Introduction

Differentiation or distinctiveness? What 246 brands say

Should a brand stand for something different from its rivals to be chosen?

It is one of the most fundamental questions in business and marketing, yet it remains unresolved. Despite years of discussion, we still lack solid evidence and data to settle it.

The debate is far from over. At Cannes Lions in June 2026, two of marketing’s best-known professors, Mark Ritson and Byron Sharp, agreed on many things, but reached no agreement on differentiation.

For decades, marketing textbooks have taught that a brand must stand for something different from its rivals to be chosen, an idea that runs from Reeves’s unique selling proposition through Kotler’s Marketing Management to Ries and Trout’s Positioning. Today, that belief is openly contested.

Ritson defends a more nuanced version of the textbook view, which he calls “relative differentiation.” Brands may not be unique, he argues, but they can offer “more of something” than their rivals, and although this kind of advantage is rare, brands should build and protect it. In his words: “My bullshit stat has always been [mental availability is] 70 to 80% of the job. I still maintain that there should be a quotient on top of that which is about differentiation.”¹

Sharp concedes that differentiation might matter in some situations, but says the bigger story is “how shockingly low it is or how little it matters.”² For Sharp, growth comes from distinctiveness, not from a unique position in the market: recognizable colors, fonts, mascots, jingles, and taglines. Communicated widely enough, these assets build the mental availability that turns into buying.

The question divides marketers between two very different growth strategies, and after years of debate, it still rests on claims like “some situations” and “shockingly low.” It deserves evidence and numbers. So we studied 246 brands across five markets, surveying 60,000 respondents, to find out whether being different from your rivals helps a brand grow.

The short answer: Ritson is right. Differentiation matters, and building mental availability around some needs can be 2.5 times easier than around others. For a challenger brand with limited resources, choosing the wrong proposition could be a costly mistake, one it may never recover from.

The longer answer comes down to three principles we uncovered in the data. They determine whether linking a brand to a particular need, whether a buying occasion, a product function, an emotional benefit, or a symbolic one such as status, can put it on a path to growth. They also show why neither pure distinctiveness nor some classic differentiation advice is a reliable route to growth.

1. Byron Sharp vs Mark Ritson – 5 big marketing truths we agree on. Uncensored CMO with Jon Evans, 7 July 2026↩
2. Andrew Tindall: What Ritson and Sharp got right (and wrong) at Cannes – The Drum, 22 June 2026↩

Most brands have clear strengths and weaknesses beyond their size

Big brands get linked to almost every need, simply because more people buy them.

Ask American cola buyers whether Coca-Cola tastes great, and 75% say yes. For Virgil’s Cola, it’s 6%. That isn’t a verdict on flavor. Every cola we measured, from Pepsi and RC Cola to Shasta and Zevia, sits where its size predicts.

But only about a third of the needs we measured are this closely size-driven. For the rest, size still matters, but much less. As we’ll see, these are the needs where challengers have a chance to stand out.

Differentiation can look “shockingly low” because, overall, size explains 90% of all brand associations in our study. But that 90% mostly explains why markets change so little: once you are big, being big helps you stay big. It says little about how brands grew into their current positions. The remaining 10%, the mental availability that size cannot explain, is where a brand can move beyond its current position and grow.

If you take size out, across 246 brands, most brands in every category have clear, distinct strengths and weaknesses. Three in four have both at the same time.

Behavio (2026)

Three properties of each need shape a brand’s chance to break through

We measured 775 needs across all 4Ps: buying occasions, product functions, emotional benefits, and symbolic ones such as status or country of origin, as well as price, place, and promotion. Then we examined how often brands manage to break through on each. The differences are striking.

Every need has three characteristics that shift a brand’s chance of breaking through on it, and their weight depends on brand size.

Behavio (2026)

Occupancy and prevalence are intuitive concepts, and neither is new to marketers. What we add is how much each matters for brands of different sizes.

Movability is less intuitive, but for challengers it matters more than occupancy and prevalence. They break through almost only on needs where the fight is still open. Where the fight is settled and size predicts everything, they almost never do.

Movable needs are where brand size matters less, and competitors actively win or lose ground. Take natural food in US dog food. One challenger scores 13 points above what its size would predict, while the market leader scores 16 points below. In other words, a brand bought by far fewer people is associated with natural food just as often as the market leader. Other brands are winning and losing ground on the same need.

Size decides great taste in colas, not natural food in dog food

Behavio (2026)

About half of all needs have medium movability, meaning competitors score, on average, 3 to under 6 points away from what their size predicts. Another 14% are highly movable, with competitors at least 6 points away, and some brands as much as 25 points above or below their predicted scores. Examples include environmentally friendly cars, installation services and expert advice in electronics, sports on streaming services, natural ingredients in cola, sensitive digestion in dog food, and immune support in supplements. Many of these are product functions.

Needs with medium, and especially high, movability are where big brands lose the usual advantage of their size, and the odds turn in favor of smaller competitors. On highly movable needs, big brands lose 2.6 times as often as they win.

Small and mid-sized brands break through 2.5 times as often on highly movable needs as on low-movability ones. For mid-sized brands alone, it is 2.8 times. No small brand founded in the last 30 years has broken through on a low-movability need.

This is where brands differ most, and where David can beat Goliath.

On highly movable needs, David can beat Goliath

Behavio (2026)

And these findings aren’t circular. We leave the brand itself out of the calculation and look only at its competitors’ positions: a brand’s own position cannot make the ground look more movable.

So one number answers two questions: movability measures how much competitors already differ on this need, but it also says how likely your brand is to break free from the position its size predicts. For small and mid-sized brands alike, the answers go hand in hand. That’s why we call such a need movable, not just moving.

The playbook that follows is simple: challengers should not look for a quiet gap where no brand is winning or losing. They should pick one of the big, active fights in their category and try to win it. Winning a fight is hard, but still easier than trying to win on dead ground.

Low-movability needs, where size predicts nearly everything, rarely reward challengers

One third of needs have low movability, and size predicts nearly everything. It’s tempting to assume that these needs are simply needs no brand has tried to claim yet. They may be unclaimed, but surely they’re not impossible to claim?

Yet many are exactly the kinds of associations that popular marketing advice has urged brands to promote for decades: deeper emotional benefits and buying occasions. Others are product needs where the main battle is already settled, and every credible brand in the category is expected to deliver. Think great taste in US colas, a morning energy boost in coffee, to refresh myself in cola, clear terms and conditions and feeling secure in insurance, and feeling like a good owner in dog food.

Category entry points tend to have low movability more often than other types of needs. None of the emotional needs we measured has high movability. Status, origin, and sustainability behave differently. We count them as symbolic needs, because they signal something to other people, and a third of them are highly movable. The same is true of CEP-like needs in cola and coffee: not a single one has high movability.

Behavio (2026)

We have seen that small brands cannot move low-movability ground. Big brands occasionally can, and so can mid-sized brands, although these are more often old brands, not new challengers. On a moment of calm and relax, only two brands hold a +6 earned strength across all markets: Douwe Egberts, a big brand, and Simon Lévelt, a Dutch coffee specialist founded in 1826.

A need with no fight in it is a trap for challengers, unless you have decades to spare

Breakthrough potential: the best need depends on brand size

Movability, occupancy, and prevalence all shape a brand’s chances of breaking through on a particular need. But the three rarely point to the same opportunity. Only 3% of brands have a silver bullet: at least one need that ranks first on all three dimensions. Every other brand must trade one strength against another.

Looking at how often small, mid-sized, and big brands break through on each dimension reveals which question each should ask first.

Small brands should prioritize highly movable needs, especially those with low occupancy. Prevalence matters less. A mainstream need would be ideal because it could give a brand room to grow from small to large around a single proposition. But less prevalent needs can still be a good choice, as long as they are highly movable and low in occupancy.

Mid-sized brands should follow the same order: prioritize movability first, then occupancy, and finally whether a need has medium or high prevalence. The one exception is low-prevalence, niche needs. To grow into a big brand, a mid-sized brand needs to win buyers it doesn’t reach yet. Niche needs simply don’t reach enough people, so they are best left aside.

For big brands, the question is less whether they can break through, and more where a breakthrough is worth the most. This flips the order of priorities: prevalence comes first. To keep growing, a market leader needs to reach buyers it doesn’t already have, and only mainstream needs offer enough of them. Occupancy comes next, with free needs first. On mainstream needs, big brands do better on low-occupancy needs than on occupied ones. Movability comes last because it involves a trade-off. A movable need offers a greater chance of winning, but also a greater risk of losing. So movability is a tiebreaker, not a data-driven priority. Whether to pursue it is a judgment call based on whether the brand needs to defend its position or win new ground.

The order of the three questions, by brand size

Behavio (2026)

Markets and categories differ widely in the potential they offer a brand. In US dog food, a small or mid-sized brand can target a need with potential #2 out of 27 combinations of movability, occupancy, and prevalence: sensitive digestion. It has high movability, low occupancy, and medium prevalence. The fight is still on, and the winners are yet to emerge.

In Dutch dog food, that fight is already settled. Sensitive digestion and natural food, the biggest battlegrounds in the US and UK, have low movability and are largely occupied. The best remaining option is #12, Netherlands produce. But it is a much harder starting point, and unlikely to support the rise of another major brand.

How growth hides the differentiation behind it

What if brands grow by building associations that their size alone cannot explain? Yet 90% of what people say about brands comes down to size. How could both be true?

Imagine launching a US dog food brand built for dogs with sensitive stomachs, cooked in facilities that make food for humans. A very promising proposition: sensitive digestion has high movability, isn’t yet fully occupied by existing players, and matters to enough dog owners to support at least a mid-sized brand.

You’re still a small brand, bought by just 2% of dog owners. You launch a great TV campaign, and 12% of dog owners now associate your brand with sensitive digestion. Only 3 of those 12 points are size-granted, what any brand of your size would typically get. The other 9 you truly earned. You’ve built a genuine point of differentiation.

Then the campaign ends. Most people who now associate your brand with sensitive digestion still don’t buy it. But over the following months, some give it a try. They may not switch to you entirely, but your brand makes their shortlist whenever their dog’s stomach acts up. Gradually, your buyer share grows from 2% to 5%.

Since the campaign ended, the number of people who associate your brand with sensitive digestion has stayed the same: 12 in 100 dog owners. But the benchmark has changed. A brand of your new size would typically be associated with sensitive digestion by 7% of dog owners. Other brands may never have actively built that association. It came naturally with the size they achieved by winning on other needs. You grew by winning on sensitive digestion. Yet from a brand research perspective, most of your original advantage has now disappeared. Of the 12% who associate you with sensitive digestion, only 5 percentage points still count as a distinctive strength.

