By Variafy team / Updated 20 October 2026

Google Ads vs GA4 conversions: a filled reconciliation worksheet

Google Ads and GA4 conversion totals can differ because the action, traffic scope, dates, counting or attribution differ. Record both report definitions first, then separate explained differences from confirmed faults and unanswered questions; a gap alone does not identify the cause.

Two translucent tracing sheets with different measurement marks overlap while their Ads and GA4 labels remain separate on a warm grey background.

On this page8 sections

What exactly are the two numbers comparing?

Fill a comparison card before deciding which number is wrong. Our suggestion is to preserve enough context that another person can reproduce both exports.

Google’s Analytics-import guide explains that Analytics can include non-Google Ads traffic and that discrepancies can occur even with a correct setup. That does not establish that every gap is harmless or that the two systems can never agree.

The following is a filled, invented diagnostic snapshot for a service business. All names, counts, dates and settings are teaching inputs. The initial screenshot showed 120 all-traffic GA4 key events and 72 Google Ads conversions. That pair is not yet a like-for-like comparison.

	Google Ads side · invented	GA4 side · invented
Action	Website enquiry success, native website action.	generate_lead key event on the same intended completed enquiry. Trigger equivalence still needs a controlled test.
Scope	One advertiser, one Search campaign, only the named action.	The named event with an event-scoped Campaign filter for that campaign; not the all-traffic total.
Period	1–30 September; account time zone Europe/Warsaw.	1–30 September; property time zone Europe/Warsaw.
Date basis	Conversions (by conv. time), rather than the initial interaction-date column.	Event-date report, with the exact report and filters retained.
Counting and goals	One after an interaction; action is primary and its standard goal is used. No custom goal in this example.	The event’s configured counting method is retained with the export; equal names do not establish equal counting.
Credit and window	Action attribution model and conversion window saved with the export.	Data-driven reporting attribution and the property’s applicable window saved; not assumed identical to Ads.
Observed result	72 conversion credits in the narrowed view, exported 6 October at 10:00 UTC.	64 key-event credits in the narrowed view, exported at the same time.

The card narrows the question; it does not certify full equivalence. In this example the operator still needs the event test and confirmation of counting and attribution differences. Our reading: “not fully comparable yet” is a useful finding, rather than a reason to average the two totals.

Are you comparing the same action, count and attribution scope?

Check the measurement route and actual report settings before editing tags. A native website action and an Analytics-based action can describe the same business event through different routes.

Google supports Analytics imports alongside Google Ads conversion measurement; it does not establish a universal rule that one route is always correct for every business. The useful question is what each configured action records and what the campaign actually uses.

Google’s conversion-column reference distinguishes Conversions from the broader All conversions, which includes secondary actions and special sources such as view-through conversions. Do not compare a single GA4 event with an unexplained total of unrelated Ads actions.

The primary/secondary guide adds a necessary exception: secondary actions included in a custom goal are used for reporting and bidding regardless of that label. For standard goals, primary actions are used for bidding when the goal is used. Keep campaign goals in the evidence, rather than assuming a label alone settles the comparison.

Google’s import guide also distinguishes One from Every after an interaction. Repeated enquiries can therefore be counted differently. Our suggestion is to document the configured rules and test the intended success event, without changing counting merely to make the totals match.

A session filter is not the same as event attribution

Analytics attribution settings affect key-event reports using event-scoped traffic dimensions such as Source, Medium and Campaign. User- and session-scoped dimensions, such as Session source, are unaffected by changes to the reporting attribution model.

The same page distinguishes Google paid channels from Paid and organic channels for conversion credit. Our reading: write the dimension scope and credit setting on the card; a google / cpc session filter does not by itself reproduce an event-attribution report.

Google’s column reference itself points both ways: it says to use Event-scoped reports, and also to filter Session source / medium to google / cpc. Our reading, based on the attribution-settings page: only event-scoped dimensions follow the reporting attribution model, so write down which one you used.

Google also documents fractional credit under data-driven attribution. A difference between attributed totals is not automatically the same number of missing people, enquiries or sales.

Do dates and processing explain any part of the gap?

Align the date basis, account/property time zones and export time before looking for missing events. These checks can explain a period boundary without repairing or replacing a tag.

The import guide dates Google Ads conversions to the click that led to the action and Analytics conversions to the action date. The column reference likewise describes click-time primary columns and provides Conversions (by conv. time) and All conv. (by conv. time).

There is different wording in the current data-discrepancies page: its opening delay explanation says ad impression date. These Google pages were checked on 19 Oct 2026. We retain the difference rather than silently treating impression and click dates as the same thing; record the actual column and measurement route you are comparing.

The by-conversion-time columns address timing, not all scope or attribution differences. Google’s column reference also excludes store-visit and store-sales conversions from those columns. This worked example concerns website enquiries, not those excluded outcomes.

Processing descriptions have different scopes too. The import guide says imported Analytics key-event data can take up to 24 hours to become available.

The discrepancies guide says to allow 24–48 hours for data synchronization and processing between Google Analytics and Google Ads before comparing results. It also gives up to 7 days for conversion data to fully appear after significant changes or account linking. The column reference additionally says reporting lags can take up to 24–48 hours.

Our reading: these statements do not give one universal missing-event deadline. Preserve the import route, recent setup changes, observed time and maturity of the selected period. Re-export the same comparison after the relevant processing check instead of treating every early snapshot as final.

