Building a ROAS Dashboard Your CFO Will Actually Trust

The ROAS dashboard setup is a control system

A dashboard with 12 polished tiles and no action thresholds is worth £0. The exception is a temporary diagnostic view built to answer a live question. Permanent tiles create reconciliation work, meeting time and false confidence unless somebody must act when their numbers cross a line.

Our position on ROAS dashboard setup is blunt: a dashboard earns its place only if every tile has a threshold attached; a tile that cannot be wrong is decoration.

Google Ads can report a number that is internally consistent but commercially misleading. Imported values, duplicated events and immature revenue can all create the gap explained in our analysis of why reported Google Ads ROAS can differ from commercial reality.

Every permanent tile needs five fields: source, calculation, comparison window, trigger and required action. Missing any one leaves room for the agency, marketer and CFO to defend different versions of the same result.

A permanent tile without a threshold will not change a budget, tracking or sales decision; one documented decision caused by such a tile in each of two consecutive cycles would prove the claim wrong.

A trusted dashboard is a control system, not a gallery.

Build four cadence tiers, not one crowded page

Breakage expires in hours, while B2B revenue can mature over months. Putting both on one screen encourages teams to overreact to recent noise and overlook slow commercial deterioration.

Use four cadence tiers containing three headline tiles each—12 tiles in total.

Daily breakage

Spend pacing: yesterday’s Google Ads cost against the approved daily pace.
Primary conversion-event count: the events Google Ads received, clearly labelled as tracking signals rather than leads.
Click-to-session reconciliation: Google Ads clicks against GA4 paid landing sessions.

These tiles answer whether delivery, tagging or the landing journey has broken. They do not justify a budget increase.

Weekly performance

Cost per raw lead: spend divided by accepted CRM lead records.
Qualified-lead count: leads carrying the agreed sales qualification status.
Cost per qualified lead: spend divided by qualified leads.

Weekly data is usually mature enough to expose waste without pretending that every lead has completed the sales process.

Monthly commercial

Ad-sourced opportunities created: CRM opportunities linked to paid acquisition.
Pipeline created: the value entering the agreed opportunity stage.
Recognised revenue from matured cohorts: finance-recorded revenue linked back to an eligible acquisition cohort.

The maturity rule matters. Revenue from old leads should not be presented as the immediate product of this month’s spend.

Quarterly trend

Rolling 90-day pipeline ROAS: pipeline created divided by advertising spend.
Rolling 90-day gross-profit ROAS: recognised revenue multiplied by gross margin, then divided by spend.
Median lead-to-win days: the time between CRM lead creation and the recorded win date.

Show four values on every tile: actual, comparator, threshold and owner. Keep diagnostic dimensions underneath, not beside, the headline number.

Cadence determines which decision belongs on which screen.

Source-of-truth contracts for ROAS dashboard setup

Finance does not trust a metric because Looker Studio renders it neatly. Trust comes from knowing which system is allowed to supply each input—and which systems are explicitly prohibited.

Adopt these 14 one-line source contracts:

Approved budget: the finance-approved planning sheet supplies it; Google Ads recommendations and Looker Studio inputs may not.
Ad spend: Google Ads supplies account-currency cost; GA4 and the CRM may not.
Ad clicks: Google Ads supplies them; GA4 sessions and CRM visits may not.
Paid landing sessions: GA4 supplies them; Google Ads clicks and CRM page views may not.
Primary conversion events: Google Ads supplies the diagnostic count; CRM leads and GA4 event totals may not.
Raw lead count: the CRM supplies accepted lead records; Google Ads conversions and GA4 events may not.
Deal-to-ad linkage: the CRM’s captured campaign identifier supplies it; GA4 channel grouping and salesperson memory may not.
Qualified-lead count: the CRM’s named qualification state supplies it; form scores and Google Ads labels may not.
Opportunity count: CRM stage entries supply it; Google Ads and GA4 may not.
Pipeline value: the CRM amount at the agreed stage supplies it; platform conversion values and manual dashboard edits may not.
Recognised revenue: the finance ledger or invoicing system supplies it; CRM closed-won value and Google Ads may not.
Gross margin: the finance-approved margin table supplies it; campaign-manager assumptions and dashboard constants may not.
Lead-to-win duration: the controlled calculation sheet uses CRM timestamps; attribution windows and GA4 dates may not.
Calculated efficiency metrics: the controlled calculation sheet supplies them from contracted inputs; native platform ROAS columns may not.

The attribution rule and version should appear as metadata, not become a second calculation inside the dashboard. Use the separate guide to revenue attribution models to choose that rule and our explanation of data-driven attribution in Google Ads to understand why platform-assigned credit is not a finance ledger.

A workable sheet blueprint has nine tabs: targets, campaign_map, ads_daily, ga4_daily, crm_leads, crm_opportunities, finance_revenue, calc and decision_log. Raw tabs remain append-only. Looker Studio reads the controlled calc output.

Join advertising data by immutable campaign ID and date. Do not blend daily spend directly into individual lead rows; that one-to-many join can repeat the same cost for every matching lead.

Use a six-field decision log: breach date, tile, threshold, owner, action and outcome. That log determines whether a tile deserves to survive the next quarterly review.

The same boundary helps scope support. Ongoing Google Ads management suits teams needing weekly operation and optimisation, while a defined Google Ads project suits a bounded setup or repair.

One metric gets one owner and one permitted source.

Give all 12 headline tiles a decision threshold

Targets belong in the targets tab, with an effective date and approver. Account economics should set them; our 2026 B2B Google Ads benchmarks can challenge an implausible assumption but should never replace the company’s unit economics.