As your brand grows, it gains associations beyond sensitive digestion. New buyers discover what your food is like when they feed it to their dogs. Their friends notice the bag, and your products appear on more marketplaces. With greater visibility and more buyers, people begin to associate your brand with the qualities and occasions they expect from any established dog food brand. Over time, they start buying you for reasons that have little to do with your original proposition.

Over time, you run more campaigns promoting sensitive digestion. Each one strengthens your lead, but each new wave of growth absorbs part of that advantage into your brand’s size. The cycle repeats until you’re so successful that everyone who cares about sensitive digestion already associates it with your brand.

But not every dog has a sensitive stomach. To reach the mainstream, you need to win on the needs that matter to the broader category. A decade or two later, you might be the market leader.

By the time a market researcher studies your brand, they may see a “shockingly” undifferentiated player, with associations on the category’s biggest, most established needs largely explained by its size. Even on sensitive digestion, the need that drove much of your growth, your brand may now score exactly as expected for its size. Your success has absorbed most of your original advantage.

Differentiation, then, is a growth path that can eventually erase the very advantage that started it.

Growth hides the differentiation that drove it

Behavio (2026)

Fortunately, our example market hasn’t reached that point yet. Sensitive digestion and natural food are still battlegrounds. Many challengers built around healthy, natural ingredients, most founded in the past two decades, have grown into mid-sized brands, and Blue Buffalo into a big one. They continue to promote their propositions through major TV campaigns, and their differentiation remains clearly visible in our data.

On natural food, Freshpet and The Farmer’s Dog in the US, and Lily’s Kitchen and Butternut Box in the UK, score up to 14 percentage points above what their size would predict. The market leaders, Purina in the US and Pedigree in the UK, have fallen behind, scoring up to 16 points below expectations.

Breakthrough potential passes an out-of-sample test

Small and mid-sized brands still retain much of the differentiation they have earned, so they are suitable for testing breakthrough potential. The strongest test is out of sample: can it predict results using data it has never seen?

In a hidden-market test, we used four markets to find the best order for assessing movability, occupancy, and prevalence, then tested it on the fifth. We repeated this five times, using each market as the hidden one in turn. For mid-sized brands, the same order won every time. Small brands had too few breakthroughs to establish a reliable order.

In a hidden-brand test, we took each of the 140 small and mid-sized brands out of the data in turn, then recalculated breakthrough potential without it. Needs in the top third broke through 2.8 times as often as those in the bottom third for small brands, and 2.1 times as often for mid-sized brands.

Undifferentiated challengers fall short on both awareness and conversion

But we didn’t just look for evidence that differentiation drives growth. We also tried to disprove it by looking for brands that grew through pure distinctiveness: brands that became easy to recognize and recall, without standing for anything specific. If that playbook worked, standing out on a particular need would have no bearing on how well-known a brand was or how often people bought it.

But our findings point in the opposite direction: among small and mid-sized brands, differentiated ones are better known, and they convert more of that awareness into buyers.

The pattern holds beyond this study, too. In our brand trackers with three or more waves, growing brands improve, on average, on 1.6 of every five associations tracked.

Behavio (2026)

Overall, we found little evidence of brands growing through pure distinctiveness: being much better known than their size would predict, with distinctive brand assets and standout advertising, but no clear strengths in their product or the other Ps. Of the 246 brands we studied, only two came close. Both were small, and neither was growing.

The other obvious candidates, brands far more famous and distinctive than their size would predict, are also strongly differentiated. Tesla is the clearest case: far above its size on distinctive brand assets and standout advertising, and on product needs such as environmentally friendly too.

Converted differentiation: from one need to broad availability

None of this proves that a brand can’t grow on fame alone. It suggests that doing so is very ineffective. Fame and distinctive brand assets seem to work much better when they support a clear lead on a specific need, chosen based on movability, occupancy, and prevalence, in the order that makes sense for your brand’s size.

Nor is this a step back to classic differentiation. The difference does not have to be unique, and once you are big, it is no longer the main reason people choose you.

This growth path is perhaps best described as converted differentiation. You need a difference to grow efficiently, and over time you cash it into broad mental availability.

Everything else in this paper adds detail, examples, and numbers to what you have just read.

Chapter 1

Once size is removed, rival brands differ clearly in buyers’ minds

Sharp argues that most of what appears in a brand image table comes down to size: big brands have more buyers, and those buyers tend to associate them with more things. As he writes in How Brands Grow, “Once the usage effect is removed, there is little sign of brands having unique images.”

Our data tells a more nuanced story. In this chapter, we show that although 90% of brand associations can be explained by size, differentiation is neither rare nor “shockingly low.” Most brands have clear strengths and weaknesses, and these often differ substantially from those of their competitors.

1.1 Most associations come with size; the rest are earned

Does Coca-Cola taste great? 75% of American cola buyers think so. Only 6% say the same about Virgil’s Cola. It is hard not to read that as a verdict on flavor.

Great taste in US colas is granted by size

Behavio (2026)

The likely explanation lies in how people answer. When a need feels typical of the category, and a brand feels typical of the category, people answer yes even without a concrete memory link. This happens in surveys, and probably at the shelf, too. We call these associations size-granted: they come with being big.

The chart shows that the number of buyers almost perfectly predicts how often people associate a brand with a need. But those making the associations are not just buyers. Many are non-buyers, especially for small and mid-sized brands. Even among non-buyers, however, associations tend to follow the number of people who buy the brand.

Non-buyers pick up clues about a brand’s size from seeing it on shelves, on marketplaces, and in their friends’ hands and homes. In other words, it is not just fame that shapes these associations, but people’s sense of how big a brand is. Awareness alone predicts size-related associations less accurately than buyer numbers do.

Tesla illustrates the point. About 80 in 100 car buyers know the brand, yet on typical car needs, its associations are no stronger than its relatively small buyer base would predict. A famous name is not enough.

Non-buyers also explain why the line is not straight but bends. Each new buyer lifts a small brand’s score more than a big brand’s. Economists call this diminishing returns.

When a small brand wins its first buyers, many more people start hearing about it and associating it with the category’s typical needs. When a big brand wins another buyer, that person likely already associated it with those needs before buying.

There is also a ceiling. No brand is linked to a need by everyone. Even Coca-Cola reaches only 75%.

For great taste in US colas, size explains the entire score. This does not mean that no one genuinely loves the taste of Coke, but all such people sit among the many who would have said yes anyway, so their love does not move the score. No cola owns a real position on taste.

Now, what does it look like when size alone can’t explain the score? Let’s look at US electronics stores, and whether people link them to expert advice.

Expert advice in US electronics is not explained by size

Behavio (2026)

Best Buy is bought by fewer electronics buyers than Amazon. Yet 64% link Best Buy to expert advice, against 37% for Amazon. On the cola chart the bigger brand always scored higher; here a smaller store scores 27 points more.

Even for Best Buy, 50 points of its score come with its size. The other 14 are truly earned: links Best Buy has built for itself. In people, that’s roughly 13.7 million US electronics buyers more than a store of its size would typically get.

Earned positions can also be negative. Some people who would agree based on brand size actively disagree. For Amazon, that deviation is 21 points: roughly 21.1 million US electronics buyers who don’t see it as a place for expert advice.

What respondents likely think when they click yes or no

So if your brand tracker shows only raw scores, you are mostly looking at your market share with dozens of needs written along the top. That’s a costly way to measure market share.

1.2 Every category, and most brands, hold earned positions

On any given need, a typical brand sits 4.2 points from where its size would put it. Most of that is noise: with a fresh sample of 1,000 people, scores would shift by a point or so, like a bathroom scale that never quite agrees with itself. But the big gaps are far too common to be down to chance:

And these gaps rarely come alone. Within a typical brand, its strongest and weakest needs sit 17 points apart. Three brands in four have both a clear strength and a clear weakness: a need people associate with them more than their size would predict, and one they associate with them less. Specifically, they score more than 3 points above their line on one need and more than 3 points below it on another. We call these earned strengths and earned weaknesses.

A point looks small on a slide, but every point is a share of the people who buy the category. A modest earned position is a crowd the size of a city; bigger ones on bigger markets are dozens of millions of people.

Large earned positions, in number of category buyers

Behavio (2026)

Some categories leave more room for brands to stand apart than others. Electronics and cars are at the loud end, with the typical brand sitting about 5 points off its line. Supplements and insurance are at the quiet end, at about 3.5 points. This difference is not down to electronics having bigger and smaller brands. How unequal brands are in size has virtually no relationship with how far they stray from their lines (a correlation of −0.05, as close to no link as the data gets).

How much brands differ, category by category

Behavio (2026)

Here is what this looks like within a single category: American cars, with brand size taken out of the equation. Each cell shows the share of car buyers who link a brand to a need beyond what its size alone would predict. A blank cell means size explains the association entirely.

Cars: what a strongly differentiated category looks like

Behavio (2026)

Cadillac owns status, but pays for it on price. Tesla owns technology and green credentials, Toyota owns value, and Ford owns the showroom. Chevrolet, the biggest of the six, owns almost nothing of its own. Where it stands out, it is by falling short: on status, design, and the environment.

Cars are easy to tell apart. People argue about them, and carmakers spend fortunes trying to stand out. Coffee, by contrast, sits on a supermarket shelf and gets bought half asleep. UK coffee is one of the flattest categories we measured.

Coffee: even the flattest category still has plenty of earned positions

Behavio (2026)

The patches are smaller, but each brand still has one: Tesco Finest on price, Douwe Egberts on the supermarket shelf and the comfort of the cup, Taylors of Harrogate on feeling special. Grind is the brand nobody links to the shelf.

1.3 Competing brands rarely share the same strongest association

As the two grids show, each brand stands out on different needs. The same pattern holds across the full study.

Competitors rarely share the same strongest association. For 70% of brands, no rival has the same top strength. And half of all strengths of 6 points or more belong to just one brand.

The pattern holds for product needs, too. In a typical category of six brands, four have a strongest product association that no rival shares. Only two share their top strength with another brand.

Brands also stand apart across the full marketing mix. Roughly a third have their biggest strength in price, a quarter in promotion, a quarter in product, and a fifth in place. In every market we studied, competitors’ strongest associations spanned at least three of the four Ps.

1.4 Non-buyers see earned positions too

Before claiming that earned positions matter for growth, we need to address one objection. These differences could simply reflect what customers say about brands they already buy. People who subscribe to Viaplay know what is on it, while those who don’t might see all streaming services as much the same. If so, earned positions would be a record of past purchases, not a route to new ones.

Yet people who don’t buy a brand account for 44% of all the links made to it. For brands bought by fewer than 20% of category buyers, it is 69%, and for earned positions above +10, 54%. In the Netherlands, 46% of streaming buyers link Viaplay to watching sports, yet only 17% subscribe. 36 points of the 46 come from people who have never paid for it.

