Google Analytics 4 ships with machine learning that most accounts never switch on: predictive metrics that estimate which visitors will buy or churn, predictive audiences you can export straight to Google Ads, automated insights that flag anomalies before you notice them, and a search bar that answers plain-English questions. Here is what each feature really does in 2026, the data thresholds that gate them, and how to put the predictions to work.
The short version
- GA4 predicts purchase probability, churn probability, and revenue.
- Predictions need 1,000 converters and 1,000 non-converters in 28 days.
- Predictive audiences export to Google Ads for targeting and exclusions.
- Everything here is in the free tier, not just GA4 360.
- 3 predictive metrics built into GA4
- 1,000 purchasers needed to train the model
- 7 days prediction window for purchase and churn
- $0 extra cost over standard GA4
The three predictive metrics
Purchase probability estimates the chance that a user active in the last 28 days buys within the next 7. Churn probability estimates the chance that a user active in the last 7 days does not return in the next 7. Predicted revenue estimates the purchase revenue expected from a user over the next 28 days. The models train on your event data and pattern knowledge from Google's wider network, which is the same machine learning family behind Smart Bidding. One vocabulary note for 2026: what GA4 used to call conversions are now key events, renamed in March 2024, so documentation and menus talk about key events wherever older articles said conversions.
The thresholds are strict. Within a 28-day window you need at least 1,000 returning users who triggered the purchase event and 1,000 who did not, and the model has to keep meeting quality checks or the metrics switch back off. There is no override. Sites below the purchase volume can anchor predictions to a higher-volume key event such as add-to-cart or lead submission, which is the standard workaround for lead-gen and B2B properties.
Predictive audiences turn forecasts into targeting
Once predictive metrics are live, GA4 offers ready-made audience templates: likely 7-day purchasers, likely 7-day churning users, and predicted top spenders, plus custom combinations with thresholds you set. Link Google Ads and the audiences sync automatically. The practical plays are the point here. Retarget only high-probability buyers instead of every visitor, and acquisition cost drops because you stop paying to chase people who were never going to convert. Use low-probability users as exclusions. Trigger win-back email flows the moment users enter the churn audience, since keeping a customer is cheaper than reacquiring one. And give predicted top spenders the VIP treatment: early access, premium offers, priority support.
Insights that find problems for you
GA4's automated insights continuously scan your data and surface anomalies, trends, and shifts on the home screen: a 34% organic traffic jump, a revenue dip that breaks the usual Tuesday pattern, a channel quietly converting three times better than the rest. The underrated move is custom insights: define your own conditions, like conversion rate falling 20% week over week, and have GA4 email you when they trigger. That configuration turns the report you forget to check into an early warning system. The search bar runs the same intelligence interactively: type "conversion rate last month" or "compare mobile vs desktop" and GA4 answers directly. Google has also been folding Gemini into Analytics, so expect the insight summaries and data questions to keep getting more conversational.
Behavioral modeling fills the consent gap
When visitors decline cookies, GA4 does not just show a hole in the data. With Consent Mode implemented, it models the behavior of non-consenting users from the patterns of similar users who did consent, which restores a privacy-compliant estimate of traffic and key events you would otherwise lose. For sites with meaningful European traffic this single feature changes reported numbers enough to alter decisions, and it is a setup step, not a purchase.
How to switch the AI on
-
Verify the purchase event
Predictions train on the purchase key event with transaction_id, value, and currency parameters. Broken ecommerce tracking means no models, whatever your traffic.
-
Turn on Google signals and link Google Ads
Signals improves cross-device data; the Ads link is what lets predictive audiences flow into campaign targeting.
-
Implement Consent Mode
Required for behavioral modeling, and increasingly the default expectation for compliant measurement.
-
Check whether predictions are live
Try creating a new audience under Admin. If the predictive templates appear, your property cleared the thresholds; if not, the gap is data volume, not configuration.
-
Create custom insights for your critical metrics
Five minutes of setup gets you alerts on the numbers that would otherwise fail silently for weeks.
What to do with the predictions
The features only pay off when they change decisions. Shift ad budget toward high-probability audiences and away from low ones. Intervene on churn before it happens instead of discounting after. Use insight anomalies as the trigger for investigating traffic problems, alongside the engagement diagnostics covered in our post on reducing bounce rate. And when the blocker is setup rather than strategy, Egochi configures tracking, activates the AI features, and builds the reporting as part of its Google Analytics service. The machine learning is only as good as the event data feeding it, and most properties we open have a tracking problem before they have an analysis problem.
Questions people ask about GA4 AI features
Are GA4 AI features free?
Yes. Predictive metrics, predictive audiences, generated insights, and natural language search all ship in the free standard GA4 property. The paid 360 tier adds data limits and SLAs, not different AI. The only real gate is meeting the data thresholds.
Why are my predictive metrics not working?
Almost always insufficient data. You need at least 1,000 returning users who triggered the purchase event and 1,000 who did not, within 28 days, and the model quality has to hold over time. If you clear the thresholds, allow a week or two for training.
How accurate are GA4 predictions?
Google does not publish accuracy figures. The models share infrastructure with Google Ads Smart Bidding and improve as data accumulates. Use predictions directionally, high probability versus low, rather than treating any percentage as precise.
Do GA4 AI features work for B2B websites?
Yes, with adjustments. B2B purchase volumes rarely clear the purchase thresholds, so anchor predictions to a higher-volume key event like lead form submission instead. Churn probability works well for spotting prospects going cold.
Do I need GA4 360 for predictive audiences?
No. Predictive audiences are available in free GA4 once your property meets the data thresholds. Link Google Ads under Admin, then Product Links, and the audiences sync automatically for campaign targeting.
