Duplication of Purchase Law

Law 03 / 20 Grow the Brand

A brand shares its buyers with every rival in the category, roughly in proportion to each rival's size, so few customers are anyone's exclusively.

  • Empirical regularity
  • 6 min read
  • Documented study
  • Reviewed by Ali Dezhami, November 8, 2026

The dilemma

Illustrative scenario

Tomás leads marketing for a mid-sized dry dog-food brand sold in pet stores and online. A competitor has just launched a grain-free line backed by a heavy video campaign, and the CEO wants to hit back: a "conquest" campaign with steep switching coupons aimed at the rival's customers, bought as a retail-media audience.

Before approving it, Tomás asks an analyst to pull a year of basket data from their largest retail partner. The result unsettles the room. Most of their buyers also bought the market leader. A sizeable share also bought the grain-free rival, more than its modest size would suggest. Only a small minority bought nothing but Tomás's brand.

The CEO calls it a loyalty crisis. The analyst calls it normal. If customers are this shared, how many coupons would land on people who already buy from them, and who exactly would be conquered?

The short answer

The Duplication of Purchase Law says that a brand shares its buyers with every other brand in its category, roughly in line with each rival's penetration: the more people buy a competitor, the more of your buyers also buy it. The pattern comes from Andrew Ehrenberg and colleagues' research on consumer purchase panels and was popularized for practitioners by Byron Sharp's How Brands Grow (2010).

How strong is the evidence?

Empirical regularity

A pattern repeatedly observed in market data. It describes what usually happens; the exact numbers vary.

This is an empirical regularity. Over a given period, the share of brand A's buyers who also buy brand B tracks B's penetration, and a single multiplier (often called the duplication coefficient) fits most brand pairs in a category reasonably well. Ehrenberg, Uncles and Goodhardt (2004) show how the Dirichlet model of repeat buying supplies expected values, or benchmarks, for duplication and related measures such as penetration, purchase frequency and sole-brand buying, so managers can see which brands depart from the norm.

Two practical implications follow. Sole-brand loyalty is uncommon over a reasonable time window, and the buyers who are solely loyal tend to be light buyers of the category. And a competitor's customers are not a separate, exclusive pool: a rival's buyers include some of yours, roughly in line with your own penetration, just as your buyers include many of theirs.

Where it bends: brand pairs that share more buyers than expected usually have something in common, such as format, price tier, function or retailer, which suggests a partition in the market. Pairs that share fewer may serve different needs or simply not be available in the same places. These deviations are the most useful output of the analysis. The law describes the typical pattern; it does not say every brand competes equally with every other.

Investigating an example

Documented study

Benchmarks for how much brands should overlap

Ehrenberg, Uncles and Goodhardt set out how managers can compare observed brand performance measures, including duplication of purchase between brands, with the values the Dirichlet model predicts. In this framework most brand pairs share customers close to the predicted level, and the signal worth chasing is where observed duplication departs from the benchmark, which can point to market partitions or distribution gaps.

Limits: the benchmarks assume a fairly stable market measured over a fixed period, and they require panel or transaction data that covers rival brands, which many companies do not have. A deviation is a clue to investigate, not a diagnosis. Category definition matters too: draw the category too broadly and you will manufacture partitions that are really separate categories.

A published study or record, linked below. Its limits are stated.

Choose your move

You are in the dilemma above. Which move would you make? Open each one to see its trade-offs.

A Send steep switching coupons to the rival's customer audience.

Part of that audience already buys your brand, roughly in line with your own penetration, so some of the discount subsidizes purchases you would have had anyway. It may still shift share among shared buyers, but only a holdout comparison will show whether the coupons pay.

B Treat the category as one shared pool: reach all category buyers and protect availability where the rival is gaining.

Consistent with the law and avoids paying to re-win your own customers. It is slower to show results and less satisfying for a CEO who wants a visible counterattack.

C Run a duplication analysis first, then focus on the rival you share more buyers with than expected.

Turns the law into a diagnostic and may show the grain-free line is a closer competitor than its size suggests. It costs time and data, and the partition may be too small to justify a separate program.

No move wins in every business. The law tells you what to check, not what to choose.

How to apply it

  1. Build a duplication table. For one period, usually 12 months, record the share of your buyers who also bought each rival. Panel, retailer or loyalty-card data work best.
  2. Estimate the expected level. Divide each observed duplication by the rival's penetration and take a typical value of that ratio as your coefficient; expected duplication is the coefficient times the rival's penetration.
  3. Flag material deviations. Look for pairs well above or below expected and ask what they share or lack: format, price tier, need state or channel.
  4. Check sole buyers. Calculate the share of your buyers who bought only your brand, and whether they are mostly light category buyers.
  5. Match the action to the finding. Use excess duplication to choose close competitors for comparative messaging, shelf placement or pricing tests, and keep broad reach for everyone else.

What to measure

Duplication with each rival compared with the expected level, sole-buyer share and penetration over the same period. Guardrail: for any conquest or switching offer, check how many recipients already buy your brand and measure incremental purchases against a holdout.

Worked example

Worked example: which other brands our buyers bought over 12 months (illustrative numbers)
Rival brandRival penetrationOur buyers who also bought itExpected at coefficient 1.5Reading
Market leader 40% 58% 60% In line
Rival B 20% 31% 30% In line
Grain-free rival 10% 27% 15% Shares more than expected
Value brand 10% 9% 15% Shares less than expected

Illustrative numbers, not data from real brands. The coefficient of 1.5 is assumed for clarity; in practice it is estimated from the observed pairs. The grain-free rival behaves like a close competitor despite its size; the value brand may serve a different shopper or be stocked elsewhere.

Where it breaks

  • Partitions are real. Format, price tier, function and channel can make some brands compete far more closely than size predicts.
  • The time window changes the numbers. Short periods show less overlap and long periods show more, so compare like with like.
  • Category definition matters. Mixing products that meet different needs creates false partitions.
  • Some markets are used one brand at a time. Contract services such as a mobile plan show low overlap in a short window even though customers switch over time.
  • Data blind spots. Your own CRM cannot see purchases of rival brands; surveys rely on imperfect recall.
  • Ethics: buying audiences of a rival's customers must respect consent and privacy rules; the law is no justification for questionable data sourcing.

What if you apply it incorrectly?

A marketer takes the law too literally, assumes every rival is equally substitutable and benchmarks only against the market leader. Meanwhile the grain-free brand, which shares far more of her buyers than its size predicts, keeps winning their purchases with a format she does not offer. The law provides the baseline; ignoring the deviations hides the competitor that actually matters.

Myth versus evidence

The social-video version

Your customers are loyal to you, and your competitor's customers are a separate group you have to steal.

What the evidence supports

Buyers are shared: most of your buyers also buy the big rivals, and each rival's buyers include yours in line with your size.

Five-minute field assignment

Using your largest retail partner's report, a marketplace dashboard or a brand-usage survey, find the share of your buyers who also bought each of your three biggest rivals in the last 12 months. Divide each figure by that rival's penetration. Are the ratios roughly similar? Write down the rival with the highest ratio and one reason it might be a closer competitor.

Checkpoint

Two questions: one about the idea, one about a decision. Answer both correctly to master this law. Questions are drawn from a larger bank and graded on our server; a retry brings different questions.

Evidence ledger

Every factual claim in this lesson, where it comes from, and how far it has been checked.

Further reading

Which other law could explain this decision?

Editorial review: Ali Dezhami, November 8, 2026. Last substantive update: November 8, 2026. Our methodology