We also compared non-buyers with the whole sample. Across 1,482 brand–need pairs, the share of non-buyers who link a brand to a need tracks the share of all buyers in the category with a correlation of 0.92. On the 179 pairs where a brand sits more than 6 points above its line, the correlation is still 0.89.

Chapter 2

Differentiation matters for growth

Size explains 90% of brand associations, which helps explain why market shares are so stable. Bigger brands come to mind across more of the category’s buying triggers, making them more likely to stay big.

In this chapter, we’ll show why the remaining 10%, the earned associations, matters for growth. It’s where brands can move beyond the position their size would normally predict and find room to grow.

But first, we test whether a brand can gain size-granted associations without earned ones, simply by becoming famous enough to appear bigger than it is.

2.1 Brands that grow on fame and distinctiveness alone are hard to find

How Brands Grow opens by telling anyone who believes that “differentiating our brand is a vital marketing task” that they are “operating under many false assumptions.” Instead, Sharp advises marketers to build mental and physical availability. Mental availability comes from “sophisticated mass marketing,” supported by distinctive brand assets such as logos, fonts, and colors, that help people recognize and remember the brand across a wide range of category entry points (CEPs). In this view, a brand’s links to CEPs come mostly with its fame, not with which CEPs it chose to advertise. That leaves brands with considerable freedom in which CEPs they communicate. Over time, they should rotate through the category’s most mainstream CEPs and potentially address those where they lag behind.

If this playbook worked, standing out on a particular need would have no bearing on how well-known a brand was or how often people bought it. But, as we’ve already shown, our findings suggest the opposite. Let’s examine the evidence in more detail.

• Differentiated brands are better known. Small and mid-sized brands with an earned strength above +6 on a product need are known by 0.6 points more people than their size predicts. Comparable brands without such an earned strength are known by 3.3 points fewer. That is a gap of about 4 points.³ The pattern holds across all eight categories.

Differentiated brands are better known

Behavio (2026)

• Differentiated brands convert more awareness into buyers. Brands with an earned strength among people who know them but don’t buy them turn more of their awareness into buyers. Their conversion is about 8 percentage points higher than that of equally well-known brands without one: roughly 17% more buyers, and over 20% more for small and mid-sized brands.⁴

• Growing brands strengthen specific associations, not their entire image at once. Beyond this study, our brand trackers show a similar pattern. Across trackers with three or more waves, growing brands improve, on average, on just 1.6 of every five associations measured.

Overall, we found little evidence that brands grow through pure distinctiveness alone. By this, we mean being far better known than their size would predict, thanks to distinctive brand assets and standout advertising, but without clear strengths in the product or other Ps. Of the 246 brands we studied, only two came close to fitting this pattern: Suri in Czech insurance and Davidoff in Dutch coffee. Both were small, and neither was growing.

Other obvious candidates are brands that are far more famous and distinctive than their size would predict. But they also tend to have clear strengths on product needs.

Take Tesla. Most car buyers know the brand in every market, yet few of them buy it. Tesla also scores far above expectations on distinctive brand assets and standout advertising. But it stands out on product needs, too, scoring up to 18 points above expectations for environmentally friendly, and up to 20 points on packed with features and a sense of status. On many other typical car needs, however, it scores below expectations.

3. 65 small and mid-sized brands with an earned strength above +6 on a product need and 75 without. The gap is positive in all eight categories (1 to 9 points) and in 23 of 29 market-by-category comparisons. p < 0.001. Earned strengths measured with each brand left out of its own line. ↩
4. 150 brands in Czechia, the Netherlands, and the UK (128 with an earned strength, 22 without), the markets where buyers and non-buyers can be told apart. An earned strength here means scoring more than 6 points above the brands’ average on at least one need, among people who know the brand but don’t buy it. Expected conversion comes from a curve of penetration against awareness, with one slope for all 40 categories and a level for each. p < 0.001. ↩

2.2 Typical attribute studies weigh prevalence and occupancy but miss movability

Let’s look more closely at how differentiation can drive growth. Return to our dog food founder in the early days, deciding which proposition gives the brand the best chance to grow.

What a typical attribute study reports: links vs importance

Behavio (2026)

A differentiation-school strategist could choose exactly for my dog based on this simple reading of the attribute table. Or they could choose low price, the second-widest gap, and follow the classic low-price strategy. In practice, more sophisticated approaches use the data to identify which needs cluster within different segments, how much those buyers spend, and which audiences the brand can realistically reach. Those factors may point to a different need.

But the underlying logic is the same: how much a need matters, and how much room there is for a brand to claim it. Both are important. They reflect two of our three growth principles: prevalence and occupancy.

Yet a third characteristic matters even more. Some needs, like exactly for my dog, may look promising on these two dimensions, but no matter how well you target them, you may never build distinctive associations around them. We call this principle the movability of a need.

2.3 Movability: challengers win where rivals already differ

Movability asks a simple question: how far do competitors sit from the position their size would predict? But this isn’t just about which brands have chosen to stand out. It reflects something inherent to the need itself: how much room it gives any brand to build an earned position. Of the three dimensions, movability is the strongest predictor of where small and mid-sized brands can hold earned strengths.

How is movability measured?

How far, on average, the other brands in the category sit from the score their size predicts. The brand itself is left out, so its own result cannot make the need look more movable.

Consider these pairs of needs. Imagine they have the same occupancy and prevalence. Where would a new player find it easier to compete: where brands have already moved beyond what their size predicts, building clear strengths and weaknesses, or where no brand has managed to break away from what size alone would predict?

Low movability vs high movability

Behavio (2026)

At first glance, the left-hand need seems more tempting. No brand stands out on it, so the first to do so might claim the territory for itself. An untouched island.

But as we argued in the introduction, needs dominated by size are not untouched islands waiting to be claimed. They are settled ground where brands struggle to stand apart. How can we tell?

Where competitors already differ, challengers hold far more strengths.

Small and mid-sized brands are 2.5 times as likely to hold a +6 earned strength on a high-movability need as on a low-movability one; for mid-sized brands alone, it is 2.8 times. Small brands on their own hold such strengths on low-movability needs too rarely to give a reliable ratio.

Highly movable product needs account for just 14% of all brand–need pairs, yet they represent 36% of major breakthroughs, where brands score more than 10 points above expectations.

This could be circular: a brand’s own distinctiveness might make a need look more differentiated. Real estate agents know a simple test. Hide a house’s price. If you can still guess it from the street it stands on, location matters. If not, it is all about the house.

We applied the same logic to brands. We set each brand aside in turn, excluding it entirely from the calculation. Using only its competitors, we measured how much brands differed from what their size would predict, for each need in the category.

We then compared two needs for the brand we had set aside: one where it broke through, scoring at least 6 points above the norm for its size, and one where it did not. Because both needs belonged to the same brand, differences in its size, budget, or overall strength could not explain the result.

In 62 out of 100 pairs, the need where competitors differed more was also the one where the brand broke through. By chance, we would expect around 50, with a maximum of 52.

That is why we call it movability. Whether a brand can move on a need depends not just on the brand itself, but also on the ground it is trying to claim. Low-movability ground is not impossible to move, but it rarely happens.

Leaders are more likely to lose ground there, while challengers gain it.

On highly movable needs, big brands lose ground 2.6 times as often as they gain it. Of 145 such brand–need pairs, 53 sit more than 6 points below what their size predicts and only 20 more than 6 points above. Small and mid-sized brands, on the other hand, have 1.25 times more wins than losses on highly movable needs.

Small brands almost never break through on low-movability needs.

On low-movability needs, not one small brand founded in the past 30 years broke through. Only two small brands did: Film1, a brand with roots in the 1980s, and Okay Electronics, founded in 1994 and recently closed.

Big and mid-sized brands occasionally move low-movability ground, but far less often than movable ground. And the mid-sized brands that do it are more often old: 9 of the 17 were founded before 1950.

What follows? Small and mid-sized challengers should prioritize movable needs. For a big brand, movable ground is a bet. On dead ground of low-movability needs, its size-granted position is well protected. On movable ground, that protection disappears, and rivals can take whatever it has not built for itself. But the brand can also pull ahead there.

2.4 Core category emotions and basic occasions tend to have low movability

Across the study, 36% of brand–need pairs have low movability: brands sit, on average, less than 3 points from what their size predicts.

These “dead” needs follow a pattern. Many are basic category occasions and functions: morning energy boost in coffee, to refresh myself in cola, or clear terms and conditions in insurance.

Others are deeper emotional needs that are central to the category: feeling like a good owner in dog food, feeling secure in insurance, to feel alive in cola, or to feel confident in supplements.

Across the study, brands hold strengths on emotional needs in 9.2% of cases, against 12.9% expected by chance.

Why do these needs follow size so closely? Not because brands have ignored them. Category functions have featured in advertising for decades, and brands have spent just as long trying to own deeper emotional needs.

Taste in cola sparked one of marketing’s most famous battles. In 1975, the Pepsi Challenge pitted Pepsi against Coca-Cola on taste, winning over some Coca-Cola drinkers. In 1985, Coca-Cola hit back by changing its recipe to launch New Coke, only to bring back the original formula less than three months later.

Two of the world’s biggest advertisers fought over taste for a decade, yet neither managed to own it. Today, every American cola brand sits within about 3 points of its predicted score for great taste, right where its size would put it.

It seems to come down to three factors:

1. The fight is settled. Brands have pushed the same need for years, and each brand’s gains have been matched by its rivals’.

2. The need is too typical of the category. Buyers see it as a basic expectation, not something that sets one brand apart from the rest.

3. Emotional needs need concrete support. In the rare cases when an emotional need that usually has low movability becomes movable in one market, the brand usually stands out on more concrete needs, too. Of the 20 brands that break through on an emotional need, 18 also break through on a concrete product need. In the UK, Butternut Box stands out for making people feel like good pet owners, as well as for natural food, high meat content, and sensitive digestion.

What kinds of needs are most likely to be movable? Some are symbolic, such as status, design, environmental friendliness, and country of origin. Unlike emotional needs, these can signal something to other people: a third of symbolic brand–need pairs have high movability, against none of the emotional ones. Others are functional, such as installation services, expert advice, kids’ profiles, and being gentle on the stomach.

Movability is not limited to product needs. Price, place, and promotion cues can be highly movable, too.

2.5 Occupancy: free needs offer room that occupied needs do not

Occupancy asks whether the market is meeting what buyers want. Some needs matter to many buyers, yet few brands are associated with them. Others are overcrowded, with brands claiming more associations than buyer demand would justify.

How is occupancy measured?

Occupancy compares two things: how strongly brands are linked to a need, and how many buyers say it matters. Where brands are linked more than the need’s prevalence suggests, the need is occupied. Where they are linked less, it is free. A brand’s needs are ranked by how much room they leave and split into thirds. The detailed calculation is in the appendix.

