How your existing users bring in new ones, with or without a formal referral mechanism.
In practice
Referral exists with or without a formal referral program — a happy user who spontaneously tells a colleague counts just as much. What sets a high-referral product apart is that it gets cheaper to grow over time: every new user brings in others, unlike paid acquisition where the cost stays flat (see CAC). A high NPS score is often the early signal that a referral mechanism, once built, will actually work — a poor NPS predicts the opposite, whatever the mechanism.
The formula
Referral share = new users who arrived through an existing user ÷ all new users in the period
- Arrived through an existing user
Anyone whose first visit traces back to a user: a referral code, a shared link with a reference, an invitation, or "a colleague told me" in the sign-up survey. Attribution is the whole difficulty — most referral happens in conversations you can't see.
- All new users
Including the paid and organic ones. The share tells you how much of your growth is free and self-generated; it is the number that should rise as the product matures.
- Viral coefficient (K)
The other lens on the same phenomenon: how many new users each existing user brings. Referral share reads the funnel from the outside; K reads it from the user's side. Both are needed, and the Tour asks about both.
Referral exists whether or not you built anything for it. What a referral mechanism does is make the invisible conversations visible and easier — a link worth sending, a reason to send it, a way to count it.
Worked example
Building the mechanism where the value already is
- A reporting tool notices in its sign-up survey that 30% of new users say "a colleague showed me a report". Nothing in the product supports that: reports are PDFs emailed by hand.
- It adds a shareable link on every report, with a rich preview and a "make your own" button — and a reference on the link so arrivals can be counted.
- Three months later: 1,000 new users a month, 410 traced to a shared report. Referral share: 41%, up from an estimated 30% — and now measured instead of surveyed.
- At the tool's €120 blended CAC, those 410 users are worth about €49,000 of acquisition a month — for a share button placed where people were already sharing.
The mechanism didn't create the referral; the product's value did. It made an existing behaviour easier and countable — which is what most "viral features" that work actually do.
Orders of magnitude
- Referral share varies from near zero (B2B tools used alone, in private) to a majority of growth (products whose output is inherently shown to others). A commonly cited signal that a mechanism is worth building is a high Net Promoter Score: people already recommending you in words will use a link if you give them one.
- Incentivised programmes ("give €20, get €20", made famous by Dropbox's extra storage and Uber's ride credits) work when the incentive is the product itself or something close to it; cash for a product people don't love mostly buys low-quality sign-ups.
- Referred users are usually worth more than average: they arrive pre-qualified by someone who knows them, activate faster and refer more in turn. That is why referral share matters beyond the CAC it saves.
How to improve it
- Find where sharing already happens — a survey question, support tickets, exports — and build the mechanism there, not in a menu nobody opens.
- Give people something worth sending: a result, a report, a personal artefact with a rich preview. A referral code is a favour asked; a shared result is a gift given.
- Make the recipient's first step trivial: land them on the thing that was shared, then on their own first value — not on a sign-up wall.
- Count it. A reference on every shared link, first-touch attribution, and a monthly look at referral share and K. "Only in communication, not the product" scores 7 in the Tour because a mechanism outside the product can't be measured or improved.
In the Tour
Does your product have a sharing or referral mechanism?
- Yes, built into the product20 pts
- Only in communication, not the product7 pts
- No0 pts
The three Referral questions climb one ladder: does a sharing or referral mechanism exist in the product (20 / 7 if it only lives in your communication / 0); is it actually used by customers, meaningfully; and is a viral coefficient or equivalent measured. It is the stage where a 0/20 is common and, unlike the others, often a decision rather than a blind spot — many good products simply never built one. The Tour's own share button, and the "take your own Tour" link on every shared result, are its answer to the first question.
Questions people ask
Is referral the same as word of mouth?
Word of mouth is referral you can't see: recommendations made in conversations, without a link or a code. Referral as a growth stage is the effort to make that behaviour easier, more frequent and measurable. A product with strong word of mouth and no mechanism is leaving its cheapest channel unmanaged.
Do referral programmes need an incentive?
Not necessarily, and the wrong one hurts. If the product's output is worth showing (a result, a design, a report), the share itself is the incentive and a reward mostly adds noise. Incentives make sense when the value is private and the ask is a favour — and they work best when paid in product (storage, credits, features) rather than cash.
What's the difference between referral and a growth loop?
Referral is one kind of growth loop — the viral one, where a user brings the next user. Growth loops also include content loops (what users create attracts search traffic) and paid loops (revenue funds the next round of acquisition). Referral is usually the cheapest loop to start, because it runs on a behaviour that already exists.