Evidence-based marketing.

We started Pinstorm in 2004 because we were tired of arguing with people who didn't have to be right. We'd spent years inside Grey and Ogilvy watching senior creative people, senior strategy people and senior account people defend work with the same three sentences: "this will build the brand", "this is on strategy", "the client loves it". None of which is the same as "this will make money".

Marketing, when it works, is not magic. It is a small number of decisions made under uncertainty, tested against a P&L, and refined until they reliably produce more revenue than they consume. Everything else — the awards, the case films, the framework decks, the LinkedIn essays about emotion-led storytelling — is theatre. Useful theatre, sometimes. But theatre.

What evidence-based marketing actually is

Three things, in order:

One. Every claim has to be earned with data. If we tell you a campaign worked, we owe you the attribution model that proves it, the counterfactual we're measuring against, and the time window we're reading. "Engagement is up 40%" is not a claim. It's a feeling.

Two. Every spend has to tie to revenue. Not to a soft KPI. Not to "consideration". Revenue. If a marketer can't draw a line between the dollar that left the bank and the dollar that came back, the spend was a donation. We don't make donations.

Three. Every theory has to be tested before it becomes practice. The marketing industry runs on inherited theories — most of them written before the internet existed. Kotler's 4Ps. Aaker's brand equity pyramid. Reis & Trout on positioning. They were useful in 1985. They are mostly useless now. (We wrote about this here.) Evidence-based marketing means we test these theories on real budgets, on real audiences, with real revenue at stake — and we discard the ones that don't survive.

What it is not

It is not data-driven marketing. That phrase has been ruined. Today it means an agency bought a new dashboard. It does not mean anyone is willing to be wrong in public when the dashboard says something inconvenient.

It is not brand vs performance. That dichotomy is fake. There is marketing that pays for itself, and there is marketing that doesn't. Both can be brand-building. Only one is responsible.

It is not attribution theatre — last-click reports glued together with multi-touch dashboards that no one challenges. Real attribution is incremental: it asks "what would have happened without this spend?" and refuses to take credit it can't prove. (See: attribution theatre.)

Five tests we put every campaign through

We use these five questions to separate evidence-based marketing from its cosmetically-similar cousins. If a programme can't answer four of the five, we don't run it.

1. What revenue number is this designed to move? Not "awareness". A number with a currency symbol.
2. What's the counterfactual? What would have happened if we'd done nothing? If we can't model it, we can't take credit for the lift.
3. What's the attribution window? Seven days? Thirty? Ninety? Different windows tell different stories. Pick one before launch and live with it.
4. What's the kill criterion? If this doesn't hit X by week Y, we shut it down. Without a kill criterion, every campaign lives forever.
5. Who eats the loss if it fails? If the answer is "the client", we are not in an evidence-based engagement. We are in a billing arrangement.

How this connects to outcome-based marketing

Evidence-based marketing is the discipline. Outcome-based marketing is the commercial model. The two reinforce each other.

When the agency is paid only when the client grows, the agency cannot afford to fool itself with vanity metrics. It is forced into evidence. Conversely, when an agency operates on evidence, it can credibly offer outcome-based pricing — because it actually knows what its work produces. Most agencies can't, which is why most agencies still bill by the hour.

The supporting body of work

We've been writing about this for two decades. The beliefs and case studies below are the working evidence — every claim, every test, every result.