Imagine you are a manager at Apple TV, which reaches 23% of streaming buyers in the UK. You need to decide whether to buy sports rights or kids’ content rights to attract subscribers. Both needs are movable, and both matter to a similar number of people. But there is another factor: kids’ content looks like a more crowded space, with more competition from other brands.

Sport is underoccupied in UK streaming; kids’ profiles are occupied

Behavio (2026)

A typical service reaching one-fifth of buyers is associated with kids’ profiles by 29% of buyers, and with sport by 23%, even though more buyers care about sport. Sport is underoccupied: buyers want more of it than services are known for. Kids’ profiles, by contrast, are already occupied. Every service in the category falls on the same side.

Free ground is where the market has yet to catch up with buyer demand. A brand that moves into this space can meet demand that is already there, without competing head-to-head with established players. NOW, a service the same size as Apple TV, sits 12 points above its size-predicted position on sport.

Small and mid-sized brands break through more often on low-occupancy needs, but the effect is smaller than for movability: 1.2 times for small and 1.5 times for mid. For big brands, the better ground depends on how prevalent the need is. Overall, they do better on high-occupancy needs, mostly on the less prevalent ones. On mainstream needs, low-occupancy needs give them more wins per loss (2.0 against 1.4).

2.6 Prevalence: a need’s reach matters most for big brands

Prevalence is the share of category buyers who say a need matters when choosing. It tells you how much a breakthrough on that need is worth. It is also the only one of the three dimensions recognized by the Ehrenberg-Bass school, which ranks category entry points by how many buyers each one reaches.

How is prevalence measured?

We ask category buyers whether a need matters when they choose. Prevalence is the share who say yes.

Prevalence matters most to big brands. A leader already sells to most of the category, so it should focus first on the needs that matter to the most buyers, turning to niche needs only after that. MediaMarkt reaches 63% of Dutch electronics buyers. High-quality products matter to 83% of buyers, while installation services matter to 45%. Both needs are highly movable, and MediaMarkt sits on its line on both. If installation became its main story, MediaMarkt would be building its brand around a need that matters to fewer people than already shop there. It would be narrowing its appeal to that of a specialist. The specialists are already there. Expert, which reaches 18% of buyers, sits 24 points above its line on installation, while Coolblue sits 16 points above. For Expert, the same need is two and a half times its size.

Prevalence does not change the odds of holding a strength, but big brands are more likely to fall behind on niche needs.

For small brands, prevalence matters least. Even the least common need in our study matters to 24% of buyers, while a small brand reaches fewer than 10%. Every need is bigger than the brand itself, so small brands hold earned strengths on niche needs about as often as they do on mainstream ones.

2.7 Breakthrough potential: small, mid-sized, and big brands should ask the three questions in different orders

Three questions shape a brand’s chances of breaking through on a particular need. Movability asks whether brands already differ on the need at all. Occupancy asks whether there is still room to build associations around it. Prevalence asks whether the need is big enough to drive growth.

The three questions rarely point to the same opportunity. Only 3% of brands have a silver bullet: at least one need that ranks best on all three dimensions. Every other brand must trade one strength against another. But the patterns in where brands of different sizes build earned strengths, and where they fall short, reveal which questions each should ask first.

Small brands: movability > occupancy > prevalence

Why this order? Almost no small brand has broken through on a low-movability need, so choosing a mainstream need with low movability would be a recipe for failure. A small brand should first ask whether a need is movable, and consider only those that are. Next, it should look for movable needs that are also free. Small brands hold earned weaknesses 1.7 times as often on occupied needs as on free ones.

Prevalence comes last. A need that is movable, free, and mainstream would be ideal: it could give a brand enough room to grow from small to large on a single proposition. But such needs are extremely rare. Most small brands will need to start with movable, free needs that are more niche. Only as they grow into mid-sized players should they shift toward more mainstream propositions.

To keep things simple, we divide the answers to each question into three bands: high, medium, and low. This gives us 27 possible combinations of the three dimensions. We then rank these combinations by movability first, occupancy second, and prevalence third, giving each need a breakthrough potential score from 1 to 27:

Breakthrough potential for small brands

Behavio (2026)

A need ranks #1 if it is high-movability, low-occupancy, and high-prevalence. High-movability, low-occupancy, and mid-prevalence needs rank second, followed by high-movability, low-occupancy, and low-prevalence needs in third place. Next come high-movability, mid-occupancy needs, regardless of prevalence (ranks 4–6), followed by high-movability, high-occupancy needs (7–9).

If no high-movability needs are available, the next best options are mid-movability needs (10–18). Low-movability needs rank last (19–27), and offer very little chance of breaking through.

Here are the top two and bottom two needs for a small brand in two categories, along with their ranks:

For small brands, movability decides which needs rank first

Behavio (2026)

Opportunities for challengers vary widely across markets and categories. In Germany, a small cola challenger’s best opportunity ranks #3. In the Netherlands, it ranks #10. In the US, it ranks #12, because no product need in the cola category is movable.

Mid-sized brands: the same order, without low-prevalence needs

A mid-sized brand first sets aside low-prevalence needs, those that matter to fewer than 60% of category buyers. To become a big brand, it needs to win buyers it doesn’t reach yet, and niche needs simply don’t reach enough of them.

It then ranks the remaining needs just as a small brand does: movability first, occupancy second, and prevalence third, using the two remaining prevalence bands.

Why the same order? Mid-sized brands do sometimes break through on low-movability needs, unlike small brands. But movability still matters much more than occupancy. A mid-sized brand is 2.8 times as likely to break through on a high-movability need as on a low-movability one, but only 1.5 times as likely to break through on a free need as on an occupied one. So when it has to choose, a high-movability, high-occupancy need is a better bet than a low-movability, low-occupancy one.

Both remaining prevalence bands, mid and high, are large enough to take a mid-sized brand to large. So prevalence comes last.

With this ranking, low-prevalence needs occupy ranks 19–27, while all other needs rank from 1 to 18:

Breakthrough potential for mid-sized brands

Behavio (2026)

Here are the top two and bottom two needs for a mid-sized brand in two categories, along with their ranks:

For mid-sized brands, niche needs drop to the bottom

Behavio (2026)

Big brands: prevalence > occupancy (low occupancy first) > movability

For a big brand, breakthrough potential is less about whether it can break through. Leaders hold an earned strength on about one need in four, almost regardless of the need. It is more about where a breakthrough is worth the most.

That flips the order: prevalence comes first. To keep growing, a leader needs to reach buyers it doesn’t already have, and only mainstream needs offer enough of them. Niche needs are also where leaders fall more than 6 points behind, 3.7 times as often as on mainstream needs. A leader that focuses on niche needs risks shrinking below the position its size would predict, and starting to look like a specialist.

Occupancy comes next, with free needs first. On mainstream needs, big brands win twice for every loss on free needs, against 1.4 times on occupied ones. On less prevalent needs, they do better on occupied ones, but those needs come last in the ranking anyway.

Movability comes last because it involves a trade-off. A movable need offers a bigger chance of winning, but also a bigger risk of losing. Big brands break through on movable needs twice as often as on low-movability ones, but also fall behind 3.7 times as often. That makes movability a tiebreaker, not the first priority. Low-movability needs are the safer choice; movable needs are the riskier bet. The ranking puts movable needs first because the goal is to keep growing, not simply to hold on to existing ground.

Sorted this way, a big brand’s needs line up like this:

Breakthrough potential for big brands

Behavio (2026)

Here are the top two and bottom two needs for a big brand in two categories, along with their ranks:

For big brands, the needs that matter to most buyers rank first

Behavio (2026)

Side by side, the three orders look like this, as in Figure 6:

The order of the three questions, by brand size

Behavio (2026)

Breakthrough potential is a starting point for strategy, not a brand’s final priority list. It reflects what one snapshot of the market tells us about each need, while marketers bring context the data cannot capture. One need may be growing quickly while another is losing relevance. A brand may not be able to deliver what buyers want. Or one need may offer far greater revenue or margin potential than another.

That’s why a need with higher breakthrough potential may sometimes be ruled out, while one with lower potential moves up the list.

A challenger has to focus its limited resources on one need. A leader, by contrast, typically runs several campaigns at once, choosing a set of needs that work well together. This makes the decision more complex and leaves room for exceptions. Focusing entirely on a niche need would risk shrinking a leader’s position. But a carefully chosen niche need can complement two mainstream ones.

2.8 Hidden-brand and hidden-market tests: the model predicts where challengers hold strengths

The strongest way to test whether breakthrough potential works is to see if it can predict data it has never seen. This is called an out-of-sample test. The usual shortcut is an in-sample test, which builds a measure and tests it on the same data. The results almost always look good because the measure is tailored to that particular dataset, including patterns that may not hold anywhere else.

So far, we’ve used all our data to estimate the chances, leaving no fresh data to test those estimates against. But there’s a simple solution: hide part of the data, one brand or even one entire market at a time. Then recalculate breakthrough potential without it. Finally, reveal the hidden data and see how well the rest of the data predicted it.

Hidden-brand test

We test whether a hidden brand actually holds more earned strengths on the needs ranked highest by its breakthrough potential than on those ranked lowest. To do this, we calculate the brand’s breakthrough potential using only the other brands, leaving the brand itself out of the analysis.

This test applies only to small and mid-sized brands. Big brands break through on many needs, so the more important questions are where they fall behind and how much each potential breakthrough is worth.

One by one, we removed all 140 small and mid-sized brands from the data: 49 small brands and 91 mid-sized ones. We made our predictions without each brand, then added it back to see how well they held up.

The results: small brands held a +6 earned strength on needs in the top third of their breakthrough potential 2.8 times as often as on needs in the bottom third. For mid-sized brands, the difference was 2.1 times.

Three of the 140 experiments:

Orthomol: its two top-ranked needs are its two strongest positions

Behavio (2026)

Then we revealed the brand: Orthomol, a German supplement brand on the market since 1991. Its two highest-ranked needs turned out to be its two strongest positions, both 11 points above the level predicted by its size.

Tesla leads on its highest-potential needs and trails on its lowest

Behavio (2026)

Tesla, in German cars since 2009. It sits 14 points above its size-predicted position on environmentally friendly and 9 points above it on a sense of status, its two highest-ranked needs. On its two lowest-ranked needs, it falls 7 and 5 points below its size-predicted position.

The Farmer’s Dog: its two top-ranked needs are its two strongest positions

Behavio (2026)

The Farmer’s Dog, a US subscription dog food brand on the market since 2016. Its two strongest product positions are also its two highest-ranked needs: 13 points above its size-predicted position on natural food and 10 points above it on sensitive digestion. It holds earned strengths on every product need, including its two lowest-ranked ones.

Hidden-market test

Here we test whether movability, then occupancy, then prevalence is the most predictive order for small and mid-sized brands. We also go one step further and challenge our own assumptions.

