Skip to main content
Coming soon Uplift Prediction gives you a preview of the impact you can expect an incentive campaign to have on its target indicator, such as revenue, engagement rate, or number of activities performed, before you commit budget or go live.

​
Choose a Target Indicator

Before generating a prediction, you set the indicator you want to forecast impact on. Example indicators include:
  • Revenue (e.g. Total Amount or Incentivizable Amount from Sale)
  • Engagement rate (e.g. share of members performing a qualifying activity)
  • Number of activities performed (e.g. sales, redemptions, or events)
  • A custom numeric or count-based indicator relevant to your program
The indicator you choose should generally align with the behavior your incentive’s rules are designed to influence.

​
Generate a Prediction

You can generate an uplift prediction by chatting with Fielo AI from an incentive’s detail page. Simply confirm the target indicator, the population the incentive applies to (based on its segments, if any), and the timebox you intend to run it for. Fielo AI analyzes historical performance for the relevant population and data set(s), then models expected performance with the incentive active against a projected baseline of performance without it.
Predictions are most reliable when there’s sufficient historical data for the population and indicator in question. If historical data is sparse, Fielo AI will let you know that its projection carries wider uncertainty.

​
Reading Your Prediction

An uplift prediction includes:
  • Baseline Projection: The expected value of the target indicator over the timebox if the incentive were not run.
  • Incentivized Projection: The expected value of the target indicator over the timebox with the incentive active.
  • Predicted Uplift: The difference between the incentivized and baseline projections, shown in absolute and percentage terms.
  • Confidence Range: A range around the predicted uplift reflecting the uncertainty of the projection.
You can also ask Fielo AI to explain the reasoning behind a prediction, or to model how the uplift might change if you adjust the incentive’s criteria, reward values, segments, or timebox.

​
Refine Before You Launch

Because predictions respond to changes in an incentive’s configuration, you can use Uplift Prediction iteratively: adjust a rule’s criteria or reward, regenerate the prediction, and compare results before activating the incentives.