£12,000 leaves the bank for Google Ads in April. Google Ads reports £73,200 of conversion value: 6.1x. GA4 assigns £40,800 to Paid Search: 3.4x. The CRM shows £19,200 of closed-won revenue: 1.6x.
Averaging them produces a fourth number—3.7x—with no defensible meaning.
ROAS discrepancies across tools arise because the tools are not auditing the same transaction. They use different outcomes, clocks, identities and credit rules. The useful question is not “Which dashboard is correct?” but “Which number fits the decision?” Reconciliation makes that choice explicit.
Why ROAS discrepancies across tools are normal
At its simplest, ROAS is attributed conversion value divided by ad spend. “Attributed” is the load-bearing word.
Every reported figure contains a measurement contract:
| Contract term | Common alternatives |
|---|---|
| Outcome | Form, qualified lead, pipeline, booked revenue or collected revenue |
| Unit | Event, person, buying account, opportunity or transaction |
| Clock | Ad interaction, website event, lead creation or deal close |
| Credit | Google-only, cross-channel, original source or latest source |
| Coverage | Observed users, modelled activity or CRM-matched records |
| Scope | One campaign, one platform or all paid media |
Change one term and the ratio can change without any arithmetic error. A platform may count an eligible conversion. GA4 may distribute the same event across channels. The CRM may merge three contacts into one buying account and wait until the opportunity closes.
The goal is therefore not identical dashboards. It is an explainable bridge from one definition to another.
That does not make every platform figure commercially sound. Our separate audit of what Google Ads ROAS can hide covers that issue. Here, the narrower job is to explain why systems disagree and decide which definition governs each decision.
What Google Ads, GA4 and the CRM actually measure
The systems sit at different points in the same chain.
| System | Starts with | Typical reporting clock | Best used for |
|---|---|---|---|
| Google Ads | Eligible ad interactions and configured conversions | Ad-interaction date in standard reporting; conversion-time views are available | Campaign, keyword, bid and creative decisions |
| GA4 | Website or app events it observes or models | Event time, with selectable attribution views | On-site behaviour and cross-channel diagnosis |
| CRM plus finance | Known contacts, accounts, opportunities and transactions | Lifecycle-stage, close, invoice or recognition date | Lead quality, pipeline and commercial return |
Google Ads is the execution view
Google Ads asks which eligible Google ad interactions led to the conversion actions configured in the account. Those actions might be qualified calls, demos or imported closed-won deals.
Its standard reporting can place a later conversion against the earlier ad interaction. Google’s own discrepancy guidance recommends conversion-time columns when comparing Ads with systems that report on the event date. That means a March campaign report rerun in May can legitimately contain more value than it did on 31 March.
This view is valuable for allocating spend inside Google Ads. It is not automatically a cash ledger.
GA4 is the journey view
GA4 starts with measured events and assigns credit under its selected attribution rules. Paid Search may receive all, some or none of an event’s value depending on the eligible channels, model, identity coverage and dimension used.
Even two GA4 reports can differ. A session-scoped acquisition report and an event-scoped attribution report answer different questions. GA4’s attribution documentation also distinguishes event time from ad-interaction time.
Use GA4 to understand landing pages, journeys and the relationship between channels. Do not assume it has seen every buyer or deduplicated several employees from one target account.
The CRM is the commercial view
The CRM follows named contacts and accounts through qualification, opportunity and close. It can remove spam, merge repeat enquiries and reveal that 40 forms produced six genuine buying accounts.
That makes it the strongest source for sales outcomes—but not automatically for marketing attribution. Original-source fields may be blank, overwritten or based on the last form submission. A forwarded link, phone call or lost click ID can separate a real customer from their digital history.
Finance then validates the amount: booked, invoiced, recognised or collected revenue. The CRM knows who progressed; finance knows what money the business can substantiate. Board reporting normally needs both.
How to reconcile ROAS discrepancies across tools
Consider a Manchester B2B software firm that spent £18,000, excluding VAT, on Google Ads in Q1. On 15 May, three reports show:
| Report | Value credited to Paid Search | Apparent return |
|---|---|---|
| Google Ads, Q1 ad-interaction date | £96,000 | 5.3x |
| GA4, Q1 event time | £38,000 | 2.1x |
| CRM, Q1 close date and original source | £54,000 | 3.0x |
None is ready to compare.
One £42,000 buyer clicked a Google ad on 28 March, returned through an email on 12 April and signed on 30 April. When the win is imported, Google Ads connects it to the March interaction. GA4 records the relevant event in April and may divide channel credit. The CRM records an April close.
One buyer. Three clocks. Potentially three credit allocations.
A clean reconciliation uses five steps:
- 1. Write the business question. In this case: “How much closed-won booked revenue came from Q1 Google Ads click cohorts by 30 June?”
- 2. Freeze the scope. Use the same Ads account, Q1 spend of £18,000, GBP, geography and VAT treatment. Do not compare campaign spend with an all-channel revenue numerator.
- 3. Align the outcome. Include closed-won booked revenue in every comparable view. Keep demo values and pipeline in separate columns; they are useful leading indicators, not interchangeable revenue.
- 4. Choose the clock and maturity date. Group customers by Q1 acquisition interaction, then allow the agreed sales-cycle window to mature through 30 June. This avoids dividing April spend by deals acquired months earlier simply because they closed in April.
