ROAS attribution windows can make one illustrative campaign look like 1.50× at 30 days and 4.00× at 90 days without changing a click, contract or pound of revenue. The longer view has not created value, and it can credit a click that did nothing. The 30-day Google Ads default is, in practice, set for ecommerce-speed buying; leave it in a B2B account and you choose to under-report every campaign that works slowly.
An attribution window is an eligibility rule, not proof of influence. It controls how long after an ad interaction a conversion may be credited. Window choice is one reason reported Google Ads ROAS can diverge from commercial return, even when every dashboard formula is correct.
Why default ROAS attribution windows misprice B2B
A prospect clicks on 1 March and signs on 15 April. A 30-day click-through window excludes that conversion; a 90-day window can include it. The signed revenue is unchanged. Only the eligible history differs.
Three controls are routinely confused:
| Control | The decision it makes |
|---|---|
| Conversion window | Whether a conversion remains eligible after an interaction |
| Attribution model | How credit is divided among eligible interactions |
| Report date range | Which dated activity appears on screen |
Google Ads documentation lists three headline defaults: 30 days for click-through, three days for engaged-view and one day for view-through conversions. Exact applicability varies by campaign and conversion source. For Search and Display click-through actions, Google lists options from one to 30 days, plus 60 or 90 days, depending on the source. Target ROAS bidding counts and optimises conversions eligible inside the selected window, so the setting is not cosmetic.
GA4 is different. Its property-level key-event lookback defaults to 30 days for first_visit and first_open, but 90 days for other key events. Since August 2026, Google Analytics conversion management has also supported per-conversion click-through windows from one to 90 days where available. A GA4 screen showing 90 days does not prove every Primary Google Ads action uses 90 days; inspect each conversion action.
The version we hit in audits is usually a Primary lead or offline-sale action created years ago and left at 30 days. The first line in our B2B Google Ads management audit is the action-level setting, not an account-level assumption. Window selection belongs in the wider B2B Google Ads strategy, alongside the commercial event being optimised.
Eligibility comes before allocation.
Choose ROAS attribution windows from the 90th percentile
Averages hide the slow tail that causes premature budget cuts. Start with matured conversion cohorts and measure calendar days from ad click to the commercial conversion used in ROAS.
Take an illustrative UK B2B software account. The conversion is a closed-won contract imported against the original ad click; the sample is 60 fully matured, matched conversions. Assume the conversion source offers 30-, 60- and 90-day settings. The three percentile comparisons are:
| Lag marker | Observed click-to-conversion lag | Comparison with the 30-day default | Practical window implied |
|---|---|---|---|
| Median (50th percentile) | 18 days | Default is 12 days longer | 30 days |
| 75th percentile | 42 days | Default ends 12 days early | 60 days |
| 90th percentile | 73 days | Default ends 43 days early | 90 days |
The median suggests 30 days is comfortable, yet it describes only the halfway point. The 75th percentile requires more than 30 days; the 90th requires a setting at or beyond 73 days. The next available setting at or beyond 73 days is 90.
Do not derive the tail from a Google Ads time-lag report capped by the current 30-day conversion window; that report cannot reveal what the setting excluded. Use matched CRM click and conversion timestamps. For ROAS, calculate two tails: one by conversion count and one by cumulative signed value. Use the later 90th-percentile day.
The governing decision rule is explicit: set the window at or beyond your 90th-percentile click-to-conversion lag. Apply these three rules:
- If both 90th-percentile lags are 30 days or fewer, keep the 30-day window.
- If either lag is 31 days to the platform maximum, select the next available setting at or beyond it.
- If either lag exceeds the platform maximum, set the platform to its maximum and run a CRM cohort report through the true 90th-percentile day. Use mature CRM ROAS for budget decisions and label platform ROAS incomplete.
The honest limit is that matched lag describes recorded converters, not causation. Sales follow-up speed, procurement delay and offer changes confound the tail. The claim is falsifiable: if three mature cohorts each place at least 90% of matched conversion count and contract value inside 30 days, the account-level case for extending beyond 30 days is wrong.
The sales cycle, not the setup screen, sets the honest horizon.
One campaign, two windows, opposite budget decisions
Take an illustrative UK cyber-security consultancy spending £8,000 a month on one Google Ads campaign. The six inputs are fixed:
| Input | Labelled value |
|---|---|
| Google Ads cost | £8,000 |
| Commercial conversion | Signed first-year contract revenue |
| Revenue eligible by day 30 | £12,000 |
| Additional revenue eligible on days 31–90 | £20,000 |
| Finance hurdle | 3.00× ROAS |
| Budget rule | Below 3.00×: cut to £4,000; at or above 3.00×: raise to £10,000 |
30-day reported ROAS: £12,000 ÷ £8,000 = 1.50×
90-day reported ROAS: (£12,000 + £20,000) ÷ £8,000 = 4.00×
At 30 days, reported revenue is £12,000 versus £32,000 signed by day 90. The 1.50× result fires a cut from £8,000 to £4,000; the 4.00× result fires an increase from £8,000 to £10,000. Those are opposite decisions for the same spend, campaign, model and revenue definition.
