What double jeopardy is
The pattern has been documented across many categories by marketing scientists, including researchers associated with the Ehrenberg-Bass Institute. In a category of broadly similar, competing brands, the brands differ mainly in how many people buy them in a period, their penetration. They differ only modestly in how loyal those buyers are, and the difference runs in a predictable direction: the brand with fewer buyers also gets bought slightly less often by them, and gives up more of their category buying to rivals.
That second penalty is what surprises people. Intuition suggests a small brand might make up for its size with a devoted following, and some marketing plans are built on that hope. The pattern says the opposite is usual: big brands tend to have both more buyers and slightly more loyal ones, and small brands have both fewer buyers and slightly less loyal ones.
What the pattern makes visible is that loyalty and size are tied together. A loyalty figure read on its own looks like a measure of how much buyers love the brand; read against the brand's penetration, it often turns out to be roughly what a brand of that size would be expected to have. What the pattern hides is the mechanism and the exceptions. It describes what happens on average without explaining why for any one brand, and it does not apply equally to every kind of market.
The pattern
| Measure | Larger brand | Smaller brand | What it means |
|---|---|---|---|
| Penetration: the share of category buyers who buy the brand in a period | Many | Far fewer | The main difference between brands of different size lies here |
| Purchase frequency among the brand's own buyers | Slightly higher | Slightly lower | The second jeopardy: the small brand's buyers buy it a little less often |
| Share of category buying its buyers give it | Somewhat higher | Somewhat lower | Even the small brand's own buyers spend more of their category buying on rivals |
| Repeat purchase over successive periods | Somewhat higher | Somewhat lower | Loyalty measures move with size, so they are hard to raise in isolation |
| Mix of light and heavy buyers | Mostly light buyers, with a heavy minority | Also mostly light buyers | Both brands depend on many occasional buyers, not on a devoted core |
The first row carries most of the difference. The rows below it show loyalty measures that do vary, but gently, and in the same direction as size. The last row matters for planning: in both large and small brands, most buyers buy the brand only occasionally in any period. A brand's buyer base is not a loyal core with a few strays but a large number of light buyers, many of whom also buy competitors, alongside a smaller group of heavier ones.
Reading it in your own data
Reading the pattern needs category-level data, because the comparison is between brands. Consumer panels and some retailers provide penetration, Purchase FrequencyAverage number of orders each customer placed in the period. and share of category buying by brand for the same period. Rank the brands in the category by penetration and look at how the loyalty measures change down the list. If double jeopardy holds, they will fall gently as penetration falls.
Then place your own brand on that line. A brand whose loyalty sits roughly where its size predicts has a normal loyalty problem, which is really a penetration problem. A brand well above the line may have a genuine niche: a distinct product, a regional base or a channel few rivals use. A brand well below it may be bought mainly for variety or on promotion, which is worth investigating before planning more promotions.
Keep the period fixed and comparable. Over a longer period more people buy every brand at least once and each buyer buys more often, so a quarterly figure cannot be compared with an annual one. Seasonal categories need like periods compared with like.
Businesses without panel data can still read part of the pattern. Their own customer records show how often customers buy and the Repeat Purchase RateShare of customers who placed more than one order in the period. between periods, though not how many category buyers never bought at all. An estimate of the category's buyers, from a market sizing exercise or category data, gives a rough penetration figure. The comparison with rivals is then a judgement rather than a measurement, but it is still a useful check before setting a loyalty target.
What it means for loyalty and growth plans
The first implication concerns targets. A small brand that sets out to match the leader's loyalty is setting a target that, on this pattern, comes mainly with the leader's size. Loyalty targets should be set relative to the brand's penetration, and an improvement in loyalty that matches a rise in penetration is the pattern working normally rather than a separate achievement.
The second concerns where growth comes from. If loyalty mostly follows size, the main route to growth is more buyers, and loyalty tends to rise a little as the brand grows. The comparison of customer acquisition and retention weighs the two investments for different business models; double jeopardy is one reason that, in frequently bought categories, acquisition of new and light buyers carries more of the growth than deepening the loyalty of heavy ones. That puts ReachUnique people who saw an ad. among the people who rarely or never buy the brand at the centre of the plan, which is the subject of the page on mental and physical availability.
The third concerns loyalty spending. Programmes that reward purchase frequency tend to reward the heavy buyers the brand already has. The guide to designing a loyalty programme explains how to target incremental behaviour rather than paying for purchases that would have happened anyway, and the guide to repeat purchase and loyalty covers how online retailers can measure loyalty without flattering themselves.
None of this means retention can be ignored. Customers who leave because of poor service, broken delivery or a bad product are lost for reasons the brand controls, and a rising Churn RateShare of customers at the start of a period who left during it. is a warning whatever the brand's size. The pattern speaks to how much loyalty a brand can expect from normal buying, not to losses caused by failures. For frequently bought goods, the FMCG marketing guide shows how these ideas shape media, retail and measurement in practice.
Worked example (illustrative)
A sparkling-water brand reconsiders a loyalty push
Fizzlebrook Sparkling Water, an invented brand, plans a £400,000 loyalty push, including a points app, after its marketing team reports that its buyers purchase it only 2.6 times a year against 4.1 times for the category leader. Before committing, it buys panel data for the category. The leader is bought by 24% of category households in a year, the second brand by 15% and Fizzlebrook by 5%. Purchase frequency falls down the list in step: 4.1, 3.6 and 2.6. Plotted against penetration, Fizzlebrook's figure sits almost exactly where a brand of its size would be expected to sit, so its buyers are not unusually disloyal. The brand moves £300,000 of the budget to reaching households that have never bought it, through broader video and retail media and wider listings in convenience stores, and keeps £100,000 for a simpler programme that targets lapsed buyers. It sets a penetration target of 7% and expects frequency to rise only slightly if that target is met.
Limits and criticisms
The pattern is strongest in repeat-purchase categories with many similar brands, such as packaged groceries and household goods, where buyers hold a repertoire and switch freely. It is weaker in niche markets, in subscriptions, where frequency barely varies, in luxury, where scarcity is part of the appeal, and in highly differentiated markets, where brands are not close substitutes. A business in those markets should test the pattern against its own data before planning around it.
It describes an average, not a law any one brand must obey. Calling it a law suggests more certainty than an empirical regularity can carry, and individual brands do sit above or below the line for real reasons. Using the pattern to dismiss any loyalty work, or to excuse a brand whose loyalty is genuinely below what its size predicts, misreads it.
It says nothing about how to gain penetration. Knowing that growth mostly comes from more buyers does not tell a brand how to reach them, what to say, where to be present or what it will cost. That work belongs to brand, media and distribution planning.
The data needed to read it properly is not always available. Panel data is costly and covers some categories better than others, and a business reading only its own records sees its buyers but not the category buyers it has never reached. Conclusions drawn from partial data should be held loosely.
Common mistakes
- Reading loyalty on its own. A loyalty figure means little until it is compared with the brand's penetration and with rivals of different size.
- Setting the leader's loyalty as a target. A smaller brand that aims for the leader's loyalty is mostly aiming for the leader's size.
- Pouring budget into heavy buyers. Most of a brand's buyers are light, and most of the growth sits with them and with people who have never bought.
- Treating the pattern as a law. It is an average; brands above or below the line deserve investigation, not dismissal.
- Using it to excuse churn. Customers lost to failures in product or service are a fixable problem, not the pattern at work.
- Applying it everywhere. Subscriptions, luxury and highly differentiated markets need their own evidence.
