Test and Learn guide: Brand lift test best practices

30/11/2019

In this article, we'll explain what mistakes to avoid and best practices to follow to create a brand lift test likely to produce meaningful results.

Choosing the right campaign

When setting up your test, ensure that the campaign you want to use:

  • Will be active for the entire time your test is running. If you need to edit any campaign schedules to accommodate your test schedule, do so.
  • Has a budget of at least USD 30,000 available to spend during the test. We rely on polls of your target audience to measure brand perception. This means we need a large audience in order to collect enough responses to generate meaningful results. We require USD 30,000 as a minimum budget to ensure we can reach enough people to do that.
  • Hasn't been running recently. This can mean creating a new campaign to test, or reactivating an old one. We recommend this because campaigns that have been running for a while before you test them are likely to have already had most of their impact. That impact won't be captured by a test that starts running towards the end of a campaign's schedule. Plus, if your campaign is almost over anyway, your results may not be very useful. We recommend testing early on so you can apply what you learn.
  • Is targeted at US English speakers. Our poll questions only appear in US English, so you can only test with speakers of that language. This also means that you must include people in the US or Canada (populous regions where US English is spoken) in your target audience to ensure that it's large enough to get meaningful results.
  • Isn't targeted exclusively/primarily at people under the age of 18. We can't poll anyone under the age of 18.
  • Doesn't use the Reach & Frequency objective. Reach & frequency campaigns can't work properly with the holdout groups required to run a lift test.
  • Isn't being used in another lift test. To maintain the scientific rigour required of lift tests, we can't have one campaign active in multiple at the same time.
  • Isn't associated with a Page that has a name that would be in breach of our ad policies when used in a poll. Remember that when your Page name is associated with a poll, it becomes subject to our ad policies. Our ad policies are distinct from our Page policies. If your Page name breaches our ad policies, we won't run your poll.

Setting an effective schedule

When setting your test schedule, ensure that it's:

  • At least two weeks in length. This is a required minimum. Your test may need more time to produce meaningful results. We recommend testing for at least four weeks. A test can run for a maximum of 90 days.
  • Aligned with the schedule of your campaign. If you don't want to adjust your campaign's schedule to accommodate your test, adjust your test's schedule to accommodate your campaign.
  • Not overlapping with marketing events that may skew results. Results from tests that ran during times when ad auctions were especially competitive (e.g. holiday advertising) or when you're spending an unusually high amount may not be widely applicable. In general, we recommend testing in situations that reflect the typical state of your advertising strategy. This will lead to the most meaningful and actionable results.

Optimising your questions for your campaign goals

There are six question formats available for brand lift test polls. You have to have a standard ad recall question, and you can choose up to two additional questions. If you're trying to decide which additional questions to add to your poll, consider where in your marketing funnel the campaign you want to test fits.

By marketing funnel, we mean the process by which someone goes from not being aware of a product to purchasing it. Brand-lift studies are best suited to measuring your ads' effects on generating and nurturing leads. Within this part of the funnel, there are a few distinct steps to consider:

  • Awareness
  • Interest
  • Consideration
  • Intent
If the campaign you're testing is focused on awareness, consider adding a standard brand awareness and/or familiarity question.

If the campaign you're testing is focused on interest and/or consideration, consider adding an abstract favourability question.

If the campaign you're testing is focused on consideration and/or intent, consider adding a recommendation and/or action intent question.

* Nguồn: Facebook