The 4.00× figure is not proof that advertising caused every contract. It is the more complete answer to the defined 90-day eligibility question. If the account uses qualified pipeline instead of signed revenue, apply a consistent lead-generation ROAS method and never switch value definitions between windows.
A cohort younger than the selected window must not trigger a revenue-led budget reduction greater than 20%. Earlier query, advert or landing-page changes need direct evidence from search terms and sales acceptance, not immature ROAS. That two-clock discipline is part of a viable B2B Google Ads strategy, not a reporting footnote.
Immature revenue is not evidence of an unprofitable campaign.
If the path from ad click to signed revenue still has gaps, Actualyse will trace each hand-off with you — book a call
The cost of long ROAS attribution windows
Ninety days buys completeness by selling speed and certainty. A longer window creates three costs.
1. Slower feedback
Spend appears immediately while eligible revenue arrives later. Under a 90-day window, a cohort aged 30 days still has 60 days to mature; judging it beside a completed cohort biases the comparison. Mark every cohort younger than 90 days as immature and reserve revenue-led scale or cut decisions for mature cohorts.
2. Blurred tests
Late contracts can arrive after several advert, offer or landing-page changes. In an illustrative timing example, a 14-day test judged on a 90-day commercial window reaches its final reading on day 104:
14 test days + 90 conversion days = 104 days
Tag results by click-date cohort and variant. Do not declare the revenue winner on day 15 merely because the test stopped buying traffic.
3. Credit assigned to clicks that did nothing
Eligibility expands faster than causal confidence. A click can sit inside 90 days of a contract without creating it. If the opportunity already existed before the click, classify the campaign as influenced rather than sourced; if the deal history contains no credible campaign role, exclude its value from campaign-sourced ROAS.
The honest limit of any long-window claim is that attribution cannot isolate cause. Sales intensity, procurement timing and price changes are three confounds. Effective Google Ads campaign setup and optimisation therefore uses a fast operational clock and a mature commercial clock.
A longer window deserves stricter evidence, not greater trust.
What does not fix a bad attribution window
Three popular account changes leave the eligibility boundary untouched.
1. Switching to data-driven attribution
Data-driven attribution redistributes credit among eligible interactions. A click outside a 30-day window remains ineligible, so a cleverer model cannot recover its revenue.
2. Expanding the report date range
Viewing 180 days instead of 30 days displays more cohorts. Each conversion is still tested against its own 30-day interaction window; the camera moved, but the fence did not.
3. Building a blended dashboard
A dashboard can reconcile labels and rearrange credited revenue, but it cannot manufacture an excluded interaction. The separate discipline of cross-channel ROAS reporting addresses duplicate channel claims, not conversions outside the source window.
Only a correctly configured conversion action changes future eligibility.
No downstream interface can recover credit the eligibility rule excluded.
FAQ
Four implementation decisions deserve fixed answers.
How many conversions are enough to estimate the 90th percentile?
Use 30 matched conversions as a working floor. At 30, the slowest three observations heavily influence the 90th percentile, so the estimate is still provisional. Below 30, round the longest credible observed lag up to the next available window and avoid pooling materially different offers.
When can we compare results after changing the window?
Google's setting applies going forward. Annotate the change date and wait one full selected window: after a 30-to-90-day change, allow 90 days before calling the first post-change cohort mature. Do not expect the new setting to backfill every excluded historical conversion.
Should every conversion action use the same window?
No. Set each Primary action from its own lag distribution. In an illustrative mapping, a demo request with a 12-day 90th percentile could use 30 days, while an imported sale with a 74-day 90th percentile needs 90 days. Keep newsletter sign-ups Secondary if they do not carry commercial value.
How should a monthly board report a 90-day window?
Use four maturity columns: 0–30, 31–60, 61–89 and 90-plus days. In an illustrative August report, May supplies mature revenue ROAS; June, July and August stay labelled immature. Budget decisions use the mature column while recent columns report delivery and lead quality.
A mature cohort is the minimum unit of honest B2B ROAS.
Summary
Apply these five rules:
- Keep 30 days only when at least 90% of matched conversion count and value arrives by day 30.
- Set the click-through window at or beyond the later 90th-percentile lag; use the platform maximum plus a mature CRM cohort when the lag exceeds it.
- Block revenue-led budget reductions above 20% until the cohort reaches the selected window.
- Compare identical spend, model and value definitions at 30 days versus the selected window before cutting or scaling.
- Treat window choice as a budget rule, never a reporting preference.
Actualyse builds measurement and attribution setups that tie B2B ad spend to real revenue. Book a call to talk through where yours stands.
