Skip to main content
AI Predictive Targeting is an add-on available in both Web Experimentation and Feature Experimentation. Contact your Customer Success Manager for details.
Kameleoon AI Predictive Targeting uses machine learning to calculate propensity scores in real time, so you can target visitors based on their likelihood to convert for a specific goal.

​
Prerequisites

Before you can use the Likelihood to convert condition in a segment, you must:
  1. Enable AI Predictive Targeting on at least one goal. See Enable AI Predictive Targeting on a goal.
  2. Have at least one trigger configured. For guidance on picking a meaningful trigger, see How to choose the right trigger.
  3. Wait for the AI model to complete its learning phase (7 days and 100,000 visits by default).

​
Create or edit a segment with Likelihood to convert

  1. Click Settings > Segments.
  2. Click New segment to create a segment, or click Edit on an existing segment.
  3. In the segment builder, expand the Conversions section.
  4. Click Likelihood to convert.
Condition selection
  1. Select a likelihood range. You can choose one of the predefined buckets (Very low, Low, Moderate, High, or Very high), each representing roughly 20% of the eligible audience. Alternatively, select Personalized to define a custom range between 0 and 100. A score of 0 indicates the lowest predicted conversion probability relative to other visitors, and a score of 100 indicates the highest.
  2. Choose the goal on which you enabled AI Predictive Targeting.
  3. Choose the trigger to associate with the prediction.
  4. Click Save to confirm.
You can also apply the Likelihood to convert condition directly to an experiment through quick targeting, without creating a reusable segment.

​
AI model status badges

Every segment that uses the Likelihood to convert condition displays a status badge, so you can monitor the AI model’s progress at a glance.

​
View learning progress

Hover over the AI Learning badge to see a tooltip with the exact training progress:
  • Time: X of 7 days elapsed
  • Traffic: X of 100,000 visits collected
AI Learning
While the model is in the AI Learning state, the Likelihood to convert condition is inactive. The segment targets visitors using only its other active conditions.

​
Monitor AI Predictive Targeting status from the Experiments dashboard

You can check an AI model’s status without leaving the Experiments dashboard:
  1. Hover over the information icon associated with an experiment.
  2. In the pane that opens, hover over the sparkles icon next to the segment.
  3. A tooltip displays the current AI status for that segment.
AI Predictive Targeting from Experiments dashboard

​
Use the segment in a campaign

After the AI model reaches AI Active status, add your segment to: