Shopify analytics vs. GA4 - what to trust, and when

Every ecommerce team eventually has the meeting: Shopify says one revenue number, GA4 says another, and someone asks which tool is broken. Usually, neither. They measure different things, differently - and knowing which is authoritative for what ends the argument.

Why the numbers differ (legitimately)

Shopify counts orders at the source - every order that hits the system, regardless of the customer's browser, consent choices, or ad blockers. GA4 counts what its tracking observed - which means it misses users who declined cookies, blocked scripts, or bought in ways the tag never saw, and its attribution models slice sessions differently. Add differing timezone cutoffs, refund handling, and session definitions, and a 5–15% gap between the two is normal life, not a bug. (In consent-strict Europe, often toward the larger end.)

What each tool is authoritative for

Shopify is truth for money and inventory: revenue, order counts, AOV, units sold, conversion rate as Shopify defines it. When finance asks what you sold, the answer comes from Shopify, full stop.

GA4 is truth for behavior and acquisition: where sessions came from, what people did on site, how channels compare, funnel drop-offs, multi-touch context. When marketing asks which channel is working, the answer comes from GA4 - read as relative signal, not accounting.

The failure mode is crossing the streams: reporting revenue from GA4 (it undercounts) or judging channel performance from Shopify's last-click attribution alone (it flattens the journey).

Where Shopify's own analytics shine

Shopify's reports get underrated. Sales by product and variant, inventory reports, returning-customer rate, and - genuinely useful and often ignored - the native site-search report showing what customers type into your store's search box. That's taxonomy input (Phase 2) straight from the source. For a quick health check on what's selling, Shopify's reports are faster and truer than anything you'll build in GA4.

The practical peace treaty

Write down, once, in a shared doc: which numbers we read from which tool, and the accepted gap between them. Check the gap quarterly - a stable gap is fine; a suddenly widening one means tracking broke, and that's a real signal worth investigating. Then stop reconciling weekly. The energy is better spent on the numbers themselves.