NOTES
- To add Predictive Audiences to your Iterable account, talk to your customer success manager.
- Predictive Goals was renamed to Predictive Audiences, and some screen shots may still reflect the previous name.
# In this article
# How Predictive Audiences works
Predictive Audiences analyzes your historical data to predict which users are likely to convert on your business goals in the future. Predictive Audiences also provides insight into the events and properties that contribute most to a prediction, identifying those that may increase or decrease the likelihood of a certain outcome. These insights can help you connect customers with experiences that match their interests and encourage conversion, while promoting the outcomes you seek.
Predictive Audiences bridges the gap between insights and actions.
Let’s say your business wants to connect customers with products they love; Predictive Audiences can help you identify who’s likely to purchase luxury brands, so you can alert them to new arrivals. Or, if your business wants to mitigate churn, you can identify users who might disengage from your messaging, so you can reconnect with them before they do.
Use Predictive Audiences to:
Identify segments of users who are likely to convert on business goals that you want to optimize (like purchases and promotional clicks), or minimize (returns or unsubscribes), so you can send them personalized messaging. By sending a segment to a dynamic list, you can ensure that campaigns go to all of the contacts who match the segmentation query at send time.
Identify which of your project's relevant user properties and first-party data (like purchase, system, and custom events) you want to increase or decrease in frequency to promote desired outcomes. For example, let's consider that you create a predictive audience to identify users who are likely to make multiple purchases a month. Predictive Audiences uncovers that it's likely that these users will also complete
OfferRedeemedevents. To increase the likelihood that users will convert, you might increase the number of special offers you send to your customers.Generate conversion probability scores for use in journeys to direct how users interact with your brand. For example, users with strong probability scores might be assigned to a journey that recognizes brand loyalty, and those with low scores to one that incentivizes conversion with promotional offers.
# Requirements for valuable predictions
At a minimum, your Iterable project should have:
- Three to six months of historical data. The amount of data required varies by industry vertical, data quality, and the audiences you're considering.
- 100,000 unique users.
NOTE
These project-level requirements are prerequisites, but Predictive Audiences also validates each audience on its own. A project can still land in Needs Review if there are an inconsistent number of qualifying users over time.
The most valuable predictions:
- Use custom events or properties that represent unique business goals, not generic conversion metrics like email opens or clicks.
- Include at least 20–30 active events that have been frequently updated in the past month.
- Use audience criteria that's neither too broad nor too narrow.
For each audience, Predictive Audiences looks at users who satisfy the full combination of conditions, not just the number of individual events or an aggregate total. Over the last 60 days, at least 1,000 users must meet your audience criteria within a 30-day lookback window on each day. If that count drops below 1,000 on any day, the audience may move to Needs Review, even when a project is large or appears to satisfy requirements.
When you define an audience:
- Optimize for a steady qualifying audience over time, not just a large one-time count.
- Treat Preview Users as an estimate—it doesn't guarantee the audience will pass validation.
- Prefer broader, outcome-oriented criteria, and use
andconditions sparingly. Stacking several required conditions can shrink the qualifying audience more than each condition suggests on its own.
Predictive audiences that apply to fewer than 1,000 users, or to almost all users in a project, don't provide a meaningful forecast. If an audience still looks broad enough but lands in Needs Review, work with Iterable to check whether project-level data readiness is affecting the audience.
# Keeping predictions current
Predictive Audiences accounts for your ever-changing data landscape by refreshing predictions weekly to reflect the most recent user data.
# Terms you should know
These terms can help you understand your predictions.
| Term | What it means |
|---|---|
| Contact list | A list you create when you select users from the Prediction chart and save them to a static or dynamic list, or export them to a CSV file or Facebook custom audience. |
| Conversion probability score | A numeric value that reflects how likely a user segment is to convert on a predictive audience in a 30-day period compared to the average of all users (reported as probability in the prediction summary, and as a value in a user profile). |
| Predictive strength | A quality gauge that tells you how effective your prediction is likely to be. See Predictive strength for possible values. |
| User field | A value you enter to represent the name of an audience in segmentation and as a field in the user profile. |
| User segment | The range, or group, of users you select from a specific predictive audience. You can export user segments for use in Segmentation and Journeys. |
# Want to learn more?
For more information about some of the topics in this article, check out this Iterable Academy course. Iterable Academy is open to everyone—you don't need to be an Iterable customer!