Predictive Modeling

What is predictive modeling?

Predictive modeling is a form of data mining technology that predicts future events or outcomes by analyzing historical and current data. So, predictive modeling involves collecting data, developing a statistical model, making predictions, and validating (or optimizing) the model over time.

Predictive modeling by Branch uses an industry-unique, predictive algorithm that incorporates historical attributions to deliver high-accuracy data where there is no universal ID.

Predictive modeling is privacy-first by design, and is possible because Branch is the top linking platform in the world. We’re able to leverage truth signals from Branch’s scale and distribution across channels to offer the most accurate and robust device-level attribution product in the industry.

How predictive modeling is useful

Predictive modeling is available to you when users opt into ad tracking via Apple’s AppTrackingTransparency (ATT) modal, and is always available to measure organic, non-paid channels. This makes predictive modeling useful in two areas:

  • Paid Ads Attribution: Although you must get user opt-in to device level ad tracking to use Predictive Modeling for paid ad attribution, there are still edge cases where the IDFA may not be available (such as web-to-app conversions, or the IDFA not populating in the tracking link). In these cases, Branch will be able to use Predictive Modeling to accurately attribute that user to the proper ad.
  • Organic Channels: Apple’s policy language applies only to tracking in the context of ad attribution, meaning deep linking and measurement of non-ads channels (email, organic social media, web-to-app conversions) can be powered by Predictive Modeling.