Why Meta, Google Ads and GA4 Conversion Numbers Never Match

Meta, Google Ads and GA4 conversion numbers don't match because each platform measures with its own attribution model, conversion window, counting method and identity data. Meta and Google both take full credit for any sale they touched, while GA4 assigns credit across channels using its own logic. None of them sees the complete customer journey, so the numbers cannot line up by design. The most reliable reference is your backend sales data, validated with independent measurement.

Why the numbers can never match

Each platform answers a subtly different question. Meta Ads Manager answers "how many conversions happened after someone saw or clicked a Meta ad." Google Ads answers the same question for Google clicks. GA4 answers "which source gets credit under my attribution model for the conversions my tag observed." Three questions, three answers, and no reason they should ever be identical.

Four technical differences drive most of the gap:

• Attribution windows. Meta defaults to 7-day click and 1-day view. Google Ads commonly uses a 30-day click window. GA4 applies data-driven attribution with its own lookback. The same purchase can fall inside one window and outside another.
• View-through conversions. Meta counts conversions from people who only saw an ad. GA4 has no visibility into ad impressions inside walled gardens, so it never credits views at all.
• Modeled conversions. Since iOS 14 and growing consent restrictions, both Meta and Google fill measurement gaps with statistical modeling. GA4 also models, but with different inputs, so the estimates diverge.
• Counting and timestamps. Platforms may report the conversion on the date of the click, while GA4 reports it on the date of the purchase. One platform may count multiple conversions per user where another deduplicates.

The double-counting problem

Imagine a shopper who clicks an Instagram ad on Monday, searches your brand on Thursday, clicks a Google ad, and buys. Meta claims that purchase. Google Ads claims it too. GA4 hands it to paid search under last non-direct click. One order, three reports, and every one of them feels entitled to the sale.

This is why platform-reported conversions commonly exceed backend revenue when you sum them across channels. Each platform is grading its own homework with rules that favor itself. That is not fraud; it is the inevitable result of siloed measurement, and it is exactly why summing platform numbers is one of the most common and costly reporting mistakes.

What each number is actually good for

A simple monthly reconciliation habit

You do not need to make the numbers match; you need to make them explainable. Once a month, line up each platform's reported conversions against your backend orders for the same period, on purchase date rather than click date. Note the summed overlap, the view-through share Meta is claiming, and the slice GA4 lost to consent. Write those figures down as your expected baseline. After that, the monthly review stops being an argument about whose number is right and becomes a quick scan for anomalies, a tag that broke, a window that changed, a sudden jump in modeled conversions, so you catch tracking failures in days instead of quarters.

What to trust, in order

1. Your backend first. Orders, revenue and margin from your store or CRM are the only numbers that reconcile with your bank account. Every other figure must be sanity-checked against them, never the other way around.
2. Platform numbers within their own walls. Meta data is fine for deciding which Meta creative wins, and Google data for tuning Google campaigns. Just never sum platform conversions across channels or treat them as incremental sales.
3. Independent measurement for budget decisions. Multi-touch attribution tools such as Triple Whale or Northbeam rebuild journeys from your own first-party data. Marketing mix modeling estimates each channel's contribution from spend and outcome data without user tracking. Lift and incrementality tests prove causation. The strongest setups triangulate all three, because each method has weaknesses the others cover.

If your MTA, your MMM and a lift test roughly agree on a channel, you can act with confidence. When they disagree, that disagreement is itself useful information about where your tracking is breaking down. This is where Admira fits. Instead of refereeing between Meta, Google Ads and GA4, it unifies multi-touch attribution, marketing mix modeling and lift testing on your own first-party data, so every channel is judged by one set of rules and reconciles against your backend. If three dashboards keep telling you three different stories about the same sale, book a demo and see the reconciled number instead.

FAQ

Is GA4 more accurate than Meta or Google Ads?

Not exactly; it is differently limited. GA4 undercounts because of consent banners, ad blockers and cross-device gaps, while ad platforms overcount their own contribution with view-through and modeled conversions. Treat GA4 as a directional cross-channel view rather than as truth, and reconcile all three against backend orders, which are the only figures that tie to your bank account.

Should I turn off view-through conversions in Meta?

Not necessarily. View-through data is genuinely useful for judging creative reach and awareness inside Meta. The mistake is treating view-through conversions as incremental sales when you compare Meta against click-based channels. Keep them for in-platform creative decisions, but exclude them when you build a cross-channel budget picture, or you will over-credit Meta.

How big a mismatch between platforms is normal?

There is no universal figure, but summed platform conversions exceeding backend orders by a wide margin is common, especially for brands running several channels with heavy retargeting. The overlap grows with the number of platforms and the intensity of retargeting, because more platforms claim the same shoppers. Consistency matters more than the exact percentage: track your usual gap and watch for sudden changes.

How do I compare channels fairly then?

Use a single independent yardstick instead of three self-interested ones. That means an attribution layer built on your first-party data, a marketing mix model, or periodic incrementality tests, applied consistently across every channel. Fair comparison requires one referee, not three players keeping their own score, and ideally you triangulate all three methods so their weaknesses cancel out.

Which number should feed my ad platform bidding?

Each platform's own conversion signal should feed its own bidding algorithm, because Smart Bidding and Meta's optimization learn from the events they can see. Keep those tags clean and well-deduplicated. Just do not sum those same numbers across platforms for reporting, and never treat them as your true revenue; that job belongs to your backend and an independent measurement layer.