Brand Loyalty Is Mostly Brand Size
Brand loyalty feels like a relationship, but decades of buying data show it mostly follows brand size: big brands have far more buyers, each only slightly more loyal.
Key takeaways
- Andrew Ehrenberg, Gerald Goodhardt and Patrick Barwise set out double jeopardy in 1990: small brands have far fewer buyers, and those buyers are slightly less loyal.
- In British washing powder data from 2005, Persil's buyers bought it 3.9 times a year and Surf's 3.4, while Persil reached more than twice as many households.
- A 2017 review led by Charles Graham reported 32 further replications and no boundary condition for the pattern.
- Most of what feels like your brand loyalty to a big brand is what buying a widely available brand looks like.
You probably think of yourself as loyal to a few brands. The washing powder you always buy, the shampoo in the shower, the bank you've used for years. Brand loyalty feels like a relationship, and companies spend heavily on the assumption that it is one: points, perks, member pricing, emails that greet you by name. The research on brand loyalty suggests something less flattering to both sides. Most of what looks like your loyalty to a big brand is a by-product of the brand being big.
Start with the case for loyalty, because it was made well, it was made by people who had studied real companies, and it still runs a great deal of marketing.
What Reichheld and Sasser asked managers to do
In 1990 Frederick Reichheld and W. Earl Sasser published "Zero Defections: Quality Comes to Services" in the Harvard Business Review. Their argument was that service companies should treat every lost customer as a defect, track defections the way a factory tracks faulty parts, and work to drive them down. A customer kept, they reasoned, becomes more profitable each year: they buy more, cost less to serve and bring in others.
The article is widely remembered for a figure about how much profit a small cut in defections could add. We could not confirm its exact wording against the original, so we leave the number out. The argument does not depend on it. Retention sounds cheaper than acquisition, loyal customers sound like an asset, and a whole industry of loyalty programmes grew up around the idea.
Seen from the customer's side, the strategy shows up as the card in your wallet and the app on your phone. Each scheme assumes the same thing: that brand loyalty is something a company can build in a person, a little at a time, with rewards and attention, and that a customer who has been built up in this way is worth more than one who has not.
That assumption is reasonable on its face. People do have favourite brands, and some of them would be hard to tempt away. A customer who has bought the same coffee for a decade looks like proof that brand loyalty exists and that it can be won.
It is a sensible-sounding strategy. The difficulty is what the buying data say about where brand loyalty comes from.
Andrew Ehrenberg and colleagues
Decades of purchase records point the other way. The pattern was first described by the sociologist William McPhee in 1963, in awareness and liking scores, and Andrew Ehrenberg and his colleagues found the same pattern in what people actually buy. In 1990, the same year as the zero-defections article, Ehrenberg, Gerald Goodhardt and Patrick Barwise set it out in the Journal of Marketing as double jeopardy.
The pattern is this. A small brand has far fewer buyers than a big one, and those fewer buyers are also slightly less loyal. Small brands are hit twice. Big brands enjoy the mirror image: many more buyers, each a little more loyal.
Byron Sharp's How Brands Grow puts real numbers on it. Here is one of the tables he reproduces, United Kingdom washing powder in 2005, from household panel data:
| Brand | Market share | Bought it in a year | Purchases per buyer |
|---|---|---|---|
| Persil | 22 per cent | 41 per cent of households | 3.9 |
| Ariel | 14 per cent | 26 per cent | 3.9 |
| Bold | 10 per cent | 19 per cent | 3.8 |
| Daz | 9 per cent | 17 per cent | 3.7 |
| Surf | 8 per cent | 17 per cent | 3.4 |
Read across the rows. Persil's share is nearly three times Surf's, and almost all of that gap is in the second column: more than twice as many households bought it. Its buyers bought it only half a purchase a year more often. The loyalty difference is real and it is small. The difference in how many people buy is enormous.
Most of what looks like your loyalty to a big brand is a by-product of the brand being big.
Sharp shows the same shape in shampoo in Britain and the United States, in cars, where bigger brands lose a smaller share of their owners, and in Australian banks, where share varies forty-fold and the rate at which customers leave varies by about a factor of two. A 2017 review led by Charles Graham, in the Australasian Marketing Journal, drew on 179 peer-reviewed publications and 32 further replications, and reported finding no boundary condition for the pattern.
Byron Sharp and colleagues
If loyalty mostly follows size, the zero-defections strategy looks different. A brand's defection rate is largely set by its size and its category, so a small brand trying to cut defections is pushing against the arithmetic. Sharp's book makes the point with cars: wiping out one large brand's defections entirely would add only about one point of market share, while far more than that changes hands every year through ordinary switching. Growth, on this account, comes mainly from reaching more buyers, not from squeezing extra loyalty out of the ones you have.
