: one part of the audience is shown an advertisement (test group), the other is not or is shown neutral content (control group). The difference in responses to the same questions about the brand between these groups is called the lift. This approach excludes distortion of the result due to the general increase in interest in the category on the market — you see the pure effect of your campaign itself.
| Metric | What it measures | Horizon |
|---|---|---|
| CTR | Click on the ad right now | Instantly |
| ROAS / conversions | Direct sales from advertising | Short-term |
| Brand Lift | Change in brand perception | Medium and long-term |
| Search Lift | Increase in brand-related search queries | Mid-term |
| Conversion Lift | Direct impact on sales through test/control | Short-term |
How test and control groups are formed
The credibility of the study rests on one condition: the groups must be statistically identical in everything except for the fact of being shown the advertisement. Platforms do this automatically through user randomization, but it's useful to understand the logic so as not to dispute the results afterward.
- A pool of users matching the campaign targeting is formed—usually 10 000 people in each group, otherwise the platform won't achieve statistical significance.
- The pool is randomly divided into a test group (will see the ad) and a control group (will not see it, will see a placebo creative).
- Both groups are asked the same questions via surveys or interactive widgets — about logo recognition, intent to purchase, and attitude toward the brand.
- A sample is taken from each group for analysis of responses — usually about half of the total number of invited participants, the rest do not respond or are filtered out due to response quality.
- The difference in the proportion of positive responses between the groups is the lift — in percentage points.
In Google Ads, the minimum budget for one question on 10 days is from 5 000 to 15 000 dollars depending on the country and its classification into group A, B, or C. In Yandex.Direct, the threshold is different: reach of at least 3 million users, total budget of at least 2 million rubles and at least 140 thousand rubles per day. This is not bureaucratic formalism — without such a sample size, the study simply will not yield a statistically significant difference, and the result will be noise.
What lift to consider a good result
There is no universal threshold of ‘good/bad’, and any benchmark without context of the category is deceitful. But reference points from real cases give an idea of scale:
- Bank «Point» — a rise in recognition by 21,4 percentage points after the campaign
- «PIK» — a rise in memorability of the key message by 82%
- «Petrovich» — brand awareness in the test group was higher by 2,5 times
How to read these figures in practice: a lift of 2-5 points in recognition for a large, already known brand — is quite a normal result, with a low growth ceiling there. And for a new player on the market, growth of 15-20 points is a realistic and achievable goal precisely because the starting base of brand knowledge is close to zero. You should look not at the absolute number, but at the direction in which the indicator moves relative to the brand's starting point.
How to interpret the results and what to do next
First, what to look at — not the overall lift, but the breakdown by audience segments and by creatives. If the lift by recognition is high, but by purchase intent almost zero — the message is remembered, but does not motivate to action, and here it makes sense to change not the media planning, but the offer itself in the creative. If on the contrary, recognition is low, but intent is high — the advertisement probably reaches an already warm audience, and there is a sense to expand the reach to cold segments.
Results do not come immediately: standard Brand Lift in Google gives data in 3-14 days, Enhanced Lift with a triple budget — in 9-14 days, Amazon Ads study by request takes 10 working days. Optimizing a campaign based on interim figures before accumulating a significant sample — a typical mistake that leads to premature and incorrect conclusions.
Cross-channel measurement and linking with other metrics
A separate issue that is rarely examined: a brand today rarely lives in a single channel. A campaign runs simultaneously in Google, VK and on reach platforms, and isolated Brand Lift for each channel gives a distorted picture — channel effects intersect and overlap with each other for the same audience. The correct approach is to combine Brand Lift, Search Lift and Conversion Lift data into a single attribution model, rather than evaluate each channel on its own scale.
Here it is important to distinguish between tasks: Sales Uplift measures short-term effect on sales and is suitable for promo campaigns and performance launches, whereas Brand Lift works over the long term and shows cumulative effect on perception. Using them as interchangeable metrics is a crude methodological error. A proper linkage looks like this: Brand Lift shows whether awareness and loyalty are growing, Search Lift indicates whether this awareness translates into active search interest, and Conversion Lift and ROAS capture whether this interest reaches the cash register.
When Brand Lift is not working and what to do with small business
The method is meaningless with small budgets and narrow audiences — if you don't have those 10 000+ users per group and the required daily budget, the platform will either refuse to launch the study or will deliver a statistically insignificant result that cannot be relied upon. Brand Lift also performs poorly in narrow B2B-niche markets with extremely small target audiences — there simply isn't enough volume for randomization.
For small businesses, real alternatives are as follows:
- Conducting surveys before and after the campaign via forms, email databases, or social networks — without strict randomization, but with a basic comparison of dynamics
- Tracking branded search traffic in Yandex.Wordstat and Google Trends as an indirect indicator of brand awareness growth
- Monitoring direct visits to the website and brand mentions in social networks before/after launching advertising
- UTM tags and comparing organic traffic by branded queries over time
This won't provide the same accuracy as native Brand Lift in the advertising platform dashboard, but it will at least allow you to roughly assess whether brand perception is moving in the right direction, without spending hundreds of thousands of rubles just to measure it.
Recent articles
MegaFon, BMW, VTB and more 240+ companies. See how it looks in the showcase reel:
Our branding agency is approached to order:
Branding agency services
