What the science says

Consensus

The Ehrenberg-Bass view starts from a deliberately deflationary claim: for most brands in most categories, growth is not won by making people love you more, but by making your brand easier to think of and easier to buy for a larger number of people. Byron Sharp's How Brands Grow (2010) is the field-defining statement, built on empirical work going back to Andrew Ehrenberg's Repeat-Buying (1972). The two levers are mental availability, the probability that your brand is retrieved from memory in a buying situation, and physical availability, the sheer ease of encountering and purchasing it, and the growth engine underneath them is penetration: reaching more category buyers, most of whom buy the category only occasionally, rather than deepening the loyalty of the heavy few.

‍

Mental availability is not left as an abstraction; it is operationalised through category entry points, the cues that call a category and its brands to mind (Romaniuk and Sharp, 2004; Romaniuk and Sharp, 2016). These are the occasions, needs, locations, moods, motives and people associated with using the category, "something quick before work", "a treat on a Friday", "when the kids are hungry", and a brand builds mental availability by becoming linked, in memory, to more of these cues, more strongly, among more buyers. The practical corollaries are distinctive assets, the colours, logos, characters and slogans that let a brand be recognised fast, and broad, consistent reach in media.

‍

What sets this tradition apart from most of marketing is the strength of its empirical spine. The double jeopardy law states that brands with smaller market share are punished twice: they have fewer buyers, and those buyers are also slightly less loyal, so measured loyalty turns out to be largely a function of a brand's size rather than of any special affection for it (Ehrenberg, Goodhardt and Barwise, 1990). The Dirichlet model goes further, reproducing a category's whole pattern of buying, penetration, purchase frequency, how buyers split their spending, from little more than the brands' market shares, and it serves as the benchmark for telling a genuine effect from ordinary statistical noise (Ehrenberg, Uncles and Goodhardt, 2004). These patterns hold across product categories, countries and decades, which makes them among the most reliably replicated findings in the whole of marketing science, a real rarity in a field full of untested stories.

Controversies

The framework is, by design, a direct challenge to the dominant tradition in branding, and that is the first controversy. The positioning and brand-equity schools, associated with work like Keller's on customer-based brand equity (Keller, 1993), hold that brands grow by building distinctive meaning, differentiation and an emotional relationship in the customer's mind. Ehrenberg-Bass replies that when you look at the purchase data, differentiation, loyalty programmes and emotional bonds mostly do not behave like growth drivers, and that brands of similar size look far more alike in how they are bought than the meaning-based story predicts. The disagreement is not a minor turf war; the two camps recommend genuinely different spending.

‍

A vivid illustration of the tradition used as a myth-buster is recent work on "sustainable" brands. Using the double jeopardy law and the law of brand user profiles across UK data, Sharp, Wheeler and Nenycz-Thiel (2023) showed that brands marketed as sustainable are not bought by a distinct, more loyal type of shopper; they behave like ordinary brands of the same size, which quietly demolishes a popular belief that a "green" or purpose position wins unusually devoted customers. (This is Anne Sharp, a separate Ehrenberg-Bass researcher, not Byron Sharp.) That is the framework at its most useful: an empirical check on a comfortable marketing story.

‍

The second controversy runs the other way, against over-claiming. Even the celebrated laws are not exceptionless. Scriven, Bound and Graham (2017), reviewing sixty-two packaged-goods categories, catalogue the recurring deviations from Dirichlet predictions and show they are not random noise but interpretable, systematic departures, some reflecting non-stationary markets, some reflecting genuine partitioning of a category into sub-markets, some the familiar niche and change-of-pace brands that sit off the double jeopardy line. Their point is balanced: brand performance stays remarkably close to the Dirichlet benchmark, but "remarkably close" is not "always", and a serious user of the framework treats the deviations as information rather than pretending they are not there.

Limitations

The sharpest limitation is scope. The laws were established, and are strongest, in mature, stationary, frequently-bought consumer-packaged-goods markets, and they are weaker or contested precisely where buying works differently: in category creation and fast-changing markets, in luxury and heavily identity-loaded purchases where a small, devoted segment really can drive disproportionate value, and in business-to-business, services, and subscription or platform businesses where individual switching costs, contracts and identity matter far more than they do for a tin of beans. In those settings the confident "grow by penetration, ignore loyalty" prescription can be simply wrong. A second limitation is that the framework deliberately downplays the psychology of attitude, identity and meaning, treating it as largely epiphenomenal to buying, which is a defensible modelling choice for aggregate CPG data but leaves it thin on the cases where meaning demonstrably moves behaviour. And there is a discipline problem in how the framework is used: its laws are so quotable that they get applied well outside the conditions that produced them.