Use these 12 threshold rules as operating defaults:

Spend pacing: yesterday below 80% or above 120% of approved pace → inspect billing, disapprovals, schedules and caps before changing bids.
Primary conversion events: 0 yesterday versus a seven-day median of at least 2 → submit a test lead and inspect Google Tag Manager, Consent Mode and form receipt.
Click-to-session reconciliation: GA4 paid sessions below 70% of Google Ads clicks for two consecutive days → inspect redirects, tagging and consent configuration before making a budget decision.
Cost per raw lead: above 120% of target for two consecutive weeks → halt campaign expansion and inspect search terms and landing paths.
Qualified-lead count: below 80% of plan for two weeks while raw leads meet 100% of plan → review qualification and routing with sales before altering bids.
Cost per qualified lead: above 115% of its ceiling across four weeks → move marginal budget to a campaign below the ceiling; hold it if none qualifies.
Opportunity count: below 75% of plan while qualified leads meet 100% → inspect sales acceptance and CRM stage discipline before blaming media.
Pipeline created: below 80% of monthly plan → freeze the next spend increase and isolate the campaign, offer or sector causing the shortfall.
Recognised revenue: below 80% of finance plan on the agreed cohort-maturity day → reconcile invoices and CRM outcomes, then restate the forecast.
Rolling pipeline ROAS: below its approved floor at two consecutive month-end reads → stop expansion and assign a recovery plan.
Rolling gross-profit ROAS: below 1.0x after cohort maturity → stop scaling unless the CFO signs a strategic exception.
Median lead-to-win days: above 125% of the previous 90-day median → extend the maturity window and exclude recent cohorts from final ROAS.

A colour may mirror a threshold, but red, amber or green is never the required action.

Thresholds turn observation into controlled action.

If the path from ad click to signed revenue still has gaps, Actualyse will trace each hand-off with you — book a call

A UK B2B worked example

Take an illustrative UK consultancy spending £8,000 in March and reading the cohort at day 90. Use these eight labelled inputs:

Google Ads spend: £8,000
Raw CRM leads: 40
Qualified CRM leads: 16
CRM opportunities: 6
Pipeline created: £40,000
Google Ads conversion value: £32,000
Finance-recognised revenue: £18,000
Finance-approved gross margin: 55%

The four approved thresholds are a £250 CPL ceiling, £550 CPQL ceiling, 4.0x pipeline ROAS floor and 1.50x gross-profit ROAS floor.

The six calculations are reproducible:

CPL = £8,000 ÷ 40 = £200
CPQL = £8,000 ÷ 16 = £500
Qualified-to-opportunity rate = 6 ÷ 16 = 37.5%
Pipeline ROAS = £40,000 ÷ £8,000 = 5.0x
Revenue ROAS = £18,000 ÷ £8,000 = 2.25x
Gross-profit ROAS = (£18,000 × 55%) ÷ £8,000 = £9,900 ÷ £8,000 = 1.24x

CPL is £200 actual versus a £250 ceiling, so it passes. CPQL is £500 versus £550, so it also passes. Pipeline ROAS is 5.0x versus a 4.0x floor, yet gross-profit ROAS is 1.24x versus a 1.50x floor.

The decision is to withhold additional budget despite healthy lead and pipeline tiles. Google Ads reports £32,000 of conversion value versus £18,000 recognised by finance; the commercial dashboard therefore carries £18,000.

The honest limit here is that day-90 revenue is confounded by sales capacity, pricing changes and deal timing. Gross-profit ROAS cannot isolate advertising causality or include every operating cost.

Commercial truth outranks platform optimism.

What does not work: impressions, average position and unused tiles

A CFO-facing dashboard becomes stronger when low-consequence metrics are removed. Apply these three removal rules:

Remove impressions from the headline view. Impressions can move because demand, budget availability or auction eligibility changed, without producing a commercial decision. Keep them in a diagnostic drill-down.
Remove average position. It is not a current Google Ads metric, and reconstructed substitutes encourage optimisation towards an unavailable objective rather than revenue or qualified demand.
Remove any tile nobody acted on during a quarter. The decision log provides the evidence. Preserve the raw field if analysts need it, but delete the permanent tile.

A red decorative tile fails the same test as a neutral one. Visibility without a predetermined response only creates another number to explain.

Unused and unactionable tiles deserve deletion.

FAQ

Four implementation decisions prevent common governance disputes.

How often should dashboard targets change?

Lock targets for one quarter. Reopen them early only if pricing, gross margin or approved budget changes by at least 10%, and version the effective date rather than overwriting history.

How much history is needed before launch?

Import at least 90 days and one complete median sales cycle, whichever is longer. With only 45 days available, label commercial tiles provisional and prohibit quarterly budget changes based on them.

What should happen when campaign identifiers are missing?

Show matched and unmatched revenue separately. If identifier coverage falls below 70% across 30 days, prohibit campaign-level revenue claims and use aggregate paid-acquisition reconciliation until capture is repaired.

How should refunds and credit notes be handled?

Deduct them in the finance extract. If adjustments exceed 5% of a cohort’s recognised revenue, restate that cohort’s revenue and ROAS rather than carrying the correction into the current month.

Dashboard governance should be specific enough to schedule.

Summary

Use these five operating rules:

Delete every permanent tile lacking a measurable trigger and predetermined action.
Separate daily, weekly, monthly and quarterly decisions into four cadence tiers.
Give each metric one permitted source; finance owns recognised revenue and gross margin.
Trigger a tracking check at 0 events versus a seven-day median of at least 2.
Stop scaling when matured gross-profit ROAS falls below 1.0x without CFO approval.

Every trusted tile can fail and force a decision.

Actualyse builds measurement and attribution setups that tie B2B ad spend to real revenue. Book a call to talk through where yours stands.