GA4 Conversion Tracking Checklist (2026)
GA4 conversion tracking gets a bad reputation from marketers who migrated from Universal Analytics and found the numbers ‘didn’t match. ‘ Most of the time, the problem isn’t GA4 — it’s an incomplete GA4 conversion tracking setup. Here’s the checklist I run through on any account before I trust a single conversion number in it.
1. GA4 Conversion Tracking: Separate Key Events from Every Click
This is the first place most GA4 conversion tracking setups go wrong. GA4 will happily track every scroll, click, and video play as an event. That’s useful for behavioral analysis — but it becomes a problem the moment you mark too many events as key events (GA4’s term for what UA called “goals”). If “newsletter signup,” “PDF download,” and “actual purchase” are all key events with equal visual weight in your reports, your conversion rate stops meaning anything specific.
Fix: Only mark true business outcomes as key events — purchases, qualified leads, bookings, demo requests. Track the smaller engagement events as regular events, and analyze them separately as micro-conversions, not blended into your headline conversion rate.
2. Check Your Data Streams Aren’t Double-Counting
If you have a website and an app under the same property, or multiple data streams pointing at overlapping traffic (a common leftover from a rushed setup), you can end up double-counting users and events. Go to Admin → Data Streams and confirm each stream maps to exactly one distinct property (a single website or a single app) and that you’re not running both gtag.js and Google Tag Manager on the same pages, firing the same events twice.
3. Cross-Domain Tracking, If You Use a Separate Checkout or Booking Platform
If your main site is on one domain and checkout runs on a different domain (Shopify checkout, a booking platform subdomain that isn’t configured as a subdomain, or a separate landing page tool), GA4 will treat that handoff as a new session unless cross-domain measurement is explicitly configured under Admin → Data Streams → Configure tag settings → Configure your domains.
This is the single most common cause of “my paid traffic shows zero conversions, but my payment processor shows sales”—the session breaks at the domain handoff, and the conversion gets attributed to (or lost in) a new, unattributed session.
4. Google Ads Linking: Import, Don’t Duplicate
Link Google Ads to GA4 under Admin → Product Links → Google Ads Links, then import your GA4 key events as conversion actions inside Google Ads—rather than building a second, separate conversion tracking setup with the standalone Google Ads tag. Running both creates two sources of truth that will never quite agree and doubles your maintenance burden every time you add a new conversion.
Once linked, check that attribution settings in GA4 (data-driven by default) match what you expect Google Ads to be optimizing toward—this affects which touchpoints get credit and can materially change which campaigns look like they’re “working.” This is where GA4 conversion tracking and Google Ads bidding actually connect.” Once your Google Ads are structured correctly, GA4 needs to track them accurately too.
5. Test Every Key Event with the DebugView Before You Trust It
Before reporting on a new key event, open Admin → DebugView, trigger the event yourself on the live site (with the GA Debugger extension or a debug parameter), and confirm:
- The event fires exactly once per action (not on page load, not twice on a slow double-click)
- The event carries the parameters you expect (value, currency, transaction ID for purchases)
- It fires on both desktop and mobile, and after any cookie consent banner interaction if you’re using consent mode
Skipping this step is how “GA4 doesn’t work” claims get started — usually it’s a mis-fired tag, not a platform problem.
6. Set a Data Retention Window You Can Actually Use
GA4’s default retention for user-level and event-level data (used in free-form Explorations) is often set to 2 months out of the box. For anything longer than that, GA4 falls back to aggregated reporting, which is fine for standard reports but limits ad-hoc exploration further back. Under Admin → Data Settings → Data Retention, extend this to 14 months if you want to run year-over-year cohort or exploration analysis later — this only affects Explorations, not your standard reports, but it’s an easy setting to forget until you need it.
The takeaway: GA4’s numbers aren’t inherently unreliable — they’re only as reliable as the setup underneath them. Most “GA4 doesn’t match reality” complaints trace back to one of these six things left unchecked, not a flaw in the platform itself. GA4 conversion tracking isn’t inherently unreliable — it’s only as reliable as the setup underneath it.