Repeat purchase and referral are two different engines, and an enterprise that has never established which one it runs is allocating capital against an assumption it has not examined.
Take next year's revenue forecast and strike out every transaction that represents a customer's second or subsequent purchase. What remains is the part of the business that does not depend on anybody coming back.
For a subscription or consumables business the page is nearly blank, which is the correct result: it locates the entire risk in duration. For an enterprise selling something a buyer acquires once in a decade — a major professional engagement, a specialised installation, a capital replacement — most of the forecast survives, and the question becomes where those buyers came from. Somebody told them. That telling is a production process, and in most organisations nobody runs it.
The two engines demand different capital allocation, governance and indicators, and most enterprises instrument whichever one their industry's convention assumes.
The Strategic Context
Revenue that recurs and revenue that propagates are not variants of the same thing. In a repeat business, value accumulates in the duration and frequency of one relationship; the asset is tenure and the dominant risk is attrition. In a referral business, value accumulates in reach and credibility; the asset is a satisfied buyer's willingness to place their own standing behind you, and the risk is that acquisition cost must be recovered inside a single transaction plus whatever further transactions it credibly causes.
Almost every enterprise runs both engines in some ratio, and very few have measured it. The mix determines how much you may rationally pay to acquire a customer, over what horizon that payment is repaid, and which function should own growth. Misjudge it and you fund loyalty mechanics for buyers who will never return, or starve the pathway quietly producing most of your new business.
What the Convention Hides
Three misreadings recur, and they survive because nothing in a standard management pack tests them.
Lifetime value is treated as a measurement. It is a forecast resting on an assumed retention rate, an assumed margin and an assumed cost of capital, and it inherits the error in all three. As one figure in a board pack it acquires an authority none of its inputs deserves.
Revenue is used where margin is meant. Valuing a customer at gross spend is right only for a business with no cost of service, and the customers who spend most are frequently the customers who cost most to serve.
Referral is treated as free. It is paid for in delivery reliability, in service recovery, and in the margin foregone to make an experience worth recounting — simply from budgets other than the one marketing reports against.
Reframing the Issue
The useful question is not how to increase lifetime value. It is which engine this business actually runs, whether incentives and forecasts match it, and what would change if the answer proved to be the opposite of the assumption. That moves a marketing metric into capital allocation, where it belongs — because the decision it governs is not what to send customers, but how much to spend acquiring them, how long to wait for repayment, and which capability to build.
Rebuilding Lifetime Value on Something You Can Defend
Any lifetime value worth putting in front of a board should be built from three things: contribution margin per period rather than revenue, an explicit retention assumption, and an explicit discount rate. The third is most often omitted, and omitting it is not a simplification — it asserts that money received in five years is worth what money received today is worth.
Consider a deliberately hypothetical customer: contribution margin of $400 a year after cost to serve, annual retention of 80 per cent, and a discount rate of 12 per cent set by finance to reflect the cost of capital. On a standard perpetuity treatment that customer is worth $400 × 0.8 ÷ (1 + 0.12 − 0.8), or $1,000. Set the discount rate to zero and the same customer is worth $1,600. Nothing about the customer changed. A sixty per cent increase in an apparently objective figure was produced by an assumption nobody stated aloud.
The figures are illustrative and carry no claim about any real business. The discipline is the point: the discount rate is a finance decision, not a marketing one, and a lifetime value quoted without one should be returned to sender.
The Referral Engine Has a Cost, a Lag and a Ceiling
If new business arrives mainly by recommendation, three properties of that engine deserve board attention.
It has a lag. The interval between delivering an experience and receiving the referral it causes is long, variable, and routinely longer than a reporting period. Feedback loops with long delays are systematically under-funded, because the spending and the return never appear in the same document. That is a structural bias, not a failure of intent.
It has a ceiling. Any customer can credibly recommend you to a bounded number of people, and in a concentrated market those people overlap. A referral engine saturates locally long before the addressable market is exhausted, and the saturation presents as a marketing problem when it is a network problem.
