Why Ads, GA4 and CRM Numbers Don't Match: Discrepancies Explained

The short answer: they will not match, and that is fine

Each system answers its own question. An ad platform counts conversions it credits to its own ads. GA4 splits conversions across all channels with its own model. The CRM records real leads and deals but only knows their source if someone passed it along.

That is why conversions from Google Ads, Meta and Yandex Direct added together almost always exceed actual leads in the CRM: each platform can claim the same sale. The goal is not to make numbers match to the last unit but to understand why they differ and keep the size of the gap stable.

Attribution windows and models

• Attribution window: how many days after a click or view a conversion is credited to the ad. Windows differ between platforms and even between conversion actions.
• Attribution model. GA4 uses data-driven attribution across channels by default, an ad platform uses its own model within its network, and the CRM often records first or last source.
• Conversion date. Google Ads assigns a conversion to the date of the ad click, while GA4 assigns it to the date of the event. A lead submitted a week after the click lands in different reporting periods.

View-through conversions

Some platforms count view-through conversions by default: a user saw an ad, did not click, but bought later. GA4 and the CRM know nothing about that impression. Check the attribution settings in the ad account and compare click-through conversions only.

Consent and ad blockers

• If a user declines analytics cookies, tags may not fire or may send only anonymized signals.
• Ad blockers and browser tracking protection block pixels and analytics scripts.
• Click identifiers (gclid, fbclid, UTM) get lost in redirects, in some browsers and when crossing domains.

The CRM gets the lead regardless; analytics does not. That explains "missing" conversions and a growing share of direct traffic.

Time zones and processing delays

The ad account, the GA4 property and the CRM server may run in different time zones. A lead at 23:30 Tashkent time can fall on a different day in a system running on UTC. Data in GA4 and ad platforms is also processed with a delay, so the last day or two keep changing.

Modeled data and thresholds

• Modeling. When cookies are declined, ad platforms and GA4 may fill in some conversions with modeled estimates. The CRM has no such modeled leads.
• Thresholding. GA4 can hide rows with few users in reports so individuals cannot be identified.
• Sampling in GA4 explorations and in Yandex Metrica on large volumes produces approximate numbers.

Duplicate leads

• A user submits a form twice, or submits a form and also calls.
• The conversion tag fires on every reload of the thank-you page.
• The event is sent from both the website and a server API without a shared ID for deduplication.

The result: analytics shows more conversions than the CRM after duplicates are merged.

How to reconcile reports

1. Assign a source of truth per metric: spend from the ad platform, sessions and behavior from GA4, leads and sales from the CRM.
2. Compare like with like: same dates, same time zone, same conversion type, click-through only.
3. Pass identifiers into the CRM: UTM, gclid, fbclid, the GA4 client ID. Then every deal can be tied to a source.
4. Send data back to the ad platforms: offline conversion imports and server-side APIs (such as Meta's Conversions API) with an event_id for deduplication.
5. Remove duplicates: block repeat form submissions, fire the tag once per lead, set merge rules in the CRM.
6. Keep a discrepancy log with the typical gap between systems. A sudden change in the gap signals broken tracking.