What is propensity scoring in marketing?

Propensity scoring assigns each user a probability score (0 to 1) for a target action: purchasing, converting, churning, or becoming a high-LTV customer. The score is built from historical outcome data using a machine learning model and used to concentrate ad spend on the highest-probability segments.

How is a propensity score built?

A propensity model requires historical labeled data: records of which users did or did not take the target action, plus behavioral and demographic features for each user. The model learns which feature combinations predict the outcome. Minimum data requirements depend on outcome frequency; rare events need larger datasets.

How do you operationalize a propensity score in advertising?

An operationalized propensity score is pushed to ad platforms as a custom audience (Meta Custom Audience, Google Customer Match) or used to adjust bid modifiers at the audience level. A score updated weekly and synced to platforms on the same cadence functions as a live targeting lever, not a static insight.

Propensity Scoring

Propensity scoring assigns each prospective customer a probability score for a target action, enabling advertisers to concentrate spend on the highest-probability segments.

Common questions

What is propensity scoring in marketing?: Propensity scoring assigns each user a probability score (0 to 1) for a target action: purchasing, converting, churning, or becoming a high-LTV customer. The score is built from historical outcome data using a machine learning model and used to concentrate ad spend on the highest-probability segments.
How is a propensity score built?: A propensity model requires historical labeled data: records of which users did or did not take the target action, plus behavioral and demographic features for each user. The model learns which feature combinations predict the outcome. Minimum data requirements depend on outcome frequency; rare events need larger datasets.
How do you operationalize a propensity score in advertising?: An operationalized propensity score is pushed to ad platforms as a custom audience (Meta Custom Audience, Google Customer Match) or used to adjust bid modifiers at the audience level. A score updated weekly and synced to platforms on the same cadence functions as a live targeting lever, not a static insight.

Propensity scoring uses historical data about customer behavior to build a model that predicts the likelihood of a future event for each individual. In a marketing context, the target event might be: this user will convert within 30 days, this user will become a high-LTV customer, or this user is likely to churn. The model assigns each user a score that can be used to inform bidding, targeting, and messaging decisions.

The data inputs vary by application. Acquisition propensity models might use behavioral signals (pages visited, time on site, content consumed), demographic proxies, channel history, and comparable customer profiles. Churn propensity models use engagement frequency, support interactions, and usage patterns. The key requirement is historical labeled data: you need a sufficient history of outcomes to train the model on.

Propensity scores become most valuable when they are operationalized: pushed into ad platform audiences so media spend concentrates on high-propensity segments, used in CRM to prioritize outreach, or used to adjust bid strategies at the user level. A score that lives in a spreadsheet is an insight; a score that is updated weekly and synced to Meta Custom Audiences and Google Customer Match is a performance lever.

Example

A subscription software company builds a 30-day conversion propensity model using trial behavior data. Users scoring above 0.7 are pushed to a Meta Custom Audience for high-intent creative. Users scoring 0.3 to 0.7 receive nurture sequences. Users below 0.3 are excluded from paid retargeting to reduce wasted spend. Trial-to-paid conversion rate rises 22 percent on the same budget.

Always on · Algorithm-led · Human-approved

Walk this operation against your account.

Thirty focused minutes with one of the founders. We map Propensity Scoring against your stack and tell you which entry point fits your stage.

Book a 30-minute call →Or write us