Master GA4 for eCommerce: Advanced Analytics Made Easy

No, and this is an important correction. GA4 reserves specific ecommerce event names and its ecommerce reports key off them. The correct name is purchase, not purchase_complete. Send purchase_complete and GA4 will accept it as a custom event that appears nowhere in ecommerce reporting, which is how stores end up with empty revenue reports despite working tags. The reserved set to use is view_item, add_to_cart, begin_checkout and purchase, plus optional steps such as add_payment_info and refund. Custom names are for genuinely custom actions only.

No. Enhanced Ecommerce was a Universal Analytics feature with its own settings toggle, and GA4 has no equivalent. Ecommerce reports populate only when your site sends the reserved events with a correctly structured items array containing item_id, item_name, price and quantity. There is no configuration screen that turns this on. If your ecommerce reports are empty, the problem is in your tagging or platform integration, and no amount of exploring the admin panel will resolve it. The quickest way to confirm where you stand is a test purchase watched in DebugView, which shows you exactly which events your site is and is not sending.

Standard GA4 properties allow 50 event scoped custom dimensions, 25 user scoped custom dimensions and 50 custom metrics, with higher limits on Analytics 360. Those caps arrive faster than teams expect, because it is tempting to register every parameter you send. Register only the parameters you will actually segment or report on, since a parameter is collected regardless and can be read in BigQuery without consuming a slot. Deleting and recreating dimensions does not recover historical data, so plan the list before you start.

No, and the eligibility bar is high. GA4 requires that within a seven day window over the previous 28 days at least 1,000 returning users triggered the relevant condition and at least 1,000 did not, with model quality sustained over time, and only purchase and in-app purchase events are supported. Many mid sized stores never qualify. Check the predictive section of the Audience Builder before planning around it, and fall back to behavioural audiences you define yourself, which work at any volume and are easier to explain.

It is a debugging tool rather than an analysis tool, and that is the right way to use it. Watching individual journeys is excellent for confirming that your events fire in the correct order, that a checkout sequence looks the way you designed it, or that a specific reported behaviour is real. It is poor for drawing conclusions, because a handful of journeys is not a sample and the human tendency to generalise from three vivid examples is strong. Use it to verify, then use aggregates to decide.

They are useful as an anomaly alarm and weak as an explanation. The system is good at noticing that a metric moved unusually, which is genuinely valuable when a tag breaks or a campaign misfires. It has no knowledge of your promotions, stock outages, price changes or seasonality, so its suggested causes are frequently wrong. Treat an insight as a prompt to investigate rather than a finding, and pair it with an alert on purchase event volume, which catches the failure that matters most.

Sooner than most teams think, for two reasons. It is the only way to retain event level history past the retention ceiling of 14 months on standard properties, and the only practical way to join GA4 behaviour against order, margin, returns and customer data from your own systems. The link is free to enable and you pay only for storage and queries. Because it captures nothing retrospectively, enabling it early costs almost nothing and enabling it late costs history you can never get back.

No. GA4 removed first click, linear, time decay and position based attribution models in November 2023. What remains is data driven attribution, which is the property default, paid and organic last click, and Google paid channels last click. Guides describing a menu of models predate that change. If your reporting genuinely depends on a different model, the route is exporting to BigQuery and building the attribution logic yourself against the raw event data, which is more work but fully under your control.

Implement User-ID, and understand what it does and does not fix. When a logged in user is assigned a stable identifier, GA4 stitches their sessions across devices and browsers into one user. Without it, identity falls back to Google signals if enabled and then to the device, so the same shopper on a phone and a laptop counts as two people. That means cross platform reporting is only as good as your login rate, which for most stores means it works for returning customers and not for first time buyers.

Only in a limited way, and the difference catches people out. When you create an audience, GA4 can populate it from up to 30 days of prior data where the conditions allow, but membership generally accrues from the point of creation onward. That makes audience creation another task worth doing early rather than when a campaign is about to launch. Build the segments you know you will need for remarketing now, so they have accumulated real membership by the time you want to use them.

Funnel Exploration with segmentation, comfortably. It requires no additional tagging beyond correct ecommerce events, it points directly at where revenue is being lost, and segmenting by device and visitor type usually turns a vague number into a specific fixable observation within an hour. Custom dimensions, predictive audiences and User Explorer are all worth having eventually, but none of them will tell you as quickly where your money is leaking as a properly segmented funnel over your core purchase journey.

Event integrity. Complete a real test purchase and watch it in DebugView, confirming that each event fires once with the reserved name and that the purchase event carries transaction_id, value, currency and a complete items array. Then raise data retention to 14 months and link BigQuery, since neither applies retroactively. Advanced analysis on unreliable event data produces confident, well presented conclusions that happen to be wrong, which is considerably worse than having no analysis at all. Repeat that check after any deployment touching the cart or checkout, because those are precisely the templates where tracking breaks without anyone noticing for weeks.