What to check before you touch anything
- What attribution model does each system use? Google Ads defaults to its own attribution model (usually data-driven), and GA4 may use a different model (or the same model with different cross-channel distribution rules) — the mismatch may be purely a methodology difference, not an error.
- Does the conversion window match between the two systems? Different windows (say, 30 days in Google Ads versus 90 days in GA4) will produce different numbers from the exact same underlying data.
- Are both systems counting the same types of conversions? If the ROAS goal in Google Ads includes only part of your e-commerce activity while GA4/BI counts all channel revenue, a mismatch is unavoidable.
- Is anything being duplicated or dropped at the technical level? Check that the GA4 tag and the Google Ads tag aren't conflicting with each other, creating either duplicate or missing transactions.
- Does the comparison period actually line up (click date vs. conversion date)? With a long purchase cycle, a conversion can land in different reporting periods across different systems depending on how the date is calculated.
- Is cross-channel interaction being accounted for (say, a user clicked an ad but purchased later via a separate direct-visit session)? Depending on the attribution model, that sale might get credited to Google Ads by one system and to a different channel by the BI system, or vice versa.
Possible approaches
- If the gap is explained by different attribution models and different windows, this usually isn't an "error" to fix — it's a fundamental property of different measurement systems. In that case, it's more useful to document the reasons for the discrepancy once for the team/client than to try to force the numbers to match.
- Google recommends using Google Ads data (or GA4 data, if GA4 is set up as the conversion source for Google Ads) primarily for bid optimization, and BI/CRM data as the source of truth for evaluating real business profitability — these are two different purposes for metrics that look similar but are calculated differently.
- If the gap is unusually large (a multiple, not a 10–20% variance), first rule out a technical cause — double-tagging, GA4 filter errors, cookie-consent traffic blocking — before chasing a methodological explanation.
- For reporting to the business, it's useful to pick one "primary" system as the source of final numbers (say, BI/CRM, if that's where real money hitting the bank account gets tracked) and clearly label the other systems as operational tools for optimization, rather than mixing numbers from different sources in one report without explanation.
- If you need to bring the numbers as close together as possible, you can unify attribution windows and models wherever technically feasible (for example, using GA4 as the single conversion source for Google Ads), while understanding that a perfect match between independent measurement systems is generally not achievable in principle.