mack grenfell

Brand Lift Testing

Stay up to date

Get notified, once a week, of the latest articles.

Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.

It's easy to get caught up in the idea that the whole purpose of digital advertising is to drive conversions. While doing so might provide short-term benefits, using digital channels to build a brand is one of the best ways to ensure long-term success.

One of the reasons some advertisers shy away from brand building is that it's difficult to measure, and that it's often not clear what success looks like. Brand lift testing looks to do away with those worries by providing a way to measure the effects of brand building activity.

The idea behind brand lift testing is very similar to conversion lift testing; if you haven't read the article on that then I'd recommend giving it a skim.

How does brand lift testing work?

Just as with conversion lift testing, we first split our target audience into two cells. There's a control cell, of people who are ineligible to see our ads, and an experiment cell, of people who are eligible to see our ads.

The splitting between control and experiment cells is handled in exactly the same way as in a conversion lift test; it's done randomly to ensure that audience selection isn't biased by Facebook's ad delivery methods.

Once the control and experiment cells are in place, ads are shown as normal to the experiment cell, and withheld from the control cell. Users from both cells are polled for the duration of the test with questions such as "Have you heard of [brand name]?", and asked to choose from the options of yes, no, and not sure.

We can then look at differences in responses between the cells to understand what impact our ads are having on people's answers to these questions. For instance, let's say we choose to ask the question "Have you heard of [brand name]?" as part of our brand lift test. This question is designed to measure brand awareness.

Say we notice that 10% of people respond yes when not exposed to our ads, compared to 15% of people who are exposed to our ads.

This means that our ads have raised brand awareness by 5% (15% - 10%). This 5% figure is known as the brand lift percentage. If there are 10,000 people in each of our control and experiment cells, then we can estimate that 5% of those people became aware of our brand as a result of our ads. This value, 500 people, is what's known as the brand lift.

The last metric to look at is cost per brand lift, which is the amount of money spend on the ads over the duration of the study, divided by the brand lift. If we spent $100 on ads during our study, in which we caused a brand lift of 500 people, then the cost per brand lift would be $0.20.

The cost per brand lift tells us the amount we need to spend on ads to make one more person aware of our brand.

Beyond brand awareness

If you want to measure brand awareness, as in the example above, you can set your brand lift study up to include the question "Have you heard of [brand name]?"

You're not limited to just this question however. Other questions that you can ask are:

Each of these questions probes a different area of how users see your brand, and can deliver very different results when asked to users.

When setting up a brand lift study, you're able to choose up to 3 of the above questions, but Facebook require Ad Recall to be one of them. This is because Facebook use the results from this internally to train the delivery algorithms for brand awareness campaigns, allowing them to show ads to those most likely to recall them.

How do you set up a brand lift study?

Unlike with conversion lift tests, brand lift tests can only be set up through Test & Learn, by selecting the option "How much impact are my Facebook ads having on brand perception?"

When setting up a brand lift test, you'll be asked to select a few options:

Once you've reviewed your test and confirmed, your test will be ready to go live at your previously selected start time.

Note that you won't be able to view your results straightaway, you'll need at least 250 responses to each question before Facebook allows you to view responses to that question. Depending on how much spend you're putting behind the test, this can take anywhere from days to weeks. Take this with a pinch of salt, but Facebook recommend spending at least 30k USD on campaigns involved in a brand lift test in order to get reliable results.

Single campaign brand lift tests

If you've read the article on conversion lift testing, you might recall that Test & Learn only lets you run account-wide conversion lift tests. Though this is what I've focused on above, Test & Learn also lets you run brand lift tests specific to a single campaign.

You can set this up in Test & Learn by selecting the "How much impact is my campaign having on brand perception?" option. The process for setting up this test is identical to setting up an account-wide test, as outlined above, except that you'll also be asked to specify which campaign you want to run the brand lift test on.

The problem with "US English"

If you're having trouble setting up a brand lift study because you're only allowed to poll users who have Facebook set to US English, and you're not targeting the US, you'll see an error. This error appears erroneously, because Facebook can run brand lift studies for users with other language settings, e.g. UK English.

To get around this error, change at least one of the campaigns involved in your brand lift test to target the US, set up the test, and then remove the US from your targeting. This will trick Facebook into allowing the test to be set up, and won't lead to any issues while the test is running.

If you're concerned about wasting any spend on the US while you're setting up your test, feel free to pause the campaigns or ad sets that have been set to target the US, until you've removed the US from targeting.