What is explained, faulty or still unresolved?

Use a separate evidence ledger so a plausible cause does not become an invented explanation. Our proposed ledger distinguishes a difference in meaning from a reproduced implementation fault.

In the invented example, the business system records 90 successful enquiries across all sources. That is an operational baseline with its own outcome and scope, not a third attribution total that Ads and GA4 must each equal.

	Finding in the invented case	Verdict and next evidence
Initial 120 vs 72	GA4’s 120 covered all traffic, whereas Ads’ 72 covered the named advertising action.	Explained comparison mismatch. Retain the original exports; use the narrower card for the next question.
Narrowed 72 vs 64	The chosen dates and time zones are recorded, but counting and credit equivalence are not established.	Unresolved. The difference of 8 is a difference in reported totals, not proof of 8 lost enquiries. Do not allocate it to consent, modelling or tags without evidence.
Business baseline (90)	Successful enquiries from every source under the business system’s rule.	An operational baseline, not an attribution total. But 120 all-traffic key events versus 90 successful enquiries also warrants a test of the event definition and repeated firing; do not treat the difference as a count of organic leads.
Trigger or duplicate concern	Indication, not proof: GA4 shows 120 key events across all traffic, while the business records 90 successful enquiries across all sources. Test one completed enquiry and, separately, repeated submissions or reloads.	Confirmed fault if a test reproduces a success event without a completed enquiry or duplicate firing for the same completed enquiry under the intended rule. Otherwise retain unresolved status.

even with a correct setup.
Google Ads Help · checked 19 Oct 2026

That short qualification is not permission to stop investigating. If a controlled test reproduces a missing success event or a duplicate trigger, record the evidence as a fault and validate the correction. If it does not, retain the unresolved status instead of assigning a cause from the size of the gap.

For a website test, use the tracking investigation. For a CRM outcome route, use the offline-conversions guide. Those are separate investigations; this card determines which evidence you need.

What can Variafy compare?

Variafy offers read-only Google Ads access after your beta invitation. GA4 is also available after your beta invitation. It reads campaign-level Ads clicks, cost and conversions alongside GA4 sessions and key events for google / cpc by session-scoped campaign, matching available campaign IDs over the chosen days.

It does not reproduce the event-scoped filter used on the card. This can surface campaigns with missing or uneven corresponding readings. It is not an exact match of individual enquiries, proof of a tag or consent fault, or an automatic alignment of every event, goal, attribution model and date basis in the comparison card.

Our suggested reading request is: “Keep Ads and GA4 readings separately labelled with their scope, selected days and missing evidence. List which campaigns show a gap and which scope questions remain open for a person to confirm.”

This workflow does not repair tracking, import data, send data or change Google Ads.

What should the client receive after the reconciliation?

Send the comparison card, the evidence ledger and a bounded next action. The useful deliverable is an explanation someone can reproduce, including what remains unknown.

Our suggested client note

“We initially compared all-traffic key events with one advertising action. The narrowed reports now show 72 Ads conversion credits and 64 GA4 key-event credits on the recorded basis. We have not established that the remaining difference is a tracking fault.”

“The site/tag operator will test the completed enquiry and confirm counting; the analyst will retain the attribution and window evidence before the next decision.”

Replace every invented count with the actual labelled reading and preserve unresolved evidence. Do not add Ads and GA4 credits together, take their midpoint or change a window merely to tidy a dashboard. To place this reconciliation in the wider monthly report, see the monthly report template for lead campaigns.

For the broader reporting vocabulary, use how to read a Google Ads report. For business economics, use cost per lead; platform-credit reconciliation does not determine which enquiries became paying customers.

How we checked

Checked on 19 Oct 2026 against 5 Google Help documents: Analytics conversion imports; data discrepancies; Analytics attribution settings; conversion tracking data and columns; and primary/secondary actions. Click-date versus impression-date wording and differently scoped processing periods are shown explicitly.

The comparison card, evidence ledger and client note are our suggestions. The entire service-business snapshot is invented, including dates, settings and counts. Variafy is our product.

Questions

Why do Google Ads and GA4 conversions differ?

They may describe different actions, traffic, count rules, attribution or time periods; missing or duplicate measurement is also possible. Record both definitions, align the intended comparison and test suspected faults. The gap alone does not establish which cause applies.

Should Google Ads and GA4 conversions always match?

No. Equality is not a universal requirement, and unequal totals alone do not prove a fault. A carefully defined comparison can still differ because its attribution or processing differs. Do not change the meaning of a metric just to force matching.

Which Google Ads date column should I compare with GA4?

Google provides Conversions (by conv. time) and All conv. (by conv. time) for conversion-date comparisons. Choose the action scope too. Its import/column guides describe click-date primary reporting, while the discrepancies page says impression date; keep the actual report basis explicit.

Is there a normal percentage difference between Ads and GA4?

One Google reference page gives an informal expectation for differences but does not state which settings it assumes, so we do not use it as a threshold. Check action, scope, counting, dates, attribution and processing, then test suspected faults; the gap’s size alone does not establish a cause.

Does the CRM settle which platform attribution total is right?

Business records can verify the recorded enquiry or sale under their own rules. They do not reproduce every platform’s attribution credit. Keep operational outcomes and labelled platform readings separate, with a specific next check for unexplained differences.