How we think about paid media.
The working version of our worldview: the same document we use to make decisions inside our clients' accounts. Written for founders who would rather understand the model than be sold to.
Brands that trust us
Most brands treat the wrong problem.
The usual story is that the platform stopped working. The reality is almost always elsewhere: creatives too alike to stand out, measurement crediting the wrong lever, a margin that cannot absorb the spend, an offer that is too weak, or an account built to spend rather than to learn.
A platform problem.
Meta and Google work. They are noisy and credit themselves generously, but they work. If their numbers move and your business does not, the problem is upstream.
A diagnosis problem.
The wrong constraint gets a confident answer. Budget goes up, creatives multiply, the account gets split. The result does not move, and "Meta is broken" enters the conversation.
The order of operations.
Diagnose first, sequence second, spend last. Almost every declining account we take over is a diagnosis that never happened, dressed up as activity.
Three diagnostics, on every account.
The same three, every time, in the same order. Each one breaks differently, each one needs a different fix.
Account structure
Is the split useful or excessive? Is the budget concentrated enough to give a clear read? Can the account grow without breaking?
The questionIs this account built to learn, or only to spend?
Creative
On Meta, the creative does the targeting. Volume is not enough: growth comes from real diversity of ideas across persona, angle, offer and format.
The questionDo we need a new idea, or an iteration?
Data integrity
We judge channels on new customers, not on the blended total. A report is there to change a decision, not to justify it.
The questionDo these numbers survive your P&L?
Margin to acquisition cost is our compass.
The gross margin from purchases in the first 30 days, divided by what the customer cost. This number matches your P&L, ignores the platforms' credit games, and tells you whether acquisition is healthy in absolute terms, not just better than last week.
Losing money
The first order does not pay for its acquisition. Usually a margin or offer problem, not an ads problem.
Under pressure
Marginally profitable. Pricing, creative or structure leave room: we find the constraint before raising the budget.
Healthy and scalable
The target zone for most brands. The goal is to stay there while spend climbs.
Under-scaled
Too efficient: money is being left on the table. The brand can absorb more spend without breaking its economics.
Acquisition before blended.
Revenue from existing customers blurs the read. We separate the two at the source and judge each on its own.
Click before view.
Sales counted after a mere view inflate Meta's credit. We keep them as a check, never as the optimisation signal.
Causality before credit.
Incrementality tests set the limit on what a platform can claim. When it disagrees with reality, reality wins.
Concepts move spend. Variations don't.
An ad concept is a combination of four choices. Change one and you have a new concept. Change a headline or a button colour and you have an iteration. Most accounts test iterations and call it creative testing. The performance gap lives at the concept level, not below it.
Persona
Who we are talking to. New buyer or loyal customer, frustrated or aspirational: each persona calls for its own angle, offer and format.
Angle
The argument. A pain, a mechanism, status, identity: what we want to convince people of before asking them to buy.
Offer
What people get and on what terms: bundle, trial, guarantee. The most neglected variable, and often the one that unlocks a new spend ceiling.
Format
Talking head, UGC, static, comparison, animation. Format is not a production decision, it is part of the concept.
Account structure
05 · Concentration over segmentationMost accounts are split so finely that nothing is readable any more. With too many ad sets, every decision is made on noise. This is not about exiting the learning phase: it is about giving the platform enough concentration to allocate budget well.
Signal behind every decision.
The account exists to teach the platform, and us, what works. We consolidate until every important decision rests on enough data to be confident.
Let the system allocate, then fence it.
Once fed, Meta and Google allocate better than manual settings. But they happily spend on demand you already have. We separate brand from the rest, and new customers from existing ones.
Segment only with a reason.
A market, a language, a product line with genuinely different economics: there, a split is justified. These are decisions, not defaults.
What guides every decision on your account.
The bottleneck changes with stage.
What takes a brand from 1 to 5 million is not what takes it from 5 to 20. Generic advice is almost always wrong by the time you apply it.
Most teams misdiagnose the problem.
A platform problem is often a creative, measurement, margin or structure problem in disguise. Time spent on the wrong problem is the most expensive line in the budget.
Good strategy reconciles with the P&L.
If the ad account looks good and the business is not improving, something is wrong. We look for the gap, not a justification.
Better data should drive better decisions.
Reporting that changes no behaviour is overhead. If a number does not move the meeting, it does not belong in it.
A structured review, built on our tooling.
Led by Martin Fournet, founder, with the system we run for our clients.
- 01Context callAbout 30 min
- 02Shopify, Meta, Google, TikTok analysis3 to 5 days
- 03Video walkthrough, in order of impact30 to 90 min
- 04Call with Martin60 min
Free · you keep the video · a written proposal afterwards, if you want one
Get my account reviewedLet's talk about your brand.
30 minutes to look at your numbers together. Free, no commitment.