A conversion lift study allows advertisers to evaluate the additional business driven by Facebook ads more accurately and make future marketing decisions based on the results.
Conversion lift studies are available without additional cost for qualifying campaigns. Please contact your Facebook sales representative to get more information.
In this article:
About measuring Conversion Lift
Measuring lift is beneficial to your campaign strategy because Facebook Conversion Lift allows you to gain insight into the causal impact of your ads on true business value, which can affect your future marketing decisions.
In the past, most online activity occurred on desktop devices, where cookie-based measurement systems were effective at tracking conversions. Today, much of our content consumption happens across different devices throughout the day, making measuring results more challenging than ever.
Conversion Lift uses a rigorous methodology to capture the impact that Facebook ads have in driving business. This methodology addresses the new measurement challenges facing today's marketers, such as
- Over-reliance on clicks: Last-click attribution is less useful for display environments, and counting clicks may not account for the value created by simply seeing an ad. By allowing advertisers to measure the impact of exposure to an ad across devices, conversion lift measurement offers a more holistic view of an ad's performance.
- The rapid shift to mobile: We live in a world where people use multiple devices throughout the day and a large share of purchases still happen in a physical shop. Facebook Conversion Lift measures people, not cookies, which allows you to track the path of conversion across multiple devices over time.
- Ineffective testing methods: Until now, there has been no universal standard for determining the direct impact of digital advertising. Few other measurement solutions determine the actual causal impact of marketing, however, conversion lift does.
How we approach conversion lift methodology on Facebook, Instagram and Audience Network
Conversion lift measures the incremental impact that your Facebook, Instagram or Audience Network ads have on your sales outcomes using an intent-to-treat methodology. Facebook uses intent to treat methodology because of the dynamic nature of our delivery system. This is a widely used and rigorous methodology that offers an unbiased way to measures sales that were caused by your ad campaign.
We start by splitting the Facebook population into two randomised groups: one who can see your ads (the test group) and one who is held back from seeing your ads (the control group). As you run your campaign, we include people in both groups who are eligible to see your campaign (the people you intended to reach) based on who you target and their activity on the platform. While the control group isn't shown your ads, the test group ads run as they normally would.
Because of how our delivery system optimises ad performance, an intent-to-treat methodology provides benefits over alternatives such as placebo ad or ghost ad lift methodologies.
- A placebo ad methodology shows an ad
that is unrelated to your product (usually for a charity) to a portion of your audience as opposed to holding out your ads from a control group. However, because our delivery system optimises ad performance differently for different ads, the two groups in a placebo ad test would not be comparable.
- A ghost ad methodology is when the delivery system attempts to serve an ad but doesn't, and instead records when an ad would have been shown. We don't use this methodology because it doesn't take into account the feedback loop from people who interact with your ads on multiple impressions and doesn't provide comparable groups.
For the above reasons, we use the intent-to-treat methodology to determine whether there was a statistically significant lift in the conversion rate between your test and control groups.
How conversion lift works
This is the process we use to implement our intent-to-treat conversion lift methodology:
- First, we randomise everyone on Facebook, Instagram and Audience Network into two groups uniquely for your study. One group will be able to see your ads (the test group) and one will be held back from seeing any of your ads (the control group).
- As you run your campaign, we include people in your lift study who are eligible to see your ads in both the test and control groups. In the test group, your ads run like they normally would. Some of the people who are eligible will see your ads and some will not. In the control group, we prevent people from seeing your ads.
- During the study, we capture sales outcomes in the test and control groups. This includes actions that people are taking based on the conversion events you are already sending us from the pixel, mobile App Events and offline conversions. We only include actions that happen during the study period in both the test and control groups.
- When the study is complete, we calculate the incremental impact of your advertising by comparing the test and control groups. Because the only difference between these two groups is that one saw your ads and one did not, we know that any impact was caused by your campaign. A positive, statistically significant lift signifies that your ads were driving business impact.
The results of the study are made available through your Facebook account representative and in Ads Manager.
Note that API results will not be available in Ads Manager and are only accessible through the conversion lift API. Please see the "Reporting" section of the Facebook for Developers Marketing API page for more information.
How we determine who's included in a conversion lift study
Facebook uses randomisation to determine whether a person is in the test or control group. A random number generator is used to create a unique salt (also known as a seed) for your study. This salt is combined with each person's unique identifier and run through a hashing function to determine which group the person belongs to. This ensures that people are randomly assigned to test or control and that they will stay in that same group throughout the test, regardless of which device or platform they are using. When new people join Facebook, they are randomised in the same manner between test and control.
We include people who are eligible to see your ads dynamically based on your targeting criteria and their activity on the platform. Test group audiences see your ads as they were normally would. Control group audiences are prevented from seeing your ads.
As lift tests use randomisation to determine whether a person is in test or control, the people eligible for your ads are comparable and will inherently be balanced on factors such as demographics, outside ad exposure and purchase behaviour.
Who should run a conversion lift study
Not all advertising campaigns are a good fit for a conversion lift study. Studies should be designed to be statistically powerful enough to provide actionable results.
To run a conversion lift study, you need
- a campaign that isn't live yet and will have sufficient statistical power for actionable results, and
- accessible conversion data, either through a Facebook Pixel, Facebook SDK or offline conversions data.
For more information on conversion lift studies and to see if one is suitable for your campaign, please contact your Facebook Sales Representative.
Conversion Lift is also available via an API. Please reference the Facebook for Developers Marketing API page for information on how to set one up.
When to set up a conversion lift study
Use conversion lift if:
- You want to see the true business value that Facebook is providing
- You want to assess how accurately your attribution model is capturing the value your ads create
- You want to measure real people, not
- You are testing major strategy changes against each other
- You are making a shift to measurement that can help you understand growth or incrementality
- You have a lot of other media in market, making it hard to attribute value accurately
- You are trying to understand the difference between correlation and causation
Don't use Conversion Lift if*:
- You are testing minor changes in strategies (split testing may be more helpful for minor changes and tactical decisions)
- You're not sharing your conversion data with Facebook
- Your campaign doesn't have enough power to measure the results you want
Please work with your Facebook Sales Representative to determine whether a conversion lift study is right for your campaign.
*If Conversion Lift isn't a feasible solution for your marketing campaign, you may want to try running a Facebook split test.
Objectives to include in conversion lift studies
There is no limit to the number of objectives you can add to a conversion lift study. If you want to measure it, you can add it. It is recommended to add all pixels, app events and/or data feeds in order to get the most comprehensive results. You should first set your primary objective: the fundamental goal of your campaign, and then determine all the other objectives you would like to measure.
Please work with your Facebook Sales Representative to discuss which objectives may be best for your campaign.
Learn more: Set up a conversion lift study
* Nguồn: Facebook