There have always been two kinds of organisation. The ones that treat brand as something worth paying for, and the ones that treat it as a cost to be worked around. That split is not new, it is not generational, and it has survived every change in the tools available to either side.
What changes is the argument. Each decade hands the second group a fresh reason to believe the work can be had for nothing, and each time it arrives looking like a revolution rather than what it is, which is the same position with new equipment.
Right now that argument is AI. Prompt a logo, generate a site, skip the studio. It is worth saying plainly that this is an old position, because the evidence on the other side of it has been accumulating for more than thirty years and almost none of it is contested.
1. The argument is older than the people having it
Desktop publishing was going to end design studios in the late 1980s, on the reasoning that anyone with a Mac and a laser printer could set type. Clip art libraries were going to do it again. Then stock photography, then template marketplaces, then crowdsourcing platforms where a logo cost ninety-nine dollars and came with forty rejected options, then offshore production at a tenth of the local rate.
Each of those genuinely changed what production costs. Not one of them changed what a brand is worth, because none of them were ever addressing the same question. Cheaper execution and better judgement are different purchases, and conflating them is the oldest error in this field.
The organisations that bought the cheap version in each of those waves were, in the main, the organisations that were already inclined to. The tool did not convert them. It gave them a justification they already wanted.
2. What has been measured since 1993
The strongest long-running evidence on brand value comes from the BrandAsset Valuator, now run as WPP BAV. The first study was fielded in 1993. It has since covered some 63,000 brands across 52 markets and more than a thousand categories, drawing on interviews with around two and a half million people. Kevin Lane Keller described it as the most extensive research programme on global branding to date, and it has been running continuously ever since.
BAV measures four pillars:
- Differentiation. Whether the brand stands apart from its competitors.
- Relevance. Whether it is personally appropriate and important to the customer.
- Esteem. The regard and respect people hold for it.
- Knowledge. How familiar people are with it.
A fifth pillar, energy, was added later to capture how innovative and dynamic a brand feels.
The interesting part is not the model. It is what happened when researchers connected it to financial results. A study published in the Journal of Marketing Research in 2008 matched eight waves of BAV data against 275 publicly traded single-brand firms. Relevance and energy had positive, significant effects on stock returns beyond what the accounting measures explained. More striking, firms that increased differentiation in one wave went on to record a mean risk-adjusted abnormal return of 0.059 in the following period, against minus 0.048 for firms whose differentiation fell.
The authors were careful, and so should we be. That is association rather than proven causation, and they say so. But it is association measured across hundreds of companies and more than a decade of data, which is a considerably higher standard of evidence than most business decisions are held to.
There is a detail in all this worth sitting with. BAV was built by Young & Rubicam, an agency that no longer exists. WPP folded Y&R into VMLY&R in 2018, then merged that into VML in October 2023, retiring a name that had been in the business for a hundred years. The agency is gone. The database is still running.
3. What the effectiveness data says about the split
The second body of evidence comes from the IPA Databank in the United Kingdom. In 2013 Les Binet and Peter Field published The Long and the Short of It, which found an optimum budget ratio of roughly 60:40 between brand building and short-term activation. Their 2018 follow-up, Effectiveness in Context, put the sweet spot at 62:38. The ratio has stayed remarkably stable across reports, and the original has been cited in more than seventy pieces of academic research and textbooks.
The finding that matters more is what they observed happening in the market. As the share of short-term, activation-led campaigns rose, the number of very large business effects recorded in IPA cases fell. The industry moved towards the measurable thing and got less effective while doing it.
This is the trap in miniature. Activation is easy to measure, so it gets the budget. Brand is slow to measure, so it gets cut. Then performance declines for reasons the reporting cannot see.
4. What the market pays for it
The third source is Kantar BrandZ, which reached its twentieth year of brand valuations in 2025 and now tracks some 14,000 brands through BrandSnapshot. Kantar reports that companies with the strongest brands have outperformed the market average consistently since 2006, and that high-equity brands have returned 88 per cent more than the S&P 500 and 251 per cent more than the broader MSCI World Index.
Three separate organisations, three different methods, three decades. A consumer survey programme, an advertising effectiveness databank, and a valuation model built for investors. They were not designed to agree with each other and they are not measuring the same thing. They point the same way regardless.
Add to that the Ehrenberg-Bass Institute, whose work since How Brands Grow in 2010 has argued that growth comes from mental and physical availability, from being easy to think of and easy to buy. It is often read as a challenge to brand thinking. It is closer to the opposite. Being easy to think of is what a distinctive brand is for.
5. Where AI actually sits in this
None of the above is an argument against AI, and it would be a strange position for a studio using these tools daily to take.
AI has genuinely collapsed the cost of production. Resizing, versioning, adapting a master idea across thirty formats, first-draft copy, exploratory visuals. That work used to absorb a great deal of time and most of it should not have. Anyone claiming otherwise has an interest in you believing it.
What AI does not do is decide what the brand should mean, which customers it is for, what the business should be known for that nobody else can claim, or whether the comfortable answer is the right one. Those are judgement calls made with incomplete information and commercial consequences. A model trained on what already exists is structurally poor at producing something that does not yet exist and should.
Generative tools are also, by construction, averaging machines. They produce the most probable output. Differentiation is the least probable one that still works. That is not a limitation to be engineered away in the next release. It is what the technology is.
Which is to say AI makes the execution cheaper and the differentiation harder, at exactly the moment the thirty years of evidence above says differentiation is the pillar that predicts financial performance.
6. Why the split persists
If the evidence is this settled, the obvious question is why the argument continues. Three reasons, none of them about the evidence.
- The payback is slow and the cut is immediate. Reducing brand spend improves this quarter and damages a year that somebody else will be accountable for. The incentive structure does the rest.
- The measurement is harder. A click is countable today. Pricing power is countable too, but almost nobody tracks it, so the easy number wins the meeting.
- Some people have simply never experienced it. If you have only ever bought design as decoration, you have no reference point for what it does when it is bought as judgement, and the argument sounds like special pleading from people with an interest in it.
That last one is worth sitting with, because it is the honest version of the sceptic’s case. Plenty of brand work is decoration sold at a premium. The objection is not unreasonable. It is just aimed at the wrong target.
7. What to do with this if you are deciding
For anyone weighing up what to spend, the research supports a fairly short list.
- Split the budget before you split the brief. Something close to 60:40 between long-term brand building and short-term activation, adjusted for your category.
- Use AI for the production, not the decision. Volume, versioning and adaptation inside the system. Positioning, architecture and what the brand stands for outside it.
- Measure the pillar that predicts performance. Differentiation is the one the BAV work connects to later financial results, and it is the first thing lost when execution is automated.
- Track your price premium. What you can charge against a comparable competitor is the most direct read on brand value available, and most organisations have never calculated it.
We have written separately on what to measure instead of vanity metrics if that last point is the live question.
The honest summary
The evidence that brand carries commercial value has been accumulating since 1993 across three independent bodies of research using three different methods. It is about as settled as anything in this field gets.
It has never persuaded everybody, and it was never going to. There have always been organisations that see brand as an asset and organisations that see it as an overhead, and that division has survived desktop publishing, stock libraries, crowdsourcing, offshoring and every other development that was going to settle it. AI will not settle it either. It will simply be the reason given for the next few years.
Which is fine. Those were never going to be the same two clients anyway.
The Animals is a brand design and digital experience studio in St Kilda, Melbourne. If you are making the case for brand investment inside your organisation, see how we approach brand strategy or get in touch.
