Why GA4 is a Game Changer for eCommerce Analytics

No. Standard Universal Analytics properties stopped processing data on 1 July 2023 and Google began deleting Universal Analytics data from 1 July 2024. There is no historical property left to compare against or export from, so the migration framing in articles from this period no longer applies. If your organisation did not export that history before deletion, it is permanently gone and your GA4 baseline starts from whenever the property was created. Plan year on year comparisons accordingly.

Three things stand out. The event based model tracks what people actually do rather than forcing everything into page views and sessions, which suits stores where the meaningful actions are product views and cart additions. Cross platform reporting combines web and app in one property, which the old model could not do. And the free BigQuery export, previously a paid Analytics 360 feature, gives every business raw event level data it owns. That last one is the most undervalued change of the three.

Worth stating plainly, because most coverage skips it. Data retention defaults to two months and tops out at 14 on standard properties, where Universal Analytics kept data far longer. Attribution options narrowed sharply when first click, linear, time decay and position based models were removed in November 2023. Reporting thresholds hide small segments. And the interface is harder to learn, with Explorations requiring real analytical thinking where the old reports gave immediate answers. It is a more capable tool that is less immediately usable.

No, and the phrase should be retired. Enhanced Ecommerce was a Universal Analytics feature with a settings toggle. GA4 has no equivalent switch, and its ecommerce reports populate only when your site sends the reserved events with a correctly structured items array, principally view_item, add_to_cart, begin_checkout and purchase. Guides describing an Enhanced Ecommerce setting in GA4 are carrying over old terminology, and following them leads teams to hunt for a screen that does not exist while the real tagging work remains undone.

Not in GA4. Google renamed conversions to key events in March 2024 across the admin toggles, report columns and API fields, reserving the word conversion for the Google Ads side. The purpose was to stop teams expecting GA4 and Ads numbers to match, since the two never counted identically. The mechanics are unchanged: you mark the events that matter and they appear in key event reporting. Any guide instructing you to mark a conversion in GA4 predates the change.

It moves the design work forward. In the old session model, reporting was largely a matter of reading what was collected automatically. In GA4 you decide what constitutes a meaningful interaction, name it, send it with the parameters you will want later, and only then can you report on it. That means a badly planned implementation cannot be fixed retrospectively, because uncollected events do not exist. Spend time on the measurement plan before the tagging, not after the reports disappoint.

It works well when users log in and User-ID is implemented, and much less well when they do not. Identity resolves through User-ID first, then Google signals if enabled, then the device identifier. Without login, the same shopper on a phone and a laptop is two users. So cross platform reporting tends to be accurate for returning, logged in customers and unreliable for first time buyers, which is the opposite of what most acquisition analysis needs. Judge the feature by your actual login rate.

They measure different things and always will. GA4 records what the browser successfully sent, so ad blockers, declined consent, network failures and shoppers leaving before the purchase event fires all cause undercounting, while a refreshed confirmation page can cause double counting. Your platform records what was charged, including offline orders GA4 never sees. Send transaction_id so GA4 deduplicates, then stop trying to reconcile totals. Use the platform for financial truth and GA4 for behaviour and channel analysis. Agreeing that split explicitly with your finance team early prevents the recurring argument about which report is correct when they inevitably diverge.

It is privacy-adapted, which is a fairer description. IP addresses are not stored, data retention is capped, and reporting thresholds suppress rows that could identify individuals. But GA4 is still a Google product processing visitor data, and in the EEA and UK you need Consent Mode v2 signals, required since March 2024 for advertisers using Google audience features, plus a compliant consent banner. Some organisations still conclude a self hosted or EU based analytics tool suits their obligations better. Treat it as a legal question, not a technical one.

Selectively. Anomaly detection is genuinely useful for catching a broken tag or a misfiring campaign. Predictive metrics are strong when they work but require at least 1,000 returning users who triggered the relevant condition and 1,000 who did not within a seven day window over the previous 28 days, so many stores never qualify. Modelled conversions fill gaps left by consent refusal but are estimates. Useful features, oversold in most coverage, and none of them a substitute for correct event tracking.

Four reserved ecommerce events firing correctly and verified through a real test purchase in DebugView. Data retention raised to 14 months and BigQuery export linked, both done early since neither is retroactive. Internal traffic filtered and actually switched from testing to active. One or two genuine key events rather than eight. One segmented funnel exploration reviewed weekly, plus an alert on purchase volume. That configuration takes an afternoon and outperforms most heavily customised properties we audit. Document the choices somewhere your team can find them, because the person who set the property up is rarely the person debugging it a year later.

For behaviour and acquisition, largely yes. For financial reporting, no, and it was never intended to be. Use your commerce platform for revenue, margin, refunds and inventory, GA4 for how people arrive and where they abandon, Search Console for whether people are finding you at all, and BigQuery when you need history beyond the retention ceiling or joins against order data. The common failure is asking GA4 finance questions, getting answers that do not reconcile, and then distrusting the tool for the things it does well.