The meeting everyone has sat through

You present the quarter. Google claims R6 million in sales, Meta claims R5 million, email claims R2 million, and the business, according to the people who actually count the money, sold R9 million in total. Somewhere at the end of the table a finance director does the addition, says nothing, and quietly files your slides under fiction.

They're right to. Every platform marks its own homework, and they all mark the same customer. The woman who saw an Instagram ad on Tuesday, searched your name on Thursday and opened a discount email on Saturday has, by the platforms' combined reckoning, bought three couches. You sold her one.

Once finance has seen that happen, every number you bring afterwards gets discounted, usually by more than it deserves, and budgets start being cut on instinct. Instinct, in a CFO, tends to begin with the line it understands least, which is generally yours.

Attribution was supposed to fix this

Attribution models were meant to settle the argument by sharing the credit out sensibly. Mostly they gave everyone something new to argue about. Last-click hands everything to the final touch, which is why branded search always looks like genius. Data-driven models spread credit by rules that live inside the platform, a bit like a referee explaining his decisions in a language only he speaks. Switch models and your best channel can become your worst by lunchtime, without a single customer doing anything differently.

None of them answer the only question finance is really asking, which is whether the sale would have happened anyway.

What finance will believe

There are three ways to measure, in rough order of effort, and each answers a slightly different question.

Attribution reporting is cheap, always on and useful for running the work: comparing campaigns inside a channel, catching the ones that have broken. It's weak evidence for defending a budget, so stop asking it to.

Incrementality tests are where trust gets built. Hold back a matched region or a slice of your audience, show everyone else the campaign, and compare the sales. The gap is what the marketing added, and because it's measured in the business's own sales, there's nothing left for anyone to discount. It takes planning and a few weeks per test, which is a modest price for a number that ends arguments.

Marketing mix modelling uses two or three years of spend and sales data to estimate what each channel contributed, including things no tracking pixel can see, like TV and outdoor. It's powerful at scale and expensive overkill when most of your budget sits in two platforms. Plenty of mid-size businesses have been sold a model when a few well-run holdout tests would have told them more.

The cheapest credibility you'll ever buy

Keep a decision log. Every meaningful change gets one line: what you changed, why, what you expected and what actually happened.

"Moved 20% of brand search budget into prospecting. Expect total revenue flat, new customers up. After four weeks: …"

It looks like admin, because it is. Six months in, though, you'll have something no dashboard can give you, which is a track record, and you can walk into the budget meeting with decisions and their consequences instead of a deck of attributed revenue. It also spares the team from rediscovering the same lesson every January, a common and surprisingly expensive hobby.

Speak finance's language, in finance's numbers

The quickest way to earn trust is to stop inventing metrics. Report against revenue, gross profit and customer numbers from the business's own systems, and put total marketing spend beside them.

This is the kind of line that survives a CFO. In FY2026, a national retail group we work with grew digital revenue by 29.2% on a 16.2% increase in media spend. Both are the business's own numbers, and nobody has to take an attribution model's word for anything.

What to do this quarter

1. Add up what every platform claimed last quarter and set it against what the business actually sold. Take the gap to finance yourself, before they find it on their own.

2. Agree one headline number with them, ideally revenue or gross profit against total marketing spend, and report it the same way every month.

3. Run one holdout test on your largest line of spend.

4. Start the decision log today. It's the only thing on this list that costs nothing.

Where we stand

Most marketing teams can measure perfectly well. What they lack is a number finance believes, and that only comes from measuring in finance's own currency.

Analytics sits at the centre of how we work at Armada. Theo Bili, our co-founder and chief analyst, runs the live experiments, decision logs and reporting himself, and because we run the media too, what a test shows on Friday changes where the money goes on Monday.

If your marketing reports and your finance reports are telling different stories, let's talk.