Google Analytics Discrepancies

Overview

Discrepancies between Convert and Google Analytics (GA4) visitor counts are common and usually expected. They stem from fundamental differences in how and when each tool counts visitors.

Convert Counts More Than GA

This is the most common scenario. Convert's tracking script loads synchronously in the <head>, counting visitors almost immediately. GA typically loads later (async, via GTM, etc.).

Why the gap exists:

Short-session visitors (bounces) may leave before GA fires
If a visitor closes the tab in the first few seconds, Convert has already counted them but GA has not
Mobile users on slow connections may abandon before GA loads
Browser "back" button usage during this window is counted by Convert but missed by GA
Ad blockers may slow down or block GA while Convert still fires

This is particularly impactful for:

Mobile users (slower connections)
Users with ad blockers
Quick-bounce visitors
Single-page views where users immediately navigate away

GA Consent Mode and `pscdl=denied`

When the project setting "Do not wait for page_view event" is enabled and GA consent is not yet granted, Convert sends experience_impression with pscdl=denied. GA receives the event but filters it out of reports because analytics storage is denied.

Solution: Enable the default setting to wait for the first GA page_view. This ensures impressions are reported only after consent is granted and data flows properly.

Note: On sites using CMPs (Consent Management Platforms), the first GA page_view can still occur before consent is granted. In such cases, the resulting experience_impression may carry pscdl=denied — this is expected CMP timing behavior.

GA Counts More Than Convert

This is rare but can happen:

GA session timeout — If configured to be very short (e.g., 5 minutes), the same visitor returning after timeout is counted as new by GA, while Convert maintains the true unique count via its cookie
Network failures — Temporary issues affecting Convert's tracking but not GA (rare since GA is already loaded by the time Convert sends)

Reducing the Gap

Option	Effect	Trade-off
Move GA/GTM bootstrap inline near top of `<head>` (keep `async`)	Shrinks or eliminates the gap	Slightly more bytes in critical path
Add `<link rel="preconnect">` to `googletagmanager.com`	Cuts DNS/TLS time on mobile	Modest gain — measure first
Fire GA in an `Initialization – All Pages` trigger in GTM	Runs as soon as GTM boots	Requires GTM container change
Accept the gap, treat Convert as the early-exit indicator	No code change	Must document rationale for stakeholders
Align GA session timeout to 20 min	Removes edge case where GA keeps sessions alive longer	Alters GA Avg. Session Duration

Key Takeaway

The discrepancy indicates Convert's tracking is working correctly — it captures all visitors including ultra-short sessions. GA's delayed tracking misses some visitors due to its loading and consent constraints. This gap can be a valuable early-exit signal in your analytics.