A marketing dashboard your CFO would sign off on

Marketing reports what the ad platforms report: attributed conversions, platform ROAS, cost per lead. Finance reports what the bank shows: revenue recognised, costs incurred, margin realised. These are not two views of the same number — they're measurements of different things using different rules.

The largest discrepancy is double counting. A customer who saw a Meta ad, searched the brand, clicked a Google ad and bought will be claimed in full by both platforms and often by the email tool as well. Each is honest inside its own attribution model, and summing them produces a figure that exceeds actual revenue.

The second is exclusion. Platform ROAS counts gross revenue against media spend, omitting cost of goods, shipping, returns, discounts, payment fees, salaries, tooling and agency fees. Finance includes all of them. A campaign can look profitable in one system and loss-making in the other with neither party making an error.

A dashboard finance will accept has a specific property: every number on it traces to a system finance already trusts. That constraint eliminates most of what marketing dashboards usually show, and what remains is short.

None of this means deleting platform reporting. ROAS, cost per click and conversion rate are genuinely useful for comparing two ad sets inside the same account over the same window, where the attribution model and cost exclusions are identical for both.

The discipline is keeping them in the operating layer and out of the executive one. Media buyers should absolutely watch platform metrics daily. Board decks should not contain them, because they'll be compared against finance's numbers and the meeting will be about the discrepancy rather than the business.

Label them clearly when they do appear. 'Meta-attributed conversions' is an honest column header. 'Conversions' is not, and it's how the argument starts.

Most companies don't need a warehouse to do this. The minimum viable version is a spreadsheet updated weekly, pulling revenue and order counts from your commerce or billing system and spend from platform invoices.

The parts that need engineering are the ones that improve the inputs rather than the presentation: server-side conversion tracking so platforms optimize toward real events, and offline conversion imports so long-cycle businesses can feed closed-won revenue back.

Add a post-purchase attribution survey — one question at checkout — for an independent signal that doesn't share a failure mode with pixel tracking. When it disagrees sharply with your platforms, the direction of the disagreement is informative even though the data is self-reported.

The measurement improvement matters less than the political one. When both functions open the same dashboard, the conversation shifts from whose numbers are right to what to do about them.

In the engagement that opened this piece, building one reconciled model surfaced $2.1M of annual spend producing applications but not funded customers. Neither team could have found it alone — marketing's data said those campaigns worked and finance's data couldn't see campaigns at all.

That's the argument for doing this before any channel optimization. Not because the reporting is more elegant, but because you cannot allocate budget sensibly while two functions hold irreconcilable beliefs about what the budget did.