Why is Causal Impact Needed in a Post-IDFA World?

5 minute read

Until recently, marketers could track customer journeys from start to end. It let them generate millions of data points to get insights on where to optimize their marketing efforts. However, Apple’s iOS 14.5 and later versions have deprecated the Identifier for Advertisers (IDFA) due to the rising focus on privacy. Without IDFA, marketers can no longer measure online activities using traditional attribution models.  

Causal impact lets marketers understand what number of their conversions belong to their campaigns. The methodology makes predictions based on real data coupled with hard-to-track marketing activities like product launches and TV campaigns. Causal impact uses historical data to predict what would have happened if there would not have been a marketing campaign.

Dividing audiences has become more challenging after the iOS 14.5 rollout. Without the IDFA, marketers should look for new ways to split their customers. For example, they can focus on splitting clients based on their geography, OS versions, types of devices, platforms, etc. However, focusing on such parameters may have some caveats. For example, while targeting customers based on their geography, customers may be in different locations on the same day, thus affecting the split test results. Running split tests based on the device models is the most effective way to market ads in the post-IDFA world. 

In incrementality testing, the audience is divided into the treatment group and the control group. While running tests, ads are delivered to the treatment group. The control group remains unexposed to ads. In incrementality testing, the causal impact is used to estimate the effect of the no-ID marketing campaigns. Unlike correlation, the causal impact proves that something is happening because of other factors. When used in mobile marketing, causal impact determines whether customers change their behavior on seeing ads. For example, a customer may buy something or install an app after seeing an ad. 

In causal impact analysis, the behavior of a treatment group is predicted based on the control group. The uplift is calculated as the difference between the actual and predicted counterfactual behavior recorded at a specific time. To measure the impact of the ads campaign while it’s running, clicks and impressions (or other KPIs of the treatment group) are coupled with several covariates of the control group. For example, when one campaign runs during a holiday, conversion rates go up. Based on this information, the covariate predicts the behavior of the treatment group during the campaign period. A different campaign can result in increased conversions on Sundays. The causal impact analysis captures this trend and accounts for the difference. All changes that cannot be explained are measured as uplift. 

In a word, casual impact analysis helps marketers evaluate their campaigns in the post-IDFA conditions when tracking user activities has become harder or impossible. If you need expert advice on the value of your marketing campaigns, contact the Atypical Digital Team.

Leave a Reply

Your email address will not be published.