The reason is usually that the experience was good and the outcome held. That is necessary and it is not sufficient, because a satisfied customer who has not heard from you in fourteen months will search when the need returns rather than recalling your name. Satisfaction produces willingness to return. It does not produce the act of returning.
The reminder is the part that is missing in most small businesses, and it does not need to be sophisticated. A periodic message with something genuinely useful in it, sent to people who have bought from you, is the highest return retention mechanism available at small scale. It beats loyalty cards, points, and every formal programme, because it costs almost nothing and it works on the actual mechanism, which is being remembered at the moment the need arises.
Time it against the natural cycle of what you sell rather than against a calendar. A business whose customers buy annually should be in contact well before the anniversary. One selling something bought when a problem occurs should be in contact often enough that you are the name in mind when it does. Getting this wrong in either direction produces the same result: the message arrives when there is nothing to act on.
Make the contact useful rather than promotional. Something they can apply, a change in your industry that affects them, or an answer to a question you keep receiving. A message that only announces an offer trains people to ignore you, and the open rate falls with each one until the list is inert. A message that helps gets read, and the business is remembered as a source of help rather than as a source of offers.
Then make returning easy. The customer who wants to come back should not have to explain their history, find an old invoice, or work out how to get in touch. Knowing what you did for somebody, when, and what they bought is what allows you to pick up the conversation where it stopped, and that record is why a customer system matters more for retention than for acquisition.
Ask after delivery, at the point where the result is visible, whether there is anything else. This single question produces more repeat work than any campaign, and almost nobody asks it because it feels like selling at a moment that should be about completion. It is not. A customer who is pleased is a customer who would rather not go through the search process again.
Watch which customers actually return, because the pattern is usually more specific than expected. Frequently one type of work produces repeat business reliably and another almost never does, regardless of how well either was delivered. That distinction is worth knowing before deciding where to spend acquisition effort, since a customer type that never returns is worth considerably less than the first invoice suggests.
And treat a customer who has gone quiet as a question rather than a loss. A short direct message asking whether their situation changed produces answers, and the reason is frequently something you can address. Businesses assume dissatisfaction, and the more common cause is that the need moved, the contact changed roles, or nobody stayed in touch.
Segment the list even roughly, because a message relevant to everybody is relevant to nobody in particular. People who bought one service have different follow up needs from people who bought another, and sending both groups the same general update produces lower engagement than two shorter messages would. This does not require software at your scale, only two lists and the discipline to keep them separate.