Referral marketing

Is the product ready for referral?

Two gates, and both are checkable this week.

The curve has to flatten. If each signup cohort keeps falling toward zero, a referral program manufactures a larger cohort that falls toward zero, faster, and you pay a reward for every one of them. Draw the curve first; how to read a retention curve is 30 minutes of spreadsheet work and it settles the question.

There is a second, less obvious reason to wait. The German retail bank studied by Van den Bulte, Bayer, Skiera and Schmitt, published in the Journal of Marketing Research in 2018, showed that referred customers' retention advantage, roughly a 40% lower churn hazard, held only while the person who referred them was still a customer. When the referrer left, the advantage went with them. Referral built on a leaky product does not just add churn, it adds churn in correlated clumps.

There has to be a moment you can name. Not a feeling that customers are happy. A specific event, with a timestamp, where the product's promise is kept: the report ran, the order arrived, the test passed, the first invoice got paid. If you cannot name that event and query it, you do not yet have anywhere to put the ask, and the ask will end up in the footer and the monthly email, which is where referral programs go to die.

The upside of getting it right is well documented. In the same bank's data, studied earlier by Schmitt, Skiera and Van den Bulte in the Journal of Marketing in January 2011 across 5,181 referred and 4,633 non-referred customers, referred customers churned about 18% less and were worth roughly 16% more over six years. Worth noting honestly: the margin advantage in that study eroded after about 29 months as the other customers caught up, while the retention advantage persisted. The lasting benefit of a referral is who stays, not what they spend on day one.

Where is the moment someone wants to share?

Almost never where the referral button is. The button is usually in account settings, and the moment is usually in the product, minutes after something worked.

Finding it takes two passes, one in the data and one in conversation. In the data, look at what already happens without any program: exports, shared links, invites to a workspace, forwarded reports, support threads where somebody copies a colleague. Those are people telling other people, with no incentive, and they point at the moment. In conversation, ask ten customers two questions: who is the last person you mentioned us to, and what were you doing when you did it. The second question gives you the moment. The words they use in answering give you the copy for free, and it is better copy than you would have written.

Then put the ask inside that moment, in the surface where the moment happens, within minutes of the event rather than in next week's send. The referral moment walks the whole build, step by step, including the version for B2B, where the moment is often a person changing jobs rather than a product event.

What should the incentive be, if any?

Three honest choices, and the order below is the order to try them in.

No incentive, with the friction removed. Make sharing one tap at the right moment, make the shared thing worth receiving, and make the referred person's arrival feel different from a cold signup. For products with a strong moment this is often enough, and it never attracts the wrong people.

One-sided, to the person arriving. A better first experience for the referred person. The referrer gets to look generous, which is the thing many of them actually want.

Two-sided, to both. The version with the clearest public evidence behind it. In Drew Houston's 2010 Startup Lessons Learned deck, Dropbox reported that a two-sided referral program permanently increased signups by 60%, that 35% of daily signups came from the program, and that users sent 2.8 million direct referral invites in the trailing 30 days to April 2010, over a period in which registered users went from 100,000 in September 2008 to 4,000,000 in January 2010. The reward was storage, which is to say the product itself, given to two people who both wanted more of it.

That last detail is the whole lesson, and it is why cash is usually the wrong answer. An incentive does not only change how many people refer. It changes which people refer, and a reward worth having to anyone at all attracts people who want the reward rather than people who like the product. When a referral program is just a coupon is the argument in full, with the three checks that tell you which kind you have built.

How do you measure word of mouth?

By asking one open question at the right time, and by comparing cohorts, not by reading a dropdown.

The demand side is not in doubt. Nielsen's Global Trust in Advertising survey of more than 29,000 consumers across 58 countries, fielded in February and March 2013, found 84% said they completely or somewhat trusted recommendations from people they knew, ahead of branded websites at 69% and consumer opinions posted online at 68%. McKinsey Quarterly, in April 2010, put word of mouth as the primary factor behind 20% to 50% of purchasing decisions. When a referral program underperforms against numbers like those, the mechanism is usually at fault rather than the appetite.

What goes wrong in measurement is narrower than people think. A "how did you hear about us" dropdown at signup, with your existing channels as the options and no open text box, asked before the person knows what to call you, will under-report word of mouth every time and then be used as evidence that word of mouth does not happen. Measuring word of mouth gives the question, the timing, and the four numbers it produces, with the formulas.

What good looks like: a retention curve that flattens, one named value event with the ask sitting inside it, a referred-customer cohort whose 90-day retention beats your non-referred cohort from the same weeks, and a word-of-mouth share you track as a trend rather than quoting as a channel.