It has an ownership gap. Acquisition sits with marketing, conversion with sales, recovery with service; the moment a satisfied customer is asked for an introduction sits between all three, which means nowhere. The engine also depends on buyers who neither complain nor recommend, and whose intentions are invisible to every instrument pointed at them [Related article: The Customers Who Generate No Signal].
What Each Engine Asks of the Organisation
A repeat engine rewards consumption: an offer designed so that ordinary use creates the next transaction. Its discipline is attrition management; its failure is complacency about a relationship that looks stable until it is not. A referral engine rewards memorability and the deliberate act of asking, and its discipline is designing delivery that survives description by somebody with no incentive to exaggerate. What makes an experience worth describing is usually the removal of an irritation the buyer expected to endure, which is a distinct subject with its own economics [Related article: Which Frictions Are You Profiting From Tolerating?].
The failure mode is the enterprise that pursues both at moderate intensity and builds neither: a loyalty scheme in a category bought once a decade, and a referral incentive in a subscription business whose real lever is the reason people cancel.
Decision Framework
Establish the mix rather than assuming it. Three tests do most of the work.
The repeat test. Across a period at least as long as the natural repurchase interval, what share of revenue came from buyers who had purchased before — buyers correctly matched across systems, not accounts?
The origin test. Of customers acquired in that period, what share traces to an identified existing customer? If the records cannot answer, that is the first finding.
The recovery test. For one transaction with no repeat and no referral, does contribution margin exceed fully loaded acquisition cost? If not, the business is solvent only on the strength of an engine nobody has demonstrated it runs.
| Repeat engine | Referral engine | |
|---|---|---|
| Acquisition spend justified by | Discounted margin over tenure | Transaction margin plus expected referrals |
| Payback horizon | Several periods | Within the transaction, or on faith |
| Primary indicator | Cohort attrition | Referral rate and origin traceability |
| Dominant risk | Tenure shorter than assumed | Referral rate falls invisibly |
Payback horizon is a cash question as well as a returns question, and the two diverge sharply when volume moves [Related article: Can Your Customers Fund the Business — and What Breaks When Volume Falls?].
From Strategy to Execution
Immediately, settle the arithmetic. Agree with finance a single construction of lifetime value, withdraw the competing versions in circulation, then run the three tests and report the mix honestly.
Over two or three quarters, build the instrument the mix implies. For a repeat engine that means attrition measured by cohort, because an aggregate rate hides a deteriorating recent cohort behind a stable older one. For a referral engine it means recording origin at first contact, while the customer still remembers, and making one person accountable for the act of asking.
Over years, positioning follows. A referral business should be building the delivery capability that produces referable outcomes and accepting the margin cost; a repeat business should be extending the reasons to stay. Both are routinely deferred, because the engine was never named.
Signals to Monitor
- The share of new customers whose origin traces to nothing specific, which measures the referral engine's invisibility rather than its size.
- Cohort attrition diverging from aggregate attrition, the earliest warning that tenure assumptions are stale.
- Rising acquisition cost accepted on the strength of a lifetime value whose discount rate nobody present can state.
- Repurchase interval lengthening, which reads as a demand problem and is often a relevance problem.
Questions for the Leadership Team
- What share of last year's revenue came from buyers purchasing at least a second time, and how far do we trust the matching behind the number?
- If every customer bought once and told nobody, at what volume would this business break even?
- Which construction of lifetime value does this organisation use, and what discount rate sits inside it?
- Who is accountable for the referral pathway, and what would their plan cost to fund properly?
- What would we stop doing tomorrow if we established that we run a referral business rather than a repeat one?
Closing Perspective
An enterprise cannot run both engines at full intensity, and the one it neglects is often the one it depends on. That neglect is rarely a decision. It follows from a metric built on revenue rather than margin, a payback horizon nobody wrote down, and a growth channel that issues no invoice and so attracts no owner.
The question in the title is not rhetorical. Every enterprise should be able to answer it with a number produced by finance rather than marketing. If no number can be produced, that absence is itself the finding.
What remains is the choice — build the engine you have, or stop paying for the one you do not.
About the author
Kevin Jogin is Founder & Principal Advisor at EraNorth. Meet the Founder.