14. Referral and expansion experiments

§The question: will users bring other users, and will existing users pay more as they get more value, and has the product earned the right to ask?

§Referral and expansion are the compounding stages, the ones that turn a good product into an engine. They are last in the funnel and last in this part for the same reason: both are built entirely on retention, and asking for them before the product retains backfires.

§ 14.1Referral: the purest loop#

§A referral loop is the cleanest example of the growth loop from chapter 10: a user's use of the product produces a new user, who does the same. When it works, growth compounds with no acquisition spend, which is why every team wants it and why so many build it too early.

§It only works on a product people already love. A referral is a recommendation, and a user will not stake their reputation on a product that has not earned it. Sean Ellis's rule is that referral amplifies existing love; it does not create it. This is why the validation signals, especially retention, are the precondition: a referral program on an unretained product asks users to recommend something they are about to abandon, and they will not.

§The design that works is usually double-sided, value for the referrer and the referred, and built into the moment of value rather than bolted on. Dropbox's referral, extra storage for both sides, is the canonical example because the reward was the product itself, given at the moment the user understood why more of it was worth having. The referral was an extension of the value, not a bribe attached to it.

§ 14.2Expansion: the highest-return growth#

§Expansion is existing users paying more over time as they get more value: more seats, more usage, a higher tier, an added product. For many businesses it is the highest-return growth of all, because the customer is already acquired, already activated, already retained, and the only question is whether the value has grown enough to justify more spend.

§Net revenue retention captures it: whether a cohort's total revenue grows or shrinks over time, after churn and expansion net out. A cohort whose expansion outpaces its churn grows revenue with zero new customers, which is the strongest position a small company can reach. Expansion experiments test the triggers: at what moment of value does an offer to expand convert, and does the expansion retain or churn.

§ 14.3The anatomy applies, with a trust guardrail#

§Both are experiments with the Part I anatomy, and both carry a guardrail that ordinary experiments do not: trust. A referral ask at the wrong moment, or an expansion push that feels like a squeeze, damages the retention it depends on. So the guardrail metric for every referral and expansion experiment is retention itself: if the experiment lifts referrals but dents retention, it lost, because it spent the foundation to juice a stage built on it.

§ 14.4When it is too early#

§The honest answer for most first-year products is that referral and expansion are too early, and building them is a distraction from retention. The signal that it is time: the retention curve is flat and high, the must-have score is strong, and users are already referring or asking to expand on their own, unprompted. Organic referral before any program is the green light to build a program. Its absence is the signal to go back to retention, because a program cannot manufacture love that is not there.

§ 14.5What you leave with#

§An honest verdict on whether the product has earned referral and expansion, from the retention curve and from whether users refer unprompted. If yes: referral built into the moment of value, double-sided, with retention as a guardrail; expansion triggered at moments of grown value, measured by net revenue retention. If no: the knowledge that the next work is retention, not a referral program. The next chapter is how to choose among all these experiments.