Where First Customers Actually Come From

The 2011 paper behind the one in the first slot tracked 5,181 customers the same German bank acquired through its referral programme in 2006 and a random sample of 4,633 it acquired by other means, following all of them for 33 months. Its headline findings are genuinely positive for referrals. After controlling for demographics and month of acquisition, referred customers were about 18 percent less likely to defect at any given moment. Their value to the bank over a six-year horizon was about 40 euros higher on a base of 253, which is 16 percent, rising to about 25 percent once you count the roughly 20 euros less it cost to acquire them. Against a 25 euro reward paid to the referrer, the authors put the return at about 60 percent over six years.

And here is the same finding from the other end. The probability of still being an active customer after 33 months was 82.0 percent for referred customers and 79.2 percent for nonreferred ones. Two point eight percentage points — that subtraction is mine. An 18 percent reduction in a hazard that is already small produces a difference you would struggle to see in a business with eighty customers, and would certainly never notice without writing it down. Both numbers are true and describe the same data. If you only ever see the first one quoted, now you know why.

Two more things the paper found that the summaries drop. The margin advantage of referred customers eroded over time while the loyalty advantage did not, so the two halves of the benefit behave differently. And the value differential showed up among high-margin customers and not among low-margin ones, and the programme was less beneficial when used to acquire older customers. The authors' conclusion is explicitly that referral programmes should be selective rather than universal.

Now the part I owe you. This is the best-identified evidence I could find on how customers arrive, and it is about a bank's formal referral scheme, not about how a new business gets its first customers. Those are different questions. I looked for administrative data or a survey with a published sampling frame on first-customer acquisition for small firms and did not find one. What is abundant is statistics of the form '85 percent of small businesses get customers through word of mouth', which circulate without a method, a sample size, or a definition of what counts as word of mouth, and which are published almost exclusively by companies that sell marketing services. I am not going to repeat one here as though it were a finding.

I want to be careful not to turn an absence of evidence into a tidy conclusion. It is entirely possible that referrals dominate small-business customer acquisition. It is also possible that owners report referrals because a referral is the only acquisition event that arrives with a name attached, while the advert that made the referrer's friend remember you does not. Nobody has measured it in a way I can check, and the people best placed to fund that research are the people selling the alternative.

Which leaves one unglamorous instruction that costs nothing. Ask every new customer who sent them, record the answer in the same place every time, and after a year you will have something nobody else has: real data about your own business, with a denominator. If a name keeps appearing, the 2018 paper says that relationship is load-bearing, and the thing to do is not to automate a referral programme but to keep that person.