Growth stops when the backlog runs dry.
The best performance campaign I ever ran hit every target and taught me nothing I wanted to know.
We cut the cost per acquisition, lifted the conversion rate, and squeezed the account until it squeaked. The client was happy. Then I stepped back and realised we had not grown the business. We had got better at catching people who were already coming. The demand was there before we showed up, and we just got better at catching it.
This is how performance marketing works, and it is why so many good marketers lose the plot. Performance is a harvesting machine. Point it at existing demand and it will bring that demand in cheaper and faster than anything else you own. What it cannot do, and never has, is make demand that was not there.
The machine only catches what is already moving
Search, retargeting and lookalikes all fish where the intent already sits. Performance needs demand it did not create.
Look at how a performance channel actually works. Search catches people already typing the words. Retargeting catches people who already visited. Even the cleverest audience targeting is fishing in a pond of people who already look like buyers. Every one of those tactics needs the demand to exist first for performance to have anything to convert. By the time performance touches a buyer, most of the deciding has already happened, somewhere the account cannot see.
Someone always raises paid social here. Cold prospecting on Meta or YouTube puts a product in front of people who were not looking, and some of them buy. Fair. That is the one place performance brushes against demand creation. But judge that cold campaign on cost per acquisition and the system hunts for the cheapest conversions, which are the people closest to buying anyway, and it defunds the creative that was doing the persuading. The machine is built to find intent, and it will find intent even in a room full of people who had none.
- A cold prospecting campaign is judged on cost per acquisition.
- The system hunts the cheapest conversions, the people closest to buying anyway.
- The creative that was doing the persuading is defunded.
Then the backlog runs dry
For a while this does not matter, because there is a backlog of existing demand to work through. The account grows, the graphs go up, and everyone assumes the machine is making the growth. Then the backlog runs dry. You have caught everyone who was already looking. And the only way to hold the numbers up is to bid harder for the same shrinking pool of in-market people, against competitors doing exactly the same thing.
Your cost per acquisition climbs, quarter after quarter, and you call it the auction getting more competitive, when you ran out of demand and never built the thing that makes more.
The machine is built to find intent, and it will find intent even in a room full of people who had none.
Creating demand is a different job
Making someone want the thing before they need it is brand work: slow, invisible on this quarter’s dashboard, and decided long before the search happens.
It is making someone want the thing before they need it. It is being the brand that comes to mind unprompted, so that when the need finally arrives, yours is the name they type and the click was decided months ago. That is brand work. It is slow, and it does not show up in this quarter’s numbers, which is why a performance-brained organisation refuses to do it.
Capturing demand
- Converts people already looking.
- Search, retargeting and lookalikes all start from existing intent.
- Fast, measurable, on this quarter's dashboard.
Creating demand
- Makes someone want the product before they need it.
- Yours is the name they type when the need arrives.
- Slow, and absent from this quarter's numbers.
What performance is for
Brand building goes quiet while the budget goes to capture. Strip out the demand you did not create and, for many brands, real growth is close to zero.
There is nothing wrong with using performance. I did it for a living. It is the most accountable, most measurable thing in marketing, and a brand that cannot capture demand efficiently leaves money on the table every day. None of this split is new, either. Binet and Field spent a decade showing that building a brand and capturing demand are different jobs. What those books rarely carry is the view from inside the machine that does the catching. The error is running performance on its own, calling that a strategy, and then blaming the algorithm when growth stalls.
The healthiest split lands near 60% brand, 40% activation.
Building a brand and capturing demand are different jobs. Across the IPA databank the healthiest split lands near 60% brand and 40% activation, varying by category. Most advertisers run closer to the reverse.
Estimated. Binet & Field, The Long and the Short of It (IPA, 2013): a modelled optimum from awarded campaigns, not an observed average · gabrielgruter.com
On my model, this is the equity line dropping off the dashboard while everything piles into activation. It is the most common shape in the market, and it is a slow-motion problem, because a brand running down its memory does not collapse. Each year the same growth costs a bit more, until the performance team is running the best account of their careers and the business is not moving.
Strip out the demand you did not create, the people who would have found you anyway, and look at what is left. This is measurable, by the way. It is called incrementality testing, and most of the accounts I have worked on never ran one, because the answer was going to be embarrassing. That number is your real growth, and for a lot of brands it is close to zero.
Strip out the demand you did not create, the people who would have found you anyway. What is left?
Notes & references
- Les Binet & Peter Field, The Long and the Short of It: Balancing Short and Long-Term Marketing Strategies, IPA, 2013.
- On incrementality and holdout testing: geo-experiment and conversion-lift studies (the Meta and Google experiment tooling), and the wider econometrics literature on the limits of last-click attribution.