Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Google Ads › Intermediate

Reconciling Google Ads and GA4 Conversion Counts

You compare an accountancy firm's Google Ads and Google Analytics conversion reports for a closed month, separate faults from structural differences, estimate how much each cause contributes and write the note you would send its finance lead. The exercise practises explaining a gap by cause rather than forcing two systems to agree.
ExerciseIntermediate levelModule 3: Measurement and account audits5 minUpdated

Why this matters

Two reports that disagree invite the wrong reactions: picking the higher number, or making one system copy the other. The Google Ads and Google Analytics comparison explains why the systems differ by design, and the platform versus analytics discrepancies guide gives the reconciliation method. Neither can tell you which cause dominates a particular account; that takes reading the settings, starting by separating faults, which you fix, from structural differences, which you document.

The two reports

Riverstone Accounting is an invented firm of accountants that wins clients through an enquiry form and phone calls. Its finance lead has noticed that Google Ads reports far more enquiries than Google Analytics credits to paid search, and neither matches the client records. The month closed beyond the longest conversion window, so both figures are final.

Illustrative example — not a real result

Riverstone Accounting: one closed month

Google Ads conversion actions (Conversions column, all marked primary):

Conversion actionSourceCountingConversions
Enquiry form submittedGoogle Ads tag on the thank-you pageOne150
generate_leadImported from Google AnalyticsOne142
Calls from adsCall assets, calls of 60 seconds or moreOne38
Total330

Other Google Ads figures: spend £6,600; view-through conversions from a Display remarketing campaign, reported in their own column, 24; cross-device conversions within the enquiry form action, 9; click-through conversion window 30 days; attribution model data-driven; auto-tagging on; account time zone London.

Google Analytics figures for the same month: generate_lead key events from all channels, 260; generate_lead credited to Google paid search in the advertising report under paid-channel last click, 118; property time zone London; consent banner acceptance for analytics storage around 85%.

Client records: web enquiries logged, 255; phone enquiries logged, 70.

Change history: generate_lead was imported eleven weeks ago; the main campaign's £22 target CPA was set five weeks ago from the reported cost per conversion.

Explaining the gap

Follow the guides' method. Rule out faults first: duplicated actions, mismatched counting, auto-tagging, time zones. Align the units. Remove what Google Analytics cannot see: view-through and cross-device conversions, and calls that never reached the website. Then estimate the rest, journeys where another channel took the credit under Google Analytics' Attribution model and visitors who declined consent, and accept a small residual. The client records count the whole population and show whether either platform is plausibly counting real enquiries.

Task

  1. Identify any fault and how it affects the Google Ads figure and Smart Bidding.
  2. From the Google Ads total, remove the fault and each figure with no Google Analytics counterpart, showing a running total.
  3. Split the remaining difference between other-channel credit and consent loss, and state the residual.
  4. Calculate the cost per enquiry as reported, as corrected and as Google Analytics would put it.
  5. Decide which number to use for bidding, for site analysis and for reporting to finance.
  6. Write a short note to the finance lead: the real cost per enquiry, why the systems differ, and what happens next.

Worked answer

Illustrative example — not a real result

Riverstone's reconciliation

Fault: the imported generate_lead and the native form tag count the same submission and both are primary. Google Analytics' 260 against 255 logged web enquiries shows the event itself is sound; the error is making it primary.

Running total:

  • Google Ads Conversions column: 330
  • Remove the duplicate import (142): 188
  • Remove calls from call assets, which have no site visit for analytics to see (38): 150
  • Remove cross-device conversions (9): 141
  • View-through conversions (24) sit in their own column and were never in the 330, so nothing is removed for them.

Remaining difference against Google Analytics' 118: 23.

  • Other-channel credit: people who clicked an ad, then returned through organic search or email to convert. About 14, from the paths exploration.
  • Consent loss: about 15% of visitors decline analytics storage, so of 141 form enquiries roughly 7 to 9 would not be observed.
  • Residual: zero to two, within the precision of the estimates.

Cost per enquiry:

  • As reported: £6,600 ÷ 330 = £20.00
  • Corrected Google Ads view (forms and calls, no duplicate): £6,600 ÷ 188 = £35.11
  • Google Analytics paid-search view (forms only): £6,600 ÷ 118 = £55.93

The note to the finance lead:

"Last month Google Ads cost £6,600 and produced about 188 enquiries: around 150 forms and 38 calls, roughly £35 each. The 330 shown was inflated because each form was counted twice after an import eleven weeks ago; that is now fixed. Google Analytics shows 118 because it cannot see calls made from an ad, credits some enquiries to the channel people returned through, and misses visitors who decline cookies. Those differences are normal. The client records remain the count of enquiries, with the corrected Google Ads figure beside them. The £22 bidding target was set on the inflated count, so we will reset it in steps; expect the reported cost per enquiry to rise although nothing has got worse."

Which cause matters most. The duplicate import explains most of the gap and is the only fault. It matters beyond reporting: Smart Bidding has optimised towards a doubled count for weeks, and its target CPA was set from that count, so real enquiries can never meet it. The Google Analytics and Google Ads integration tutorial covers the primary and secondary choice that would have prevented it.

Which number for which decision. Bidding: the corrected Google Ads figure, which the bid strategy sees. Site analysis: Google Analytics. Finance: the client records, with the corrected Google Ads figure beside them.

Common mistakes

  • Explaining before checking for faults. Here the fault is the largest cause.
  • Subtracting view-through conversions from the Conversions column. They are reported separately and were never part of the total.
  • Bidding on the imported action alone so the numbers match. That adopts one system's limits.
  • Leaving the target alone after the fix. A target set on a doubled count is now impossible to meet.
  • Reporting a platform number to finance as the enquiry count. The client records are the count.

Carry the habit into the account audit capstone, where measurement is the first gate, as in the Google Ads audit walkthrough, or return to the intermediate Google Ads level.

Pages this lesson applies

Terms used on this page

Also related

Related