Overview
Kohort models your campaigns using either daily or weekly predictions. The right choice depends on your app's data characteristics - specifically how noisy or sparse it is. Our goal is to deliver the most accurate and stable predictions for your use case.
Daily Predictions
Daily predictions offer granular, day-level forecasting and work best for apps with consistent, stable transaction patterns. Most IAA games perform well in this setting.
Weekly Predictions
Weekly predictions aggregate data over a wider window, smoothing out day-to-day noise. They are recommended for:
High IAP revenue apps where transaction signals are inherently noisy
Apps with sparse conversions where daily volume is too low for reliable daily estimates
The trade-off is less granularity, but significantly more stable and consistent results.
How We Determine Your Prediction Frequency
Your prediction frequency is determined in consultation with your Customer Success representative based on:
Signal noise: The noisier your data, the stronger the case for weekly predictions
Transaction density: Sparse daily conversions favour weekly aggregation
Revenue model: Apps driven by high-percentage IAP revenue typically benefit from weekly modelling
Accuracy performance: If daily predictions consistently fall below accuracy targets, weekly predictions may be trialled as an alternative
There are no rigid rules. This is an ongoing optimisation based on how your app's data behaves in our models.
Data Requirements
Both prediction modes require a minimum volume of data for reliable estimates. Noisier apps naturally need more data before predictions stabilise. Your Customer Success representative will advise on data readiness and any adjustments needed.
What This Means for You
Your prediction frequency is configured by the Kohort team, no action is required on your end. If you'd like to know which mode your app is using or have questions, reach out to your Customer Success representative.