• Model Data Packages

  Polaris is a first-of-its-kind incrementality measurement product. The insights we provide through a powerful blend of media mix modeling (MMM) and experimentation are incredibly valuable. That said, those insights, in the form of well-understood KPIs, are only as good as the model. Therefore, we feel it is important to supply model information so the insights can be better understood. It is by no means mandatory to understand the model information in its entirety or at all, and it does requireFew readers

• Model Components

  Basics The media mix models trained by Polaris include several components. Each component captures a different performance factor. Those performance factors combine to make up your total country-wide performance. Organic Factors Organic performance factors include: Seasonality (day of week, month of year) Trends (country-wide, app-wide, category-wide, industry-wide) Promotions (discounts, product releases, events, specials) Organic activities (non-paid social media posts, viFew readers

• Model Coefficients

  Basics Coefficients can be thought of as the relative incremental contributions of each channel, campaign, and site for each country and metric. Coefficients are one of the keys to any media mix model and are direct precursors to the incrementality metrics that are the final output provided by Polaris. You can think of coefficients as a further breakdown of the marketing component of the model. Within the marketing component, a specific media mix combiFew readers

• Marketing Input Multicollinearity

  Basics Multicollinearity can be a problem in MMMs because the more correlated the spend/impressions of one channel is with other channels, the more difficult it is for the model to isolate the impact of that channel. That said, the longer the time period of the input data, the less likely multicollinearity will be a significant problem. The reason is, there are so many factors at play that determine spend and impressions for any given campaign, most of which are outside of the control ofFew readers

• Backtesting Error

  Basics Model error refers to the difference between backtesting predictions of the model and the actual observed values. It is meant to provide some insight into how well the model is able to predict a future holdout test set with its current fit. This is not a comprehensive evaluation of the model’s accuracy and cannot be used to validate a media mix model (MMM). Validation is only possible with ground truth incrementality, which is why experiments are so important. In fact, backtesFew readers