The hidden-brand test left out each brand when calculating its breakthrough potential, but it still used an order of the three questions chosen based on all brands. What if that order was wrong?

This time, we tested all 12 possible combinations. The three questions can be asked in six different orders, and for occupancy, we can prioritize either free or occupied needs. Each time, we used four of our five markets to identify the best order, then tested it on the fifth, hidden market. We repeated this process five times, hiding a different market each time.

For mid-sized brands, the same order came out on top every time: movability, then occupancy, then prevalence. For small brands, movability came first in three of the five markets. But there were too few breakthroughs to determine whether occupancy or prevalence should come second.

Together, the hidden-brand and hidden-market tests provide the strongest validation possible with a single wave of data. They show that the associations a brand builds matter, and that, at least for challengers, the choice is not arbitrary.

2.9 The growth cycle: how differentiation lifts challengers, then disappears

How well do these numbers actually reflect what happens in a category? US dog food offers one example.

For decades, Purina and Pedigree dominated the category. But this century, several challengers, including four in our study, broke through: Blue Buffalo, Freshpet, The Farmer’s Dog, and Taste of the Wild. Each has built a sizable buyer base since it was founded.

Their stories have a lot in common. Each started with the same belief: ordinary kibble was not good enough. But each found a different way to challenge it.

In 2003, a family whose dog had cancer couldn’t find food they trusted, so they made their own. Blue Buffalo was born, offering natural kibble with real meat as its first ingredient.

In 2006, three pet food industry veterans took a different approach. They founded Freshpet, ditched kibble altogether, and began selling fresh, cooked food from refrigerated units in grocery stores.

Then, in 2015, a New Yorker whose dog kept getting an upset stomach was advised by her vet to cook for her. That home cooking became The Farmer’s Dog. After more than 200 manufacturers turned them down, they found a human-food factory to produce their meals and began delivering them directly to customers by subscription.

Taste of the Wild was the exception. Unlike the other three, it came from an established pet food maker. Launched in 2007, it built on the grain-free recipes of a few smaller pioneers, drawing on the idea of an ancestral diet.

Four brands, four different answers to the same question: what does healthy, natural food for dogs look like? Each has carved out a distinct position in the market. Here’s where those positions stand today:

Healthy, natural dog food: each challenger leads on different needs

Behavio (2026)

Does that mean The Farmer’s Dog has been the most successful brand in healthy, natural dog food? No. It means only that The Farmer’s Dog currently has the strongest position in buyers’ memories.

To understand why, let’s return to the growth cycle introduced earlier. When a brand builds associations in buyers’ memories beyond what its size would predict, it moves through five steps:

1) Earn. The brand builds associations beyond what its size would predict. Most of these come from people who don’t buy it yet.

2) Convert. Some of those people become buyers.

3) Absorb. As the brand grows, more people encounter it: on more shelves, in friends’ homes, and across more online stores. Buyers and non-buyers alike begin associating it with the same needs they associate with established dog food brands, far beyond its original proposition. Its line rises across every need, including the one that first set it apart. The original advantage is absorbed into the associations that come with greater size, and the difference disappears.

4) Get copied. Once a challenger proves that a particular need can drive growth, competitors typically move into the same territory. This further erodes the brand’s original advantage.

5) Settle. The need becomes established territory. Every brand sits close to the level of associations its size would predict. To keep growing, the brand must find another need to break through on.

The Farmer’s Dog sits between the earn and convert stages. Its position in buyers’ memories runs well ahead of its sales, so more of its growth potential is still stored in people’s minds than for any other brand in this comparison.

Two things stand between that memory and sales. The first is time: people change their dog’s food only occasionally. The second is availability: The Farmer’s Dog falls below the level its size would predict on being easy to buy in stores and available in supermarkets.

In March 2026, the brand started selling through Walmart.com, reducing one of these barriers and giving more of its earned memory a chance to turn into sales. Getting onto physical store shelves would likely be its biggest growth opportunity right now.

The Farmer’s Dog is in a strong position right now, and the more competitive the need, the stronger its advantage:

The more movable the need, the bigger The Farmer’s Dog’s earned lead

Behavio (2026)

Breakthrough potential points to sensitive digestion ahead of natural food. The reason: sensitive digestion is still an open space, while natural food is already occupied by competitors. This fits the brand’s stated targeting of owners whose dogs have health issues, including digestive problems.

Yet its TV ads focus on real food, freshness, and longer lives, rather than digestive health. This could explain why its lead on sensitive digestion is smaller than its lead on natural food. A more focused message around digestive health could help it build an even stronger position.

Freshpet, meanwhile, has moved further along the cycle, into the absorb stage. It pioneered fresh dog food, but today trails The Farmer’s Dog on natural food by a wide margin. This suggests much of the memory it earned has already converted into buyers.

Its weaknesses now extend beyond the product itself: it also falls behind on promotion and online availability. To avoid losing its lead to the second mover, Freshpet needs to start rebuilding its memory advantage. That means investing in high-reach campaigns and strong creative work.

Blue Buffalo is the largest of the three, generating more revenue than either rival. Its size also puts it in a strong position to copy the fresh food proposition and capture much of its growth potential. In 2025, it launched its own fresh line, targeting what General Mills describes as a $3 billion segment.

Much of its original advantage in better kibble has been absorbed by now. Yet Blue Buffalo remains strong on product in buyers’ memories, even as fresh food grows and dry food declines. It also has a much stronger position on place, both in buyers’ memories and on actual store shelves. That’s an advantage The Farmer’s Dog cannot easily or quickly replicate.

And the losers of the product fight?

Taste of the Wild’s earned weaknesses across all needs suggests its sales may be declining. One possible explanation is the FDA investigation into grain-free dog food in 2019, after which sales of the brands named in the investigation fell.

It’s also possible that much of Taste of the Wild’s sales comes from outside buyers’ memories, through pet shop recommendations or automatic reorders. This is common for small, fast-growing brands, but would be much harder to sustain for a brand of its size.

Royal Canin also looks unlikely to see significant growth in its current position. Whatever advantage it once had appears to have been absorbed. Its positioning centers on exactly for my dog, a need where brands barely differ. Despite its many breed-specific product lines, it has failed to build a distinct position there.

With weak awareness and advertising, it will struggle to shift buyers’ perceptions. Royal Canin may therefore be a candidate for repositioning.

And what about Purina? As the category leader, it built its lead on complete, affordable food available in every grocery aisle. It still wins on exactly those fronts today: price and supermarket availability.

But no product need in the category is large enough to fuel further growth for a brand of Purina’s size. Its strength on price also keeps it out of the new fight. Across our study, brands that win on low price tend to lose on perceived quality. Purina is a clear example, sitting well below the level its size would predict on natural food. That’s often the trade-off that comes with owning the low-price position.

So does Purina need to respond at all? It has already launched a deep-frozen fresh food line, Purina ONE Fresh Entrées, but under a smaller sub-brand. Whether it can bring its mass production up to human-grade standards while keeping prices affordable is another question. That would be no easy feat.

Ultimately, Purina does not need to win on product. It only needs to lose less ground there, as long as it holds on to its lead in price and availability.

Averaging each brand’s cues within each of the four Ps gives us a snapshot of where each brand stands:

Each US dog food brand has a different 4P profile

Behavio (2026)

The pattern we see in US dog food plays out across many other categories too. Together, these stories help explain the numbers and breakthrough odds from the previous chapters. The challengers did not seem to grow by finding an empty gap or simply looking different. They picked their fight and had the product, organization, and promotion to win it.

The next chapter explains a choice this chapter made without much fanfare: treating price, place, and promotion as part of mental availability. Shouldn’t mental availability be limited to category entry points, with price left out? And shouldn’t physical availability be measured on actual shelves?

Chapter 3

Mental availability includes links across all 4Ps, not just CEPs

Mental availability is the chance that a brand is noticed or thought of in buying situations. It rests on memories that link the brand to the different cues that come up in these situations. The more cues a brand is linked to, and the fresher those links, the better its chance of coming to mind and being bought.

On the page in How Brands Grow where mental availability is first introduced and defined, Byron Sharp describes it as the network structure in buyers’ memories and illustrates it with a bank. The first association on his list is a branch near work, a place association. Three others of the twelve are also about place: a branch on high street, internet banking and can get cash out. Three are about the product: home loans, Visa and my first credit card. Another three are distinctive assets, which fall under promotion: the color, the logo, and the staff uniform (p. 177).

However, in How Brands Grow Part 2, in a chapter written by Jenni Romaniuk, the definition of mental availability narrowed. It became tied to category entry points (CEPs): situations and needs that arise before any brand enters the picture, such as saving for a home or moving to a new city. These are what trigger a search in memory. Almost none of the bank associations from the first book would still qualify. A branch near work belongs to one specific bank, so it cannot trigger a search for any bank.

In this view, only CEPs “get a brand into the race.” Brand attributes linked to product features or price, described as heuristics, baseline category competencies, or reasons to reject a brand, only help buyers decide “which one to buy when there is a mental availability tie.” And if a brand is hard to find, that is a physical availability problem, assessed through store audits, not a question of memory.

In this chapter we show that the original, broader view makes more sense, both in the data and in practice. CEPs are more likely than other attributes to have low movability, where every brand gets the link simply because of its size. Brands also build strong, readily accessible memories around price and place, just as they do around other needs relevant to a buying situation.

3.1 Category entry points corrected brand-centric attribute lists

Before challenging category entry points, it is fair to recall what a typical brand attribute battery looked like before they came along. How Brands Grow provides an example: trusted, efficient, rapport, relevant, solution, innovative, essential.

How often have you sent a parcel or bought a phone and thought, even subconsciously, “Ah, essential”? Attribute batteries relied on absurdly brand-centric language, as if buyers had enough spare glucose for their brains to conjure up words like “rapport” while standing at the shelf.

Turning the battery around to focus on buyers and their real needs was absolutely the right correction. To whiten my teeth or to get through the afternoon slump can be genuine needs that lead people to look for a product or brand to meet them.

The second key idea is that a need can simply be an occasion: a Friday night in, a kid’s birthday party, when friends drop by unannounced. There need not be a reason to believe or a goal behind it. A brand can be linked to a situation in life, and that alone can matter.

Romaniuk’s 7W framework (when, while, with what, with whom, where, why, how) is a useful way to build a long list of situations in which your brand could be the answer.

But Romaniuk goes further, drawing a distinction between CEPs and other attributes. Only CEPs count toward mental availability. Other associations are classified as “baseline category competencies” or reasons to reject a brand. This is where the framework starts to get a little complicated, unclear, and shaky.

