The brand-versus-performance argument is unusually resistant to evidence, because the two sides are not measured on comparable terms. Performance advertising reports its results daily. Brand advertising reports its results slowly, indirectly, and in a form that is easy to dismiss.
So the argument gets settled by whoever has the more confident dashboard, which is not the same as whoever is right.
Why the usual framing is wrong
The debate is normally posed as a philosophical one: is advertising about building memory or about capturing demand? Both sides have a literature and neither will win, because the answer depends on the business.
A more useful framing: brand and performance are not two kinds of advertising, they are two constraints. Performance advertising converts existing demand. Brand advertising creates future demand. Which you need depends on which of those is currently scarce.
Which constraint are you under?
Three diagnostics, all readable from data you already have.
Does performance efficiency degrade as you scale?
Increase performance budget by a meaningful amount and watch acquisition cost.
If efficiency holds, you are capital-constrained: there is more profitable demand available than you are currently funding. Brand investment is not your bottleneck — more performance budget is, and the case to make internally is for funding, not for awareness.
If efficiency degrades sharply while conversion rates hold steady, you are demand-constrained: you have reached most of the people who already want the category, and the additional spend is buying progressively less interested audiences. This is the clearest signal that brand investment has a job to do.
How long is the purchase cycle?
The longer the gap between first awareness and purchase, the more of the decision happens outside any measurable window.
For an impulse purchase, being present at the moment of intent is most of the battle, and performance advertising does that well. For a considered purchase researched over weeks, the set of options a buyer considers is often formed before they search at all — and being in that set is a brand outcome that no attribution window will show you.
How much of your demand is branded?
Look at the proportion of your search traffic arriving on branded terms, and its trend.
Branded search is demand you have already created. A rising share suggests brand equity accumulating; a flat or falling share while spend rises suggests you are buying transactions without building anything that outlasts the budget.
It is an imperfect proxy — it is affected by competitors bidding on your name and by category growth — but it is available to everyone, and its trend over quarters is informative even when its level is not.
A decision table
| Situation | Likely constraint | Where the next unit of budget goes |
|---|---|---|
| Efficiency holds as spend rises | Capital | Performance, and make the case for more of it |
| Efficiency degrades, conversion rate steady | Demand | Brand, or new markets |
| Efficiency degrades, conversion rate falling | Targeting or offer | Neither — fix conversion first |
| Long cycle, low branded search share | Demand | Brand, patiently |
| Short cycle, high branded search share | Capital | Performance |
The third row is the one most often misdiagnosed. Falling conversion rates alongside rising costs usually means the offer or the page is the problem, and spending more on either kind of advertising will simply buy more of an unprofitable outcome.
Why attribution systematically undercounts brand
Because brand advertising's effect appears as an improvement in everything else.
When brand investment works, more people recognise you, so your performance ads get better click-through rates, your landing pages convert better, and your branded search volume rises. Attribution credits those improvements to the performance channels where they show up.
This is not a flaw that better attribution fixes — it is what attribution is for. It assigns credit to observed touchpoints, and brand effects are diffuse by nature. The consequence is that a budget allocated purely on attributed performance will underfund brand indefinitely, and the numbers will look excellent right up until growth stalls.
How to measure brand without pretending
Three methods, in ascending order of effort.
Branded search volume over time. Free, imperfect, and directionally useful. Watch the trend against brand spend, controlling for category growth as best you can.
Market-level holdouts. Run brand activity in one set of comparable markets and not in another, then compare total outcomes — not attributed ones. This is the most credible method available to a mid-sized advertiser, and it is the same design described in measuring incrementality on a small budget.
Marketing mix modelling. Once there is enough history and enough variation, a model can estimate brand contribution alongside everything else. It requires years of data, which makes it a method for businesses that already have a history rather than for settling this quarter's argument.
What about published ratios?
They exist, they are widely quoted, and they are derived largely from big consumer-goods advertisers with mature brands, long histories and patient capital.
That does not make them wrong. It makes them a poor template for a growing business with a short cash cycle, because the constraints are not the same. A business that cannot fund next quarter from this quarter's revenue is not in a position to make an investment that pays back over years, however well evidenced the payback is.
Read them as evidence that brand investment has real returns. Do not read them as an allocation to copy.
The practical answer for most businesses
Below a certain size, the separate brand budget is a distraction. The higher- return move is to make your performance advertising distinctive enough that it builds memory as a by-product — which is the same argument as testing fewer, more genuinely different creative concepts.
Distinctive performance creative does both jobs at once. Bland performance creative does one job expensively, because the auction charges for attention that the work fails to earn on its own.
Once efficiency starts degrading at scale while conversion holds — the demand-constrained signal — that is the moment a separate brand line earns its place, and the moment the argument stops being ideological and becomes arithmetic.