In short
Propensity modeling is a statistical method that estimates how likely a user is to take a specific action, such as buying, upgrading or leaving. Each user gets a score based on past behavior and traits. Teams use these scores to decide who to target, with what, and when.
Why it matters for apps
Propensity scores help teams spend effort where it changes outcomes. Instead of messaging everyone, you focus on users close to an action.
- One model per action: A propensity to upgrade is a different score from a propensity to churn. Each needs its own target and data.
- Target the persuadable: Users with a very high score may act anyway. Users with a middle score are often where a nudge makes the difference.
- Good inputs matter: Scores are only as good as the behavior you track. Clean, consistent event data is the starting point.
- Common mistake: Treating the score as the goal. A score has no value until it drives a decision you can test against a control group.
Industry examples
An investing app scores users on their likelihood to open a savings plan and shows a goal challenge to the middle group (hypothetical).
A news app scores readers on their likelihood to subscribe and offers a free premium week to those near the threshold (hypothetical).
Build it with StriveCloud
StriveCloud does not calculate propensity scores. It supplies the inputs: events, stats per day, week, month or all time, and segments that update as users act, which webhooks can send to your own model. Teams can then use segments in StriveCloud to target the groups their model selects with challenges, rewards or messages.
Want to try this? Book a demo
Frequently asked questions
What is a propensity score in marketing?
A propensity score is a number, often a probability, showing how likely one user is to take a given action, such as buying, upgrading or churning.
What is the difference between propensity modeling and churn prediction?
Churn prediction is one type of propensity model, focused on leaving. Propensity modeling covers any action, such as purchases, upgrades or referrals.
What data do you need for propensity modeling?
Historical behavior linked to a known outcome: events like logins, purchases and feature use, plus whether each user later took the target action.