Sharp is also candid about where the pattern bends. His own banking table runs mildly against double jeopardy, and he says so. The tradition documents two kinds of exception. Niche brands sell to fewer people who are more loyal than their size would predict, often because they are hard to find or made for a specific group. Change-of-pace brands are the reverse: plenty of people try them, few come back often, as with a seasonal flavour.
There is a serious argument on the other side as well. Research linking customer satisfaction to company performance finds a real association, and our own review of that literature records it as replicated, modest and highly conditional. The double jeopardy tradition does not deny it. Its objection is to the causal reading, that a company can raise satisfaction and so raise its share. Both positions are on the record. Our reading is that double jeopardy describes markets very well on average, and that the satisfaction research is a fair warning not to treat service quality as irrelevant just because loyalty mostly follows size.
The evidence is also heavily concentrated in one research tradition. Roughly a third of the book's sources come from Sharp's own institute or its lineage, and the 2017 review was written by researchers in the same school. That does not make the pattern wrong. It does mean the strongest claims about what managers should do with it have mostly been made by the people who discovered it.
What a loyal customer actually looks like
So what does this mean for you, the person who feels loyal?
Brand loyalty, for most everyday products, is less like a relationship and more like a habit spread across a small set of brands. People who buy Persil also buy Ariel; people who bank with one bank often hold a card with another. Your loyalty to a big brand is real in the sense that you buy it often. It is not, mostly, a reward the brand earned from you against the odds. It is what buying a widely available, frequently noticed brand looks like.
That changes how a loyalty scheme should be read from the customer's side, and how much of your own brand loyalty you should take credit for. On this account, the members of a large supermarket's points programme will be disproportionately its existing regular shoppers, who would have kept coming anyway. The scheme rewards the loyalty that size already produced. Whether it creates any new loyalty is a separate question, and the double jeopardy tradition is sceptical that it can create much.
None of this means service quality is irrelevant. A brand that treats people badly will lose them, and Sharp does not argue otherwise. The claim is narrower and more useful: across thousands of brands, the gap in brand loyalty between big and small brands is small, and the gap in how many people buy them is large. Which leaves an open question for the reader as much as the marketer. If the loyalty you feel is mostly the size of the brand looking back at you, how much of your weekly shop is choice, and how much is simply what was easiest to see?
Common questions
What is the double jeopardy law in marketing?
It is the pattern, described by Andrew Ehrenberg and colleagues, in which small brands have far fewer buyers than large brands and those buyers are also slightly less loyal. Large brands get more buyers and slightly more loyalty, so loyalty mostly tracks size.
Is brand loyalty real?
Yes, in the sense that people repeat-buy brands. But for most everyday products the differences in brand loyalty between competing brands are small, and much of what looks like loyalty to a big brand reflects how many people buy it.
Do loyalty programmes work?
The double jeopardy research suggests they mostly reward customers who were already buying often. They may have other uses, but the evidence that they create large new loyalty is weak.
Sources
Sharp, B. (2010). How Brands Grow: What Marketers Don't Know. Oxford University Press. Read in full by this publication, 2026-10-08. Ch. 2, double jeopardy, Table 2.2 (United Kingdom washing powder, 2005, household panel data); p. 25, banking; p. 34, car defection.
Reichheld, F. F., & Sasser, W. E. (1990). 'Zero defections: quality comes to services.' Harvard Business Review, 68(5), 105-111. Cited for its argument only. The widely quoted profit figure was not confirmed against the original and is not printed.
Ehrenberg, A. S. C., Goodhardt, G. J., & Barwise, T. P. (1990). 'Double jeopardy revisited.' Journal of Marketing, 54(3), 82-91. doi:10.1177/002224299005400307. Bibliographic record confirmed; cited for the formalisation and the name.
Graham, C., Bennett, D., Franke, K., Henfrey, C. L., & Nagy-Hamada, M. (2017). 'Double jeopardy, 50 years on: reviving a forgotten tool that still predicts brand loyalty.' Australasian Marketing Journal, 25(4), 278-287. doi:10.1016/j.ausmj.2017.10.009. Abstract and record only. Review base of 179 peer-reviewed publications and 32 further replications; reports no boundary condition.
McPhee, W. N. (1963). Formal Theories of Mass Behavior. Free Press. Credited for first describing the pattern, in awareness and liking scores. Secondary record only.
A note on scope: the satisfaction research is described from this publication's own review of that literature, which records the association as replicated, modest and highly conditional. Roughly a third of How Brands Grow's sources come from the author's own institute or its lineage.