GA4 vs Facebook Ads conversions never match because they answer different questions. Meta credits conversions inside its own attribution window after a click or a view, including modelled ones. GA4 credits only conversions it sees in the browser, split across every channel by one model. TikTok, Snapchat and Google Ads each add their own rules. Reconcile them in a table beside backend orders instead of forcing them to agree.
- Meta, TikTok and Snapchat can count view-through conversions. GA4 never sees them.
- Google Ads dates a conversion to the click; GA4 dates it to when it happened.
- Backend orders are the referee for totals. GA4 is the referee for channel mix.
- Track each platform-to-GA4 ratio weekly; investigate when it moves, not because it isn’t 1.
Every agency has had the call: Meta shows one purchase number, GA4 a much smaller one, and the shop a third. "Which one is lying?" None of them. They are three different measurements, and your job is to know why they differ and how much difference is normal for this account.
Why the numbers differ
The causes, side by side
| Ad platform | GA4 | Effect on the gap | |
|---|---|---|---|
| Who gets credit | Each platform credits itself for conversions inside its window | One model splits credit across all channels | Platform totals added together often exceed real orders |
| View-through | Can count conversions after an ad view with no click | Can’t see views at all | Widest for video and awareness buys |
| Window | Set per ad set or conversion action | Lookback of 30, 60 or 90 days for key events | Long consideration cycles widen it |
| Date | Google Ads uses the click date | GA4 uses the conversion date | Days don’t line up; weeks get closer |
| Consent and blockers | Server events can still reach the platform | No tag, no event, unless modelling applies | Widens where consent rates are low |
| Cross-device | Logged-in users matched across devices | Cookie-based unless User-ID or signals help | Mobile click, desktop purchase goes missing in GA4 |
| Tagging | Counts the conversion regardless of UTMs | Broken UTMs move it to Direct or Unassigned | Platform looks fine, GA4 channel looks empty |
| Event source | Pixel and server events, de-duplicated by the platform | The GA4 tag’s own purchase event | Either side can double or drop independently |
| App vs web | Includes app and SKAN conversions | A web property sees web only | App-heavy accounts look broken in GA4 |
GA4’s model is set in Admin > Data display > Attribution settings: data-driven, paid and organic last click, or Google paid channels last click. Changing the model rewrites past reports as well as future ones, and changing the lookback window only applies going forward, so note both before you compare months.
Write down each platform’s rules first
For every platform in the report, note
- The attribution window and whether view-through is on (Meta shows it in the Attribution setting column)
- The exact conversion event used for optimisation and reporting
- Whether the column includes app or offline conversions
- The account time zone and currency
- For Google Ads: whether you read Conversions (primary actions only) or All conversions
- For GA4: whether you read session last click (Traffic acquisition) or the attribution model (Advertising)
The platforms change their defaults from time to time, so read the setting from the account rather than from memory, and record it in the report notes.
Build the reconciliation table
This is the table we put in front of clients. Fill it weekly, per platform, from the same date range and time zone.
Weekly reconciliation template
| Platform reported | GA4 (session last click) | Ratio platform ÷ GA4 | Normal range | Note | |
|---|---|---|---|---|---|
| Meta | Purchases, attribution setting noted | Purchases where session source is facebook or instagram | Calculated | From your trailing weeks | Changes, outages, offers |
| TikTok | Purchases, window noted | Purchases where session source is tiktok | Calculated | From your trailing weeks | |
| Snapchat | Purchases, swipe and view noted | Purchases where session source is snapchat | Calculated | From your trailing weeks | |
| Google Ads | Conversions (primary), click date | Purchases where session source / medium is google / cpc | Calculated | From your trailing weeks | |
| All paid | Do not add up | Sum of the rows above | n/a | n/a | |
| Backend | Orders and revenue | All GA4 purchases | Backend ÷ GA4 | From your trailing weeks | Tracking health |
- Fix the plumbing first: UTMs that land in the right channel (the Unassigned fix), no duplicate purchases (duplicate purchases) and no test traffic.
- Use the same week, time zone and currency for every column. Compare weeks, not days.
- Calculate the ratio for each platform. As a starting point, collect six to eight weeks before you call any range "normal".
- When a ratio jumps outside its range, check tracking before performance: a changed URL template, a tag release or a consent change usually explains it.
- Report the backend number as the total, GA4 for channel mix, and each platform’s number as what that platform optimises toward.
What to tell the client
Keep it to three sentences. "The shop’s orders are the real total. GA4 shows how those orders split across channels using one rule for everyone. Each ad platform shows what it believes it influenced, including views, which is what its algorithm optimises toward, so we never add those numbers together." Then show the ratio trend, because a stable gap is a measurement fact and a moving gap is news.
Where this goes wrong
- Adding platform conversions together. Meta, TikTok, Snapchat and Google Ads can all claim the same order.
- Changing the GA4 attribution model mid-quarter. The change rewrites past reports too, so last month’s deck no longer matches GA4.
- Reading yesterday at breakfast. GA4’s prior day typically settles by mid-afternoon property time, and key-event credit can change for up to 12 days.
- Mixing Google Ads columns. Conversions counts primary actions only; All conversions adds secondary ones. GA4 imports in Google Ads also follow GA4’s rules.
- Judging app campaigns on web GA4. If the purchase happens in the app, use your mobile measurement partner’s numbers for that campaign.
Do it faster with Whathead
Filling this table means pulling five sources every Monday. With Whathead, Claude or ChatGPT reads GA4 and the ad platforms through one MCP connection and lays the numbers side by side, without changing anything:
- “Compare GA4 purchases with Meta-reported purchases by day for the last 14 days.”
- “Build last week’s reconciliation table: Meta, TikTok, Snapchat and Google Ads reported purchases, GA4 purchases by session source for each, the ratio, and the attribution setting each platform used.”
Keep the tracking side honest with the paid media tracking QA routine. Whathead is our product.
GA4 vs ad platform conversions: common questions
Why does Facebook show more conversions than GA4?
Meta counts conversions within its attribution window after clicks and, depending on the setting, views, and it models some it can’t observe. GA4 only credits conversions it sees, by its own model, across all channels.
Which number is correct, GA4 or Meta?
Both are correct for their question. Use backend orders for totals, GA4 for channel mix, and Meta’s numbers for optimising inside Meta.
Why does GA4 show different conversions than Google Ads?
Google Ads dates conversions to the click and counts by its own conversion settings; GA4 dates them to the conversion and may credit other channels. Imported GA4 conversions follow GA4’s rules.
Can I make GA4 match Facebook Ads?
Not exactly, and you shouldn’t try. Fix tagging and duplicates, then track the gap as a ratio and investigate when it moves.
How big a gap between GA4 and Meta is normal?
It depends on the account’s mix of view-through, consent rates and buying cycle. Measure your own range over several weeks rather than using someone else’s number.
Sources
- GA4 Help (Checked 19 Nov 2026): attribution models, attribution settings, modelling and data freshness, consent mode behavioural modelling, data freshness
- Google Ads Help (Checked 19 Nov 2026): why Google Ads and Analytics data differ, import Analytics conversions
Written by Ahmed Shahin, founder of Whathead. Last reviewed 19 November 2026.
Fill the reconciliation table from one chat
GA4, Meta, TikTok, Snapchat and Google Ads numbers side by side, read-only.



