Branch의 예측 모델링은 다른 MMP가 제공하는 것과 어떻게 다릅니 까?
Traditional MMPs are in the ads business. Predictive Modeling is possible because Branch is the top linking platform in the world, which allows us to build a sophisticated attribution model that includes truth signals from every channel and platform. These signals are device-level and privacy-safe, and no other MMP has them.
For example, we can see when they are accessing the device from common public networks like the home or office, as well as their private addresses when they are on cellular. With this, we can detect if a user switches from cellular to their home network between the click and the app session to match correctly where other simple technologies will fail. Importantly, we can use this insight to improve accuracy both when making an attribution, and when deciding not to attribute.
The best analogy is one based on statistics. When you’re running an experiment where there’s uncertainty, the sample size matters. You need enough samples to accurately determine the distribution. Traditional MMPs use one sample (the IP of the last click) to match the IP of the app session. Branch predictive modeling uses a huge number of samples, collected over the user’s anonymous profile, making it much more accurate than a single data point.
Updated over 2 years ago