Every advertiser has had this meeting. Meta Ads reports 40 purchases. Google Ads reports 31. GA4 reports 22. The finance team asks which number is true, the agency blames the platform, and someone suggests pausing the campaign “until tracking is fixed”. Often, nothing is broken — the three systems are simply counting different things, by different rules, over different windows.
That does not mean every gap is innocent. Sometimes a tag really is misfiring, a purchase really is counted twice, or a consent banner really is blocking half your data. The skill is telling the two apart quickly, before you cut a profitable campaign or scale a broken one. That skill is the GA4 conversion discrepancy diagnostic.
The direct answer: why GA4 and your ad platform disagree
A GA4 conversion discrepancy happens because ad platforms and GA4 count conversions differently: platforms credit their own ad interactions inside their own attribution windows, while GA4 counts key events from site sessions under its own GA4 attribution model and lookback window. A gap is normal; a gap that suddenly jumps, moves in one direction only, or appears in one channel is the signal of a real fault.
The six reasons the numbers differ
1. Different attribution models
Meta may credit a purchase to an ad clicked seven days ago, while GA4’s data-driven model spreads credit across several touches — search, social, email — and may show fractional credit rather than one whole conversion per channel. Neither is lying; they are answering different questions. Google’s own documentation explains how to change the reporting attribution model for key events and what each model does, which is the first place to check before assuming a fault.
2. Different windows
Meta’s default click window, Google Ads’ conversion window and GA4’s 90-day key-event lookback window rarely match. A customer who clicks an ad, returns two weeks later via email and buys will be counted differently by each system — and on different days. Comparing a 7-day window in one tool with a 30-day window in another guarantees a gap.
3. Different definitions of a conversion
Is a “purchase” the thank-you page loading, the payment clearing, or the order appearing in your shop system? If Meta counts a browser purchase event, GA4 counts a key event on a different trigger, and your store counts only paid orders, three honest numbers will disagree. Refunds, failed payments and test orders widen the gap further.
4. Consent, blockers and browsers
GA4 depends on a browser tag firing. Ad blockers, Apple’s privacy features, rejected cookie consent and a slow thank-you page all stop that tag. Server-side events — Meta’s Conversions API, for example — often survive where the browser tag does not, which is why Meta can report sales GA4 never sees. This gap grows over time as privacy controls tighten, and it affects some audiences (iPhone users, privacy-conscious European shoppers in Germany, France or the Netherlands) more than others.
5. Modelled conversions
Where GA4 cannot observe a key event directly, it may model it — estimating conversions from observed patterns. Google’s modelled key events documentation notes that attributed conversion data can keep updating for up to 12 days after the event. Comparing yesterday’s GA4 figure with yesterday’s ad-platform figure therefore compares a still-moving number with a settled one.
6. Genuine tracking faults
Finally, the real faults: a purchase event firing twice (browser plus server, without deduplication), a tag firing on the wrong page, a thank-you page that loads before payment completes, or a GA4 key event marked on an event that also fires for non-buyers. These faults produce gaps that are large, sudden or channel-specific — and they are fixable. OWOX’s comparison of Facebook Ads and GA4 reporting walks through the same distinction between expected differences and real measurement problems.
GA4 conversion discrepancy: the 6-point diagnostic
Work through these in order, before changing any budget:
| Check | What to look for | What it tells you |
|---|---|---|
| 1. Same dates, same window | Are both reports using identical date ranges and attribution windows? | A window mismatch explains most small, steady gaps. |
| 2. Same definition | Does each tool count the same event, on the same trigger? | Definition drift is the commonest hidden cause. |
| 3. One source of truth for revenue | How many paid orders does your shop or CRM actually show? | The order system, not any dashboard, is the anchor number. |
| 4. Tag firing check | Does the purchase event fire once, on the real thank-you page, in a live test order? | Double or missing fires are real faults — fix first. |
| 5. Browser vs server | Are browser and server events deduplicated by the same event name and ID? | Undeduplicated events inflate the platform that receives both. |
| 6. Trend, not snapshot | Is the gap stable over weeks, or did it jump on a specific date? | A stable gap is usually attribution; a jump points to a change — a site update, a consent change, a tag change. |
What to do with each result
- Stable, moderate gap, same trend direction: pick one governing number for budget decisions — usually paid orders from your own system — and use the other tools as cross-checks. Stop chasing an exact match; two different models will not produce one.
- Gap jumped after a site or tag change: roll back or inspect that change first. A new checkout, a new consent banner or a theme update is the likeliest culprit.
- One channel’s gap is far larger than others: inspect that channel’s tagging specifically. A Meta gap that dwarfs a Google gap, for example, points to pixel/CAPI setup, not to GA4 generally.
- Platform reports far more than orders exist: check deduplication and test orders immediately. Counting the same sale twice makes every ROAS figure — and every scaling decision — wrong.
Examples help here because the pattern travels. An ecommerce store in London, UK, selling at £60 average order value, a clinic in Sydney, Australia, booking consultations at $150, and a retailer in Milan, Italy, selling at €80 will all show different absolute gaps — different channels, different privacy settings, different devices — but the diagnostic is identical: align dates, align definitions, anchor to real orders, test the tag, check deduplication, read the trend.
GA4 vs Meta Ads reporting: a practical rule
Use GA4 vs Meta Ads reporting differences as a health check, not a verdict. Watch the ratio week to week. If Meta reports roughly 1.3 times GA4 every week, that ratio is your baseline — monitor it, and investigate only when it moves sharply. If it jumps from 1.3 to 2.1 in a week, something changed, and the six checks above will usually find it within an hour.
Frequently asked questions
Should GA4 and Meta Ads ever match exactly? No. Different models, windows, definitions and privacy losses make an exact match essentially impossible, even with perfect tagging. Aim for a stable, explainable gap and a trustworthy order count, not identical dashboards.
Which number should I use for budget decisions? Your own order and revenue system, reconciled weekly against each platform. Use platform numbers to judge relative performance between campaigns inside that platform, and GA4 to understand cross-channel journeys — never compare a campaign’s Meta number directly with another campaign’s GA4 number.
How long should I wait before comparing? Avoid the most recent week for GA4 channel attribution, since modelled and attributed figures can update for days afterwards. Compare settled weeks, and re-check a recent gap after it has had time to stabilise.
Does server-side tracking fix the discrepancy? It narrows it — server events survive browser blocking — but it does not remove attribution differences. Server-side tracking plus proper deduplication gives you more complete data; it does not make two different models agree.
Stop arguing about which dashboard is “right” and start diagnosing. Align the dates, align the definitions, anchor to real orders, and let the trend tell you whether you have an attribution difference or a fault worth fixing. Start with the full pixel, CAPI and GA4 tracking stack, check your deduplication if sales look double-counted, explore AgencyAxiom’s tracking services, and book a free consultation for a tracking audit of your own numbers.