Guide 89 · Admin · GA4 + Tag Manager
Predictive audiences in GA4: use the suggested templates only when the property qualifies
A predictive audience has at least one condition based on a predictive metric, such as likely 7-day purchasers. Use the suggested template when it reads Ready to use.
After the trace, the Predictive tab shows which suggested audiences are Ready to use, and the one you picked matches a predictive metric your property actually generates.
Download the reference: the predictive audiences reference (JSON) (541 bytes, MD5 748611622c047ec2fd70eeeb8cd48b14). It holds the worked example from this guide with every figure recomputed. The file was built for this guide and checked the same day; it was not exported from a live property.
What you need
- A property eligible for predictive metrics, with enough purchase history.
- Marketer access or above at property level.
- The list of suggested predictive audiences, open in the builder.
The trace
- Open Admin, then under Data display click Audiences, then New audience. Under Suggested audiences, click Predictive. The templates listed there are the predictive audiences Google suggests for your property.
- Read the readiness label. Suggested predictive audiences that meet the prediction-modelling prerequisites are labelled Ready to use. A template without that label is not ready, and building from it produces an audience that never fills.
- Pick one template to start with. Likely 7-day purchasers includes users who are likely to make a purchase in the next 7 days. Likely 7-day churning purchasers includes purchasing users likely not to visit in the next 7 days. Predicted 28-day top spenders is the third common template.
- Open the template in the audience builder and read its condition. For likely 7-day churning purchasers, the condition includes a purchase event, a lifetime value above 0, and a churn probability above the 80th percentile. The percentile is part of the definition, not a number you invent.
- Check the worked example in the download. Out of 3,120 purchasers, the top 20% by churn probability is 624 users. The 80th percentile cut keeps 624 of 3,120, which is 20.0%. If your audience holds far more than a fifth of purchasers, the condition was changed.
- Save the audience, then watch whether it accumulates users. If the property becomes ineligible for the predictive metric, exported predictive audiences stop accumulating new users and no new predictions are generated. That is the signal to check eligibility again, not to rebuild the audience.
Dead ends
DEAD END — BUILDING A PREDICTIVE AUDIENCE ON AN INELIGIBLE PROPERTY. Predictive audiences depend on the underlying predictive metrics being eligible. Without eligibility there are no predictions to base a condition on.
DEAD END — EDITING THE PERCENTILE WITHOUT A REASON. The 80th percentile in the churning template is the definition of likely to churn. Changing it to the 50th percentile doubles the audience and removes its meaning.
DEAD END — EXPORTING AND NEVER CHECKING. An exported predictive audience that stops growing is telling you the property lost eligibility. Check the metric, not the export.
End state
The worked example keeps 624 of 3,120 purchasers above the 80th percentile, exactly 20.0%. The template was Ready to use, its condition was read before saving, and eligibility is the thing to watch.
Sources
- Analytics Help — Predictive audiences — predictive audience definition, Ready to use label, the 80th percentile churn condition, and the eligibility warning
Checked 11 November 2026. The click paths above are the ones in Google's help and developer pages, opened 11 November 2026. No step in this guide was run inside a live Analytics property or a live Tag Manager container, because both need a Google sign-in and this site does not sign in as anyone. The figures on this page were recomputed in Python 3 and are shown in the downloadable file, which was checked the same day. Google moves menu items without warning; if a path on this page no longer matches, the linked help page is the current map.