3.2 Only the wording separates a CEP from an attribute

For Romaniuk, reliable and good quality are “baseline category competencies,” not CEPs, because they cannot exist independently of a brand or product. Nobody needs good quality in the abstract. They need something to be good quality. A buyer does not start the day thinking, “I need a good-quality coffee brand.” They start by needing a morning energy boost. No brand is required to define that need: coffee might provide it, but so might exercise. This is where memory retrieval begins. The argument, then, is that only CEPs count toward mental availability.

It sounds plausible, but the distinction between the two mostly comes down to wording.

A simple copywriting change can move an attribute into CEP territory. Reliable does not qualify because reliability needs something, such as an outdoor-wear brand, to belong to. Rephrased as to stay dry when the weather turns, it suddenly fits the Why type of CEPs. That kind of rewording can be useful. But nobody knows exactly how memories are worded or stored in our heads, and they are most likely stored as neither version. Both are highly relevant when buying outdoor gear, far more so than essential. Reliable carries much the same meaning with less effort for the brain to decode. That is hardly a reason to exclude it from mental availability.

3.3 CEP-like needs more often have low movability

Romaniuk recommends that CEPs make up most of a brand tracker’s attributes, 60–70%, because they are what make up mental availability. The remaining attributes, baseline category competencies such as reliable or good quality, are there to check that the brand keeps up, “to ensure the brand does not underperform.” This is presented as a rule of thumb, without supporting evidence.

The data tell a different story. Many textbook CEPs have low movability: every brand gets linked to them in proportion to its size, just as it does with a baseline competency. Take coffee, for example.

A morning energy boost is as clear a CEP as any. Yet in the UK and the Netherlands, it is the least movable product need on the list: brands sit just 1.8 and 2.0 points from their expected positions. Every brand gets the link its size predicts, and none gets more. In a tracker, this is exactly the kind of cue that only checks whether a brand is keeping up.

Premium quality, which does not qualify as a CEP, is the most movable product need in Dutch coffee, with brands sitting 6.7 points from their expected lines. In Germany, the two most movable needs are environmentally friendly (7.4) and gentle on the stomach (6.8), neither of which qualifies as a CEP either. These are the needs where coffee brands differ, yet a tracker following the rule of thumb would only be checking that brands keep up.

Coffee is not an exception. The line between a CEP and an attribute is blurry, so we sorted all 397 product needs using the distinction most often applied in practice. On one side are needs, goals, feelings, and occasions that exist independently of any brand: a pick-me-up, to watch sports events, or for an aging dog. On the other are qualities, ingredients, origins, features, and services: premium quality, German origin, expert advice, or high meat content. The CEP-like needs have low movability 40% of the time, compared with 28% for the others. We set aside 33 needs that could go either way. To watch on multiple devices in streaming, for example, is a situation the buyer is in, on the train or in front of the TV, but also a feature of the app. Counted on either side, the result barely moves: 40% against 28–30%. Meanwhile, the needs that a CEP list would demote are almost twice as likely to be movable: 18% versus 10%. The pattern holds across all eight categories. In coffee and colas, not a single CEP-like need is movable.

CEP-like needs more often have low movability

Behavio (2026)

None of this makes occasions useless. Great for evening relaxation in Dutch streaming is a genuine occasion. It matters to 85% of buyers, and it is a competitive battleground: brands sit 6 to 8 points from their expected positions, and five of the six claim it less than its prevalence would suggest. Netflix sits 9.7 points above its line, Film1, with 6% of buyers, sits 8.3 points above, and Viaplay sits 10.8 points below. An occasion like this belongs on every tracker’s long list. But it earns its place the same way premium quality or expert advice does: it matters, and brands can compete over it.

And all of this was product. Price and place can be just as movable: in some markets and categories, low price and easy to buy in-store are more movable than almost any product need in the study. Should they be part of mental availability too?

3.4 Price and place memories are retrieved as fast as CEPs

Technically, price and place could serve as CEPs: when money is tight, or without having to go far. In the Ehrenberg-Bass model, they never do. In every example, they are classified outside mental availability, as heuristics, competencies, or physical availability. They come into play only after the CEPs have done the retrieval, helping decide “which one to buy when there is a mental availability tie.”

Nobody has found evidence of this sequence in the brain. Neither cognitive science nor neuroscience shows that brand memories are retrieved in stages by type:⁵ first a CEP, then, once a few brands come to mind, price and quality. As far as we know, memories are not sorted by type. They most likely activate in parallel, all at once.

Our own data points the same way. Every response to the 4P cues came within 3 seconds, usually much faster. We discarded the few slower responses. Even so, brands still own strong positions on price and place. These are fast memories, not the result of deliberation.

So this line does not hold either: “you still need the Mental Availability to be in the race, even when the heuristic helps make the choice.”

The first time you find yourself broke, there is no established memory linking an occasion like when I am broke to a brand. You are simply broke, and several things activate at once, including the type of product you want and the need to have it cheap. As a result, cheaper brands come to mind. A price or a place memory can be exactly what puts a brand in the race.

5. The brain does work in two steps, just not by type of memory. First, the hippocampus brings back whatever is linked to the situation, so it could be a need, a price, a place, all at once. Then the ventromedial prefrontal cortex puts everything that came back on one scale, and the result is the feeling I want this one. Neither step looks at CEPs first and price later (Shadlen & Shohamy, 2016; Bartra, McGuire & Kable, 2013). ↩

3.5 Price and place in memory differ from actual price and distribution

The usual reply is that place and price are not memories at all. Place is physical availability, price is what appears on the label, and both can be checked through store audits and price lists. True, but none of this argues against price and place memories being crucial to mental availability. A shop opens twenty meters up the street. If nobody knows it is there, your distribution has improved, but your sales have not. A brand can cut its price and still be remembered as expensive.

The gap is the point, not a flaw. Marketers already accept this for promotion. An ad can reach a million screens, yet very few people may notice or remember it. Only those who remember it can buy because of it. The same gap exists across the other three Ps. Audits and channel data are essential for many tactical and operational decisions. But if you have only one number to watch, the Ps in mind are the better choice: a brand is rarely seen as easy to buy unless it is, while it can sit on every shelf and still go unnoticed.

3.6 The 4Ps map to what marketers control, so each gap points to a team that can act

The marketing 4Ps, product, price, place, and promotion, are the four main levers marketers can pull. They were never defined by what matters to buyers, but by what a manager can control. They divide up the manager’s job, not the buyer’s mind.

That is also the standard criticism: the 4Ps are too manager-centric, built around what marketers control rather than what people want. This is probably why they never became a framework for sorting attributes in brand measurement, a field that sees itself as the buyer’s advocate. Instead, what gets measured has followed each era’s view of the customer’s mind: functional benefits in the rational era, brand personality in the era that followed, and CEPs today.

We interpret the 4Ps through the lens Bob Lauterborn proposed, as the customer’s 4Cs. Seen from the buyer’s perspective, they cover nearly everything that shapes a choice, while most batteries cover only part of it:

Promotion is the one people are least willing to admit influences what they buy. Yet there is plenty of evidence that it does: familiarity builds trust and makes the choice feel safer.

Of course, buyers’ memories are unlikely to be sorted into these four boxes in the brain. But companies organize their work, and often their teams, around the 4Ps: product, pricing, distribution, and marketing. Most marketing teams no longer control the other three, but they are well placed to show how all four work together in the minds of current and future buyers. It is a perspective the board rarely gets from anyone else.

Measuring the other Ps is also the only way to properly interpret marketing’s own results.

3.7 Promotion can only be judged alongside product, price, and place

Most marketers have sat through this meeting. The campaign tested well. Awareness is up, and brand scores are up. Sales are not. Then someone asks what marketing did with the money.

There are two possible explanations, and they call for opposite actions. Either something else is holding the brand back, whether the product, the price, or how easy it is to buy. In that case, more advertising will not fix the problem. Or nothing else is holding it back, and the memory is simply being built before the purchases follow. In that case, the campaign needs more time. On a dashboard that measures only promotion, these two situations look identical.

Measured across all four Ps, these situations look very different. Three challengers in the German car market show how differently a brand can be held back. Tesla sits 9 points above its size on promotion, but price holds it back, worst of all on good value for money. Hyundai faces the opposite problem: it performs far above its size on price but is weakest on promotion, with both its look and its advertising below what its size would suggest. BYD is strong on price, too, but held back by place: servicing, showroom purchases, and test drives. For each brand, the biggest gap sits with a different team.

Three challengers, each held back by a different P

Behavio (2026)

The reverse happens, too: promotion can carry a brand that sits below its size on other Ps, whether by design or not. Coca-Cola in Germany, the Apple Store in the US, and Mercedes in Germany all sit well below their size on price, and well above it on promotion. But each is carried by something different. Coca-Cola is recalled unprompted 13 points more than its size predicts, and its advertising stands out. The Apple Store is carried by its look, its brand elements 11 points above its size, while it is recalled barely more than its size predicts. Mercedes is carried by its advertising, which sits 11 points above its size, while its awareness and recall are no higher than its size predicts. For all three, being below their size on price is part of their position. Mercedes, for example, sits 23 points above its size on status and 22 points below it on low price. We explore the nuances of interpreting these profiles in Part Two.

Three brands carried by promotion despite weak price scores

Behavio (2026)

None of these conclusions is possible from a promotion dashboard alone, or from the usual price and distribution data. Marketers usually respond that they already know their price and distribution. But a store count does not tell you how many buyers find the brand easy to buy, and a price list does not tell you how many consider it good value. Only when all four Ps are measured in the same units and against the same size line can promotion be assessed alongside the other three. Then even a flat quarter after a campaign can be read: the ads had no effect, or they held the line while the product, price, or place got worse.

Even if most marketers control only promotion, measuring the other three Ps using the same cues, protocol, and scale lets them make better decisions about how to use it.

3.8 The 4Ps plus ease of retrieval predict brand size in five markets

A careful reader may have noticed two unfamiliar cues in the last two figures: known and unprompted recall, both measured relative to brand size. Unlike the other cues, which measure prompted links between a need and a brand, these two measure prompted awareness and unprompted recall, each compared with what the brand’s size predicts. We added them after investigating whether the 4Ps really explain a brand’s sales.

That check is harder than it sounds. Brand size drives around 90% of every brand metric, so everything correlates with sales. Even a low-movability cue can track sales at 0.99. Liking explains 95% of the differences in sales between brands. Most of that is liking following sales, but what explains the rest? Most likely, something in the concrete 4P cues. Liking rolls all of it into a single number, leaving nobody any wiser about which factors matter. And “be more liked” is not something a brand can act on.

If the Mental 4Ps track sales just as well, that changes the picture. Every detail is captured in the number: the positions a brand owns, how movable they are, and their occupancy. We averaged the cues within each of the four Ps, then averaged the four scores into one. All scores were standardized within each category. Together, the Mental 4Ps explain 92% of the differences in sales between brands, nearly as much as liking.

