Skip to content

Blog · Analytics

GA4 predictive metrics for ecommerce: what they forecast and when you can use them

Google Analytics 4 can estimate which customers are likely to buy, lapse or spend in the near future. These predictions are useful—but only once your store sends clean purchase data and has enough returning customers for Google’s models to qualify.

StoreBuild editorial team · 8 min read

Published

Recommendations here are our editorial judgement from the documented facts — not rankings, test results or scores. We haven’t tested these platforms hands-on.

The idea

What GA4’s predictive metrics measure

Google documents three predictive metrics for websites: purchase probability (the probability that a user active in the last 28 days will log a purchase key event within the next 7 days), churn probability (the probability that a user active within the last 7 days will not be active within the next 7 days) and predicted revenue (the revenue expected from purchase key events within the next 28 days from a user active in the last 28 days).

These are predictions about individual users, built by Google’s machine-learning models from your event data. They are not a sales forecast for your whole business, and they do not tell you why someone will buy or leave.

This article is editorial guidance based on the official documentation listed below. We haven’t tested these features hands-on, and we don’t quote industry statistics we can’t trace to a reliable source.

Before you plan around them

Eligibility: why many small stores won’t see them yet

Google’s documentation sets minimum thresholds. In the last 28 days, over a seven-day period, at least 1,000 returning users must have triggered the relevant condition (purchase or churn) and at least 1,000 returning users must not have. Model quality must also be sustained over time. Check the linked page for the current requirements—Google can change them.

Purchase probability and predicted revenue also require the purchase event (Google now recommends purchase over the older ecommerce_purchase) to be sent with its value and currency. If your store doesn’t send a correct purchase event, the models have nothing to learn from.

Many independent stores have fewer returning buyers than these thresholds. That isn’t a failure—it means the simpler methods later in this article will serve you better for now.

Six steps to make your data prediction-ready

  1. 1. Connect GA4 the way your platform documents

    Use your platform’s documented connection rather than pasting tags by hand where possible. Mixing two methods can double-count purchases.

  2. 2. Confirm the recommended ecommerce events

    Google’s ecommerce guide lists events such as view_item, add_to_cart, begin_checkout and purchase. Place a test order and check in GA4’s real-time or DebugView reports that purchase arrives once, with value, currency and items.

  3. 3. Mark purchase as a key event

    Predictive metrics refer to purchase key events. Check that purchase is marked as a key event in your property settings.

  4. 4. Respect consent rules

    Where consent is required, users who decline analytics may not be measured, and your numbers will be lower than your store’s order count. Don’t bypass a visitor’s choice to fill the gap.

  5. 5. Wait, then check eligibility

    Eligibility is assessed over time. Check whether the metrics appear in your audience builder and Explorations rather than assuming they will by a particular date.

  6. 6. Compare predictions with what actually happened

    Before acting on a predictive audience, compare a past prediction period with the orders you actually received. Treat the metric as one signal, not the truth.

Based on official documentation

How each platform documents sending data to GA4

Summarised from official documentation. The exact events sent depend on the app, plugin or settings you use—verify with a test order.

  • Built in
  • Needs an eligible paid plan
  • Via an app (may have its own fees)
  • Via a WordPress plugin or extension
  • Depends on your host
  • Your responsibility
Connecting GA4
ShopifyThrough Shopify’s Google & YouTube app, as documented in the Help Center (Via an app (may have its own fees))
WixConnect a Google Analytics property through Wix’s marketing integrations (Built in)
WooCommerceGoogle Analytics for WooCommerce extension (Via a WordPress plugin or extension)
Ecommerce events (purchase, add to cart and similar)
ShopifySent by the connection—check which events in your setup (Via an app (may have its own fees))
WixWix documents ecommerce event tracking with a GA property (Built in)
WooCommerceExtension documents ecommerce event tracking (Via a WordPress plugin or extension)
Using predictive metrics
ShopifyInside GA4, once your property is eligible (Your responsibility)
WixInside GA4, once your property is eligible (Your responsibility)
WooCommerceInside GA4, once your property is eligible (Your responsibility)

Practical uses once you qualify

Google documents predictive audiences such as likely 7-day purchasers, likely 7-day churning users and predicted 28-day top spenders. You can use them for analysis in GA4 or, where you have linked Google Ads and your consent setup allows, for advertising.

Reasonable uses include checking which products likely purchasers are viewing, sending a helpful (not pushy) reminder to opted-in customers who appear likely to lapse, and avoiding spending on ads aimed at people who would probably buy anyway. Measure each change against a comparable period or a control group.

Avoid presenting predicted revenue as income you have earned or will earn. It is a probabilistic estimate for planning conversations, not a basis for loans, payroll or stock commitments on its own.

If you don’t qualify

Forecasting without predictive metrics

A weekly spreadsheet of orders, revenue and returning-customer share often answers the same questions for a smaller store. Compare this week with the same week last year where you have history, and note promotions, stockouts and holidays so you don’t mistake them for trends.

Your store admin’s own reports remain the source of truth for orders and refunds; GA4 shows behaviour on the site. When the two disagree, investigate the tracking before drawing conclusions. Our inventory forecasting and cash-flow articles cover how to turn these numbers into stock and spending decisions.

Before you rely on a GA4 prediction

Tick these off before you commit.

  • A test order produces exactly one purchase event with value, currency and items.
  • Purchase is marked as a key event in your GA4 property.
  • Only one method sends ecommerce data, so purchases aren’t double-counted.
  • Consent choices are respected, and you understand the resulting gaps.
  • The metric actually appears for your property—you haven’t assumed eligibility.
  • You’ve compared past predictions with real orders before acting on them.

Common questions

Keep exploring

Official sources

Platform facts on this page reference the documentation below. Features, plans and availability change — check the provider before you decide.