- 5. Match records and explain the residue. Join click IDs and campaign parameters to CRM contacts, accounts and opportunities. Label unmatched value, cross-channel credit, deduplication, timing and later status changes separately. Do not bury them in an “other” row.
After alignment, the Q1 click cohort contains four won deals worth £96,000. On the agreed basis, £96,000 divided by £18,000 is 5.3x. The CRM and Ads can now be reconciled to the same cohort and outcome.
GA4’s 2.1x has not become “wrong”. It remains a Q1 event-time, cross-channel attribution view. It should not be averaged with 5.3x.
This reconciliation establishes a consistent attributed result. It does not prove that every pound was incremental, nor does it calculate profit after delivery costs. Those are different questions.
Which ROAS number should you trust?
Trust the source closest to the decision, then use the other systems as checks.
| Decision | Primary source | Required check |
|---|---|---|
| Did a mature ad-sourced cohort create revenue? | CRM joined to finance | Source completeness and cohort maturity |
| Should next quarter’s paid-search budget rise? | Finance-validated CRM cohorts | Ads trends, open pipeline and sales capacity |
| Which campaign or search term should change tomorrow? | Google Ads | Qualified-lead and win quality from the CRM |
| Which landing page or channel path is weak? | GA4 | Traffic quality in Ads and downstream CRM outcomes |
| Is total marketing spend efficient? | Finance-led portfolio reporting | Consistent cost and revenue scope |
| Did advertising cause additional sales? | A controlled incrementality test | Observational dashboards as supporting evidence |
This is a decision hierarchy, not a league table. For a board budget, a deduplicated and matured CRM cohort validated against finance normally outranks an event-value dashboard. For a bid change tomorrow, Google Ads has the most useful detail, provided downstream quality acts as a guardrail.
Never sum revenue claimed by several platforms. Never average their ROAS figures. Both operations combine incompatible credit systems and can count the same commercial outcome more than once.
Portfolio decisions also require clear distinctions between blended and platform ROAS and between ROAS and MER. Choosing the most flattering dashboard resolves neither.
When a ROAS mismatch means tracking is broken
An expected difference is stable and explainable. A broken chain creates a sudden, widening or unallocated difference.
Investigate when:
- Google Ads conversions jump after a release, while GA4 events and CRM records remain flat.
- GA4 form events continue, but new CRM contacts fall on the same date.
- Paid leads rise while click IDs and campaign-source fields become blank.
- Imported closed-won value exceeds the revenue finance can validate.
- One form, landing page, device type or campaign accounts for most of a new gap.
- Mature cohorts stop progressing even though platform conversion value stays healthy.
Monitor relationships, not just totals: Ads conversions to CRM records, GA4 lead events to matched contacts, matched click-ID rate, contacts to unique accounts, and CRM wins to the finance ledger. Establish a baseline for each conversion action and investigate material movement from that baseline. There is no universal acceptable discrepancy percentage, especially when one B2B deal can move a monthly ratio by several points.
Each report should state its outcome, unit, date basis, time zone, attribution rule, spend scope, value basis, maturity window, match rate, owner and refresh date. Keep the native figures beside the reconciled view so nobody mistakes a transformed number for raw data.
Where the join fails between the click and the sales record, the practical route for carrying ad interactions through to qualified B2B revenue is a separate implementation task. The same measurement contract should govern an in-house media buyer or external Google Ads campaign management: whoever changes spend must know which signal is directional and which one authorises budget.
Apply the same standard to published B2B client results. A useful result names the spend period, commercial outcome, value basis and observation window; a bare “6x ROAS” does not tell a founder enough.
FAQ
Why is Google Ads ROAS higher than GA4?
Google Ads and GA4 can use different eligible interactions, attribution rules, conversion actions and reporting clocks. First align the conversion, date basis, time zone and credit rules. A remaining explained gap is not evidence that either tool is malfunctioning.
Should Google Ads and GA4 match exactly?
Only under tightly aligned conditions, and exact equality is still not the objective. Processing delays, identity coverage and modelling can produce small differences. Track the expected relationship for each conversion action and investigate changes, rather than forcing one total to equal the other.
Is CRM ROAS the most accurate figure?
The CRM is usually strongest for qualification, opportunities and closed-won outcomes. Its acquisition attribution is only as reliable as its source capture, account deduplication and sales-stage hygiene. Validate revenue against finance and report the unmatched share.
How should a B2B company report ROAS with a long sales cycle?
Use acquisition cohorts and an explicit maturity window based on actual time-to-close data. Show platform and GA4 measures as early indicators, then restate the cohort when enough opportunities have resolved. Do not judge April acquisition from every deal that happened to close in April.
Summary
- ROAS discrepancies across tools usually reflect different measurement contracts, not bad arithmetic.
- Google Ads guides in-platform action; GA4 explains measured journeys; CRM and finance validate commercial outcomes.
- Reconcile the outcome, scope, clock, cohort, identity and credit rule before comparing ratios.
- Use the source closest to the decision and keep other systems as checks.
- Investigate sudden or unexplained gaps; preserve stable, explainable differences.
</article> e every dashboard to agree.
Actualyse builds measurement and attribution setups that tie B2B ad spend to real revenue. Book a call to talk through where yours stands.