Could anything improve the result? We tested everything else in the data, including liking. It all moved with the 4Ps and added no new information, with one exception: unprompted recall. Among mid-sized brands, recall behaves differently from the 4Ps. Whether a brand sits at 10% of buyers or 30% can depend on having more 4P links, or simply on being easier to recall. So we added unprompted recall, adjusted for brand size, to the promotion box. Once past buying is accounted for, what remains is ease of retrieval: how readily the brand comes to mind, which promotion can largely influence. Throughout this paper, ease of retrieval is the concept and unprompted recall, adjusted for size, is how we measure it. Adding it improves the result by three percentage points. More concretely, it brings 11 more brands in every 100 to within 5 points of their actual size.

Together, the 4Ps and ease of retrieval explain 95% of the differences in sales between brands, matching liking. But unlike liking, we know what is inside.

This isn’t tailored to one country. We developed the model in the US, froze it, and used it unchanged to predict brand sizes in Czechia, Germany, the Netherlands, and the UK. It explained 93% to 96% of the differences in each market, as it did at home.

The US model predicts brand size in four other markets, unchanged

Behavio (2026)

All of this rests on one step we have so far taken on trust: separating what a brand owns from what its size gives it. The next chapter shows how we account for size, and why most trackers cannot.

Chapter 4

Why standard trackers miss earned positions, and how we measured them

Roughly 90% of any brand metric is the brand’s size. Only the remaining 10% tells you something about the brand itself. This changes how every metric in a tracker should be interpreted, not just attribute scores. In our data, awareness explains 80% of the differences in sales between brands, unprompted recall 87%, consideration 93%, and liking 95%. Loyalty, perceived uniqueness, promoter score, and many other well-known metrics follow the same pattern. Taken at face value, they mostly measure buyer numbers in disguise. Until you separate the effect of size, you cannot tell what else these metrics reveal about a brand.

But accounting for size properly is harder than it might seem. We tried to keep it simple, but found that the simple approaches were misleading. What held up was a specific curve for each cue, fitted across all categories and markets at once.

4.1 Separating earned from size-granted answers depends on a well-drawn size line

An earned memory is a deviation: how far a brand sits above or below what a brand of its size would typically score. So everything depends on getting one thing right: the expected score that brand size alone would predict. Drawing that line is a statistical task. The full method, including the reasoning and equations, is in the methodological appendix. Here are the principles behind it and how we arrived at them.

Look at these two needs, where no brand clearly owns a position. But the effect of size is very different for each cue. Great taste climbs from 6% for the smallest brand to 75%. Natural ingredients tops out at 48%, even for Coca-Cola, which reaches nearly nine in ten Dutch cola buyers. In both cases, the need is almost fully size-driven. But more people feel great taste is typical of colas than feel natural ingredients are.

Size grants some cues far more than others

Behavio (2026)

About 90% of the differences between scores is size-granted, but not all of it comes from size alone. 76 points come from brand size. Another 14 come from the need, because how much size gives a brand depends on how typical the need is for the category. So every cue, in every category and market, needs its own curve. A tracker measuring 20 attributes needs 20 separate curves to calculate each brand’s deviations correctly.

There is no shortcut. Any method that skips this step will invent strengths and weaknesses, sending brands down paths they cannot win. Mental advantage, the most widely used size adjustment in brand tracking, can misjudge a brand’s true position by up to 15 points. The next subchapter explains why.

So what is the best way to draw the curve? Not by deciding its shape in advance. We tested different shapes and kept the one that predicted scores most accurately. The test is simple. Hide one brand’s score, draw the curve using all the other data, and ask it to predict the hidden score. The difference between the prediction and the actual score is the error. If you repeat this for every score in the study, the average error shows how well the curve captures what brand size alone would predict. We tested nine shapes. The best missed by 5.1 points on average. Without any size adjustment, the average error was 17.7 points.

The model is hierarchical, meaning each curve can also draw on information from the others when needed. Many trackers have relatively few brands to work with. Our study includes up to 12 brands per category, but usually just six. A curve drawn through six points can swing sharply when one point moves. When the six brands provide little evidence about the curve’s shape, the model draws on how the same type of cue behaves across the other 39 category-market combinations. When the six brands show a clear pattern, the curve gives their data more weight. For the smallest brand in each category, this reduced the average prediction error from 5.0 to 3.7 points.

Winning a comparison is not enough, so we ran many tests designed to find where the method might fail. For example, we drew 20 new smaller samples of 1,000 people each. A brand’s score varied by about 1.2 points between samples. Or we reduced the questionnaire to 12, 8, and 6 cues. The results matched those from the full questionnaire with correlations of 0.99 or higher. Full lists of test is the the appendix.

The method is reliable enough for us to confidently treat deviations greater than 3 points as earned memories.

4.2 Mental advantage judges a brand against itself, not against its size

The most widely used size adjustment in brand measurement is a multiplicative model, described in Better Brand Health by Jenni Romaniuk and called mental advantage. In a table, its results look just like ours: +8 here, −4 there.

It was designed as a messaging check within the “rotate the mainstream needs” playbook. For that purpose, it can work, with some caveats we’ll come to below.

Its defining feature is that all the scores sum to zero. If you add up a brand’s advantages and disadvantages, they cancel each other out. This changes the question the model answers. It does not ask, “Mirror, mirror on the wall, who is the fairest of them all?” It asks, “Which is my fairest feature?” That answer can be useful. It shows which features are relatively strongest and weakest for a brand. But your fairest feature might still be plainer than the plainest feature of the fairest brand in the land. Mental advantage will never tell you that.

A brand that is strong across the board sees its ordinary positions as weaknesses. In the Netherlands, Coca-Cola scores exactly where its size would predict on I can afford it regularly, a cue where no brand stands out. Yet mental advantage labels this a 9.6-point disadvantage. Netflix scores 5 points above what its size predicts on easy to buy online, but mental advantage calls it a 9-point weakness.

Neither brand has anything to fix here. Both are simply so strong on other cues that a model forced to sum to zero has to push some scores below the line.

The reverse happens, too. A brand that is weak overall mistakes a weakness for a strength. Purina, the leader in US dog food, falls short of its size on 14 of its 20 cues, so mental advantage expects less from it. On to experience closeness, it reports a small advantage of +1 point. But a brand of Purina’s size would typically score 54 on this cue. Purina scores 48, 6 points below where its size would predict.

The chart shows how mental advantage’s expected values jump up and down from brand to brand. That is because each brand’s expected score comes from its own total links, split across cues the way the whole category splits them, not from its size in the market. Mental advantage does not isolate what a brand has built beyond its size. It adjusts a brand for its size and its other strengths at once, and cannot tell the two apart.

Mental advantage and the size line judge Netflix in opposite directions

Behavio (2026)

Summing to zero has two more costs. First, mental advantage cannot tell you whether a brand is strong overall. In our study, 33 brands score clearly above what their size would predict across their cues. Mental advantage identifies none of them, because every brand’s scores must sum to zero.

Second, whatever you gain on one cue, you lose on the others. If a campaign lifts one cue by 10 points, mental advantage shows a gain of about 8 points on that cue. It then subtracts those 8 points from your other cues, taking a little from each, even though buyers’ perceptions of you on those cues have not changed.

Across our data, mental advantage points in the wrong direction 240 times. It labels 141 genuine strengths as weaknesses and 99 genuine weaknesses as strengths. That is 10% of all earned positions in our data.

4.3 Random speeded pairs reduce the biases of lists and scales

Most brand trackers use one of two methods. Agreement scales show a brand followed by a list of attributes to rate. Pick-any lists show a cue followed by a list of brands to choose from. We used a third method: random, speeded pairs. Respondents see dozens of pairs, one at a time, with a short cue on the left and a brand logo on the right. They answer yes or no. The pairs appear in random order, and any response taking longer than 3 seconds is discarded. We believe this makes a substantial difference when measuring earned positions precisely, especially for small and mid-sized brands.

Three ways to ask the same question

Behavio (2026)

Why go to the trouble? Every list creates a context, and context introduces bias. A brand is judged alongside the names around it, and a cue is judged against the cues that came before it. The answer starts to reflect the screen as much as the memory itself. Many mechanisms can introduce this bias, and they are likely to affect small brands more than large ones. The size curve cannot correct for a bias that does not follow brand size.

Random pairs also solve a problem every tracker owner knows. In scales and lists, answers influence one another. If you change the questionnaire between two waves, the data shift even when nothing has changed in the market. With random pairs, each answer stands alone. You can add or remove a brand or cue without changing anyone else’s answers. When we shortened the questionnaire to 12, 8, and 6 cues, the results for the remaining cues matched those from the full version with correlations of 0.99 or higher. This independence also made the experiment in Section 2.8 possible: hiding a brand and asking the model where it should fight. It also allows trackers to maintain a long-term time series even as the brands and cues they measure change.

Chapter 5

Conclusions and next steps

Many influential ideas in marketing science rest on patterns observed across brands, often in selected case studies rather than whole categories, and rarely checked on data the analysis has not seen. This study measures every pattern across whole categories in five markets, takes each brand’s size out need by need, and confirms the key results on brands and markets left out of the analysis. The evidence strongly supports all six main claims.

The first claim is methodological: brand image only makes sense relative to brand size, and every need needs its own size expectation, because each need rewards size differently.

The second describes differentiation: 90% of what people say about brands can be explained by size, yet once size is taken out, most brands hold clear strengths and weaknesses of their own.

The third concerns where differentiation happens. Occupancy, prevalence, and movability all shift the odds. The new one is movability: on some needs brands consistently sit far from their size, on others they do not. Movability is measured on a brand’s competitors, never on the brand itself. The effect also holds out of sample: with each brand set aside, the need where competitors differed more was the one where the brand broke through in 62 of 100 pairs, against about 50 by chance. In all five hidden-market tests, the same order of the three questions came out on top for mid-sized brands.

The fourth is that category entry points have no special status among brand memories. Memories of price, place, and promotion come back just as fast, and CEP-like needs are more often low in movability than other product needs (40% vs 28%). The result holds whichever way the 33 borderline needs are counted.

The fifth is that differentiation can turn into size as a brand grows. An earned strength helps a brand win buyers; as the brand grows, the strength becomes part of what its size grants it. That is why a snapshot study finds differentiation “shockingly low”: 90% of associations reflect size even where differentiation drove the growth that created that size. As an explanation, it fits the data. How often this happens, and how fast movable needs settle, is worth measuring over time.

The sixth is that building memories on the needs with the highest breakthrough potential grows penetration more efficiently, against the claim that differentiation hardly matters and any message works about as well. The data points clearly one way. Small and mid-sized brands hold earned strengths 2.5 times as often on high-movability needs as on low-movability ones. Big brands lose ground on those needs 2.6 times as often as they gain it, and lose 3.7 times as often on niche needs as on mainstream ones. Differentiated brands turn roughly 17% more of their awareness into buyers, and only 2 of 246 brands came close to growing on fame and distinctiveness alone.