Open questions

The open questions are mostly about boundaries and prediction. Exactly where and why do the laws break, and can the systematic deviations be turned into a positive theory of when a category will partition or destabilise (Scriven, Bound and Graham, 2017)? Do category entry points predict growth prospectively, so that a brand deliberately building links to more entry points reliably grows faster, rather than the measure simply describing brands that have already grown, since the entry-point methodology is comparatively new and still becoming a standard? And how should the framework be reconciled with the genuine cases, luxury, fandom, mission-driven and identity brands, where loyalty and differentiation clearly do carry weight, without either camp pretending the other has nothing? The most honest current position is that the tradition has unusually strong laws with unusually clear boundary conditions, and that the interesting work is at those boundaries.

So what

Treat this framework as your empirical default and your hype filter. Its greatest practical value is negative: it is the most reliable check available on the marketing industry's habit of attributing growth to purpose, emotional connection or loyalty, mechanisms that sound compelling and mostly do not survive contact with purchase data. So the default playbook is to grow by reaching more category buyers, especially the light and occasional ones, to build mental availability by owning more of the category entry points, the real cues and occasions that bring the category to mind, and to make the brand physically easy to buy and instantly recognisable through consistent distinctive assets. Track brand health by penetration and entry-point coverage rather than by love scores. Then hold the whole thing inside its boundary: this is close to law in mature, frequently-bought categories, and only a loose analogy in luxury, identity, business-to-business and subscription markets, so calibrate accordingly.

‍

For companies and brands, one scope note comes first, because it decides how much of this to trust. The concrete playbook below is a consumer-brand playbook, aimed at products and services sold to large numbers of individual buyers, and it is at its strongest for everyday, frequently-bought ones. If you sell business-to-business, industrial supply, or a relationship-heavy service where a handful of accounts and long contracts drive the revenue, this is a different world, and the prescriptions here should be treated as loose analogies at most rather than rules. With that boundary set, the moves for a consumer brand follow directly. Favour broad reach over narrow targeting, because most of your future buyers are light buyers who rarely think about the category. Audit your category entry points: list the cues and occasions buyers actually use, see which ones your brand is linked to and which are unclaimed, and build toward the gaps. Invest in distinctive assets and use them consistently, so recognition is instant. And apply real scepticism to loyalty-programme returns and to "purpose" as a growth lever, not because they never matter but because the data rarely support them as the engine of growth. The same caveat closes the loop: in business-to-business, in a relationship-driven service, or in a genuine subscription, platform or luxury business, weight loyalty, retention and differentiation far more heavily, because there the switching costs, contracts and identity effects the framework downplays are real.

‍

For political parties and campaigns, there is a useful and carefully-bounded analogy. A party or candidate needs to be mentally available at the moment of decision, linked to the issues, moments and identities that bring politics to mind, and broad reach across the wider electorate usually matters more than deepening the enthusiasm of an already-committed base, which is the political echo of penetration over loyalty. Distinctive assets, consistent colours, symbols, faces and slogans, help a party be recognised and recalled fast. The boundary here is important and honest: elections are winner-take-all, intensely identity-loaded and often turn on turnout, so the packaged-goods laws travel only loosely, and the base-versus-reach balance is a genuine strategic judgement rather than a settled law.

‍

For government and public communication, the lens is practical for any service or behaviour you need people to choose or use. Be mentally available at the moment of need, linked to the situations and cues when the service becomes relevant, "when you lose your job", "when the baby arrives", and physically available, genuinely easy to access, since a service people cannot easily reach or recall will be underused whatever its merits. Track reach and coverage of the relevant entry points rather than satisfaction alone.

‍

The dual-use point here is mild and mostly about honesty in your own house. The framework's real gift is that it is falsifiable and frequently falsifies pleasant marketing beliefs, so the responsible use is to let it discipline your spending and puncture your own hype, while resisting the temptation to wield its laws as slogans in categories they were never tested on. Using robust laws outside their range is its own kind of overclaiming.

‍

The summary is that mental and physical availability, made concrete through category entry points, give brand growth an unusually solid, data-backed spine: be easy to think of and easy to buy, for as many people as possible, and distrust the stories that say love and differentiation are what grow you. Keep that as your default and your corrective. But keep it inside its evidence: iron law in mature packaged-goods markets, useful analogy with clear exceptions elsewhere, and never a reason to ignore the loyalty and identity effects that genuinely drive some categories.

‍