One level of proof for the last two claims remains: a longitudinal study showing that gains in specific earned memories predict penetration growth, after controlling for other factors. Such studies are rare in marketing science. They need years of consistent measurement and very large samples, because only a small share of brands grow in any given year, and even the largest tracking databases seldom measure the same needs consistently enough. We are working on a way to make this test affordable.

Such a study would add detail on when and how earned memories turn into buyers. It is very unlikely to overturn the patterns here. For example, if challengers broke through on low-movability needs as often as on others, the snapshot would already show it.

About

About the authors

Vojtech Prokes and Jiri Boudal are the co-founders of Behavio, a research technology company.

They studied sociology and market research together before pursuing separate careers in marketing and research. In 2015, they reunited on a glacier alongside two former classmates, one completing a PhD in behavioral economics, the other in biostatistics. Somewhere along the climb, the four decided to start a company that would pioneer new approaches to brand and advertising research, combining behavioral science with non-declarative measurement.

Their ambition was to turn that science into clear, actionable next steps for marketers, not another 100-slide report.

Behavio now serves 600 clients and measures brands in over 40 markets.

Clients include Decathlon, IKEA, and dozens of the kind of brands this paper is about: underdogs that became the biggest in their categories across the continent, such as Bunq, Notino, and GymBeam.

Many people at Behavio contributed to this study.

Appendix

Methodological appendix

A. Data and sample

The study covers 246 brands in 8 categories and 5 markets: the US, the UK, Germany, the Netherlands, and Czechia. Each category in each market is one cell, 40 in all, with about 6 brands each. In every cell, 1,500 nationally representative respondents were screened for category buyers, 60,000 interviews in total, collected from June to August 2026.

Brand size is penetration: the share of category buyers who bought the brand in the past year, or in the past 10 years for cars. Across the study it runs from 1% to 89%.

Each cell has 18 to 20 needs across the 4Ps, and every brand was scored on every need: 4,758 brand–need scores. With relative awareness and relative unprompted recall added for each of the 246 brands, the study holds 5,250 measurements. Product needs alone give 2,442 brand–need pairs, 484 for small brands, 902 for mid-sized, and 1,056 for big. These 2,442 pairs are the base of every breakthrough table in Appendix D.

We are open to sharing the data for research and scientific purposes.

B. How the size curve was chosen

The model. A hierarchical (mixed-effects) logistic regression that models each association score as a function of the logarithm of brand penetration, defined as its share of category buyers. The shared component is estimated once using restricted maximum likelihood (REML), based on all 4,758 scores across 40 category-market cells.

The equation. For each cue, the model estimates the odds of a brand being linked to it:

logit(score) = μ_p + β · log(buyers ÷ 20) + u(category) + v(cue)

Here μ_p is the baseline level for each of the 4Ps when a brand has 20% of category buyers. This baseline is shared across the study. β = 1.18 describes how quickly scores rise with brand size: doubling a brand’s buyers multiplies its odds of being linked to a cue by about 2.3. u(category) allows each category-market cell to start at a higher or lower level and rise faster or slower than the study-wide average. v(cue) allows each cue to differ in the same way within its category.

Why hierarchical. The category and cue effects, u and v, are not estimated independently. Each is pulled toward zero depending on how much its own data, usually from just six brands, can be trusted. Roughly, the weight given to a local estimate is: w = τ² / (τ² + error²). Here, τ² measures how much categories or cues genuinely differ, while error² measures the uncertainty in the local estimate. With few brands or noisy scores, w is low, so the curve relies more on the rest of the study. The slope also depends on how widely the brands vary in size. When their sizes are spread out, the curve can follow their pattern. When their sizes are clustered together, it borrows more heavily from the study-wide slope.

What is estimated where. μ_p, β, and the variances are estimated once for the whole study and are not re-estimated for each new category. For each new measurement, only u and v are estimated, using the data from that category’s own brands.

What is left out. The model deliberately excludes a brand-level random effect. This keeps a brand’s overall strength in its deviations rather than averaging it away. The question remains: “Does this brand score above what its size would predict?” rather than “Does it score above its own average?”

Awareness and unprompted recall are not associations and get their own curve: one slope for each, estimated once across all categories, and a level set within the category.

Chapter 4 gives the model and its equation. The tables below show what it was tested against.

Four ways to define a deviation, applied to the same 4,758 scores:

The category average leaves size in: its deviations correlate 0.845 with brand size. The index against the leader flatters small brands: across all brands, the average deviation comes out at +6.2 instead of zero. The multiplicative model, used by Romaniuk and Sharp among others, lands close to ours (correlation 0.82), but it takes brand size from the brand’s own image scores, so every brand’s average deviation is zero by construction. Of the 33 brands that sit more than 3 points above their curve on average in our model, it flags none.

RTL+, bought by 22% of German streaming buyers, is linked to wide selection of content by 38%. The category average puts it 10.3 points below, the index against Netflix, the leader, 10.6 points above, and the size curve 1.1 points below.

Curve shapes. Each candidate was scored the same way: one score held out, the curve fitted on the rest, the held-out score predicted. The error is the average miss in points.

The halo version wins this test and is still left out. With one score held out, the halo learns the brand from its other scores. With the whole brand held out, the gap nearly closes: 5.25 with the halo, 5.34 without. A halo would also change the question from “above what its size predicts” to “above its own average,” so a brand weak on every need would come out normal.

The logarithm was not imposed. Left free in the Box-Cox family, where an exponent of 0 is exactly the logarithm, the exponent lands at 0.012.

Two results sit on thresholds. On fresh samples, a pair changes deviation band in 11.3% of cases and movability band in 13.8%, almost always right at a cut-off: 73% of the movability changes lie within half a point of 3 or 6. The deviation itself barely moves. The bands are cuts through a continuous number, so a pair right at a cut-off can be read either way.

The curve is evidence only between the smallest and the largest brand measured in a cell.

C. How movability, occupancy, and prevalence are computed

Earned association. For brand b and need n: earned = observed − expected, where expected is the score the size curve gives a brand with b’s share of category buyers. This curve is fitted on all brands in the cell. A deviation within ±3 points counts as noise.

All three are computed on product needs, from a brand’s competitors, with the brand left out of the curve. A dimension never sees the score it is meant to explain.

Movability of need n for brand b is the average, over the other brands j in the cell, of |observed_j − expected_j|, in points, with every expected value taken from a curve fitted without brand b. Low: under 3. Medium: 3 to under 6. High: 6 or more. It has no fitted parameter.

Beyond ±6 is measured with the brand left out of its own line. Chance comes from 20 datasets regenerated with binomial noise around the fitted curves.

Occupancy of need n for brand b sets how strongly the need is linked to brands against how many buyers care about it, in four steps:

1. Saturation s = the score that the curve fitted without b gives a brand with 20% of category buyers.

2. Across b’s product needs, fit a least-squares line:

logit(s) = a + c · prevalence

3. The gap is how far saturation falls short of what the need’s prevalence leads you to expect:

gap = (a + c · prevalence) − logit(s)

4. Within the brand, needs are ranked by gap and split into thirds: the top third is low occupancy, the middle third medium, the bottom third high occupancy.

The 20% reference is close to the median brand (25%) and inside the range of measured brands in 37 of 40 cells. At 10% or 30%, needs rank almost identically (rank agreement 0.968 and 0.981).

Prevalence is the share of category buyers who say a need matters when they choose. High: above 80%. Medium: 60–80%. Low: below 60%. The cut-offs are terciles of Behavio’s brand tracking benchmarks, not of this study, where 42% of product needs fall below 60%. We use prevalence for product needs only. Buyers rated price, place, and promotion items as well, but we leave those ratings out. Most buyers say price matters to them, so a high score there says little about the market. Most say promotion doesn’t, yet many other studies show that it does shape their decisions.

Across all brands, prevalence barely changes how often they break through: 12%, 13%, and 14% of pairs do, from the lowest to the highest third. It changes how often they fall back. Big brands lose on 22.0% of low-prevalence needs and on 5.9% of high-prevalence ones, 3.7 times as often. Mid-sized brands lose 2.1 times as often (13.2% against 6.4%). Prevalence also shows what a breakthrough is worth. Among 150 brands in Czechia, the Netherlands, and the UK, brands whose aware non-buyers link them to high-prevalence needs turn awareness into buying better (r = 0.29, p = 0.001).

The three are independent. Within a brand, the rank correlation of occupancy with prevalence is 0.01, with movability 0.05, and between prevalence and movability −0.15.

Breakthrough potential sorts a brand’s product needs into 27 combinations, 3 bands on each dimension, in the order its size sets. Small brands: movability, then occupancy, then prevalence. Mid-sized brands: the same, with low-prevalence needs moved to the end. Big brands: prevalence, then occupancy, then movability.

Brand size comes from penetration: small under 10%, mid-sized 10–29%, big 30% or more. Each product need gets a band on movability, occupancy, and prevalence (high above 80%, medium 60–80%, low below 60%). Its potential, #1 to #27, is the position of that combination in the order table for the brand’s size.

D. Breakthrough rates by brand size and need dimension

Base: 2,442 product brand–need pairs. A win is an earned association of +6 points or more; a loss is −6 or less. Each percentage is a share of the pairs in its row, with the number of cases in brackets.

For small brands, the first sort is movability, so D4 repeats D1. For big brands, it is prevalence, so D4 repeats D3.

A ratio counts as reliable when both sides have 10 or more cases, and as indicative with 5 or more. Below 5, it is not used.

Small brands have too few wins to carry a ratio of their own. Pooled with mid-sized brands, high movability gives 20.0% wins against 8.2% on low movability, 2.5 times, on 40 and 39 cases. Adjusted for brand size (Mantel-Haenszel), the ratio is 2.8.

The pair and triple tables (movability × occupancy, movability × prevalence, all three together) are in the data download. Too many of their cells hold fewer than 5 cases to print.

E. Robustness checks: sampling noise and the ±6 threshold

The deviations are not noise. The whole dataset was regenerated 20 times from the fitted curves with nothing but binomial sampling noise:

The observed spread of deviations is 5.66 points; sampling error accounts for 1.21. 95% of the variance is real.

The strict version makes the result stronger. With every brand left out of its own cell’s curve, deviations above +10 rise from 208 to 348. The paper reports the smaller numbers, except in three places that use the strict version, with each brand left out of its own line: the awareness gap in Section 2.1, the breakthrough rates in the movability table in Appendix C, and the hidden-brand test in Section 2.8.

Why ±6. The noise of a single measurement crosses no threshold from 3 points up. At +3, a quarter of all scores would count as wins. At +10, only 4% are left. ±6 is the lowest threshold where wins and losses cluster on needs clearly more than chance, with enough cases left for every combination.