The message in question is the least glamorous thing a business can send: a second contact about a quote that got no answer. No new offer, no discount, four sentences. It is also, on the only large dataset anyone has published, the message that seven businesses in ten never send. The interesting part is not whether it works, but that the arithmetic is so lopsided that whether it works barely matters.
Yesware, which sells email tracking to sales teams, analysed more than 500,000 messages sent by its users during the first quarter of 2014. Two findings are worth carrying. Of the threads that got no reply, 70 per cent stopped after the first attempt and only 19 per cent went as far as a second message. Where a second was sent, roughly 21 per cent of those threads produced a reply, and about a quarter answered eventually across the whole sequence.
Handle it carefully, because it is vendor telemetry rather than research. The population is people who had already bought a sales-email tool, the period is twelve years old, and a reply is not a sale. What it supports is narrow and still worth having: a second message to a silent thread is not a rounding error, and almost nobody sends one.
Now the claim you will actually meet. "Eighty per cent of sales require five follow-ups" appears in nearly every sales article written since 2010, and traces back to Marketing Donut, a British small-business publisher, whose own wording is that "different studies carried out at different times, in different places, by different market research companies over a number of years" reveal it. No study is named — no sample, no method, no date. We hit the same dead end tracing speed-to-lead statistics and quoting multipliers. In this field the most repeated numbers are reliably the ones with least behind them.
Since the true effect size cannot be recovered from published sources, stop asking how well follow-up works and ask how well it would have to work. That has a clean answer, and it needs three numbers you already hold: the minutes a follow-up takes, what an hour of that person costs you, and the gross contribution of an average won job. Multiply the first two, divide by the third, and you have the recovery rate at which the habit pays for itself.
Below is that calculation for three shapes of business in this part of the world, assumptions stated so you can substitute your own.
| Business shape | Follow-up cost per quote | Contribution per won job | Break-even recovery rate |
|---|---|---|---|
| Trade firm, Nouméa — 15 quotes a month, two follow-ups totalling 8 min, owner time at 3,000 XPF an hour | 400 XPF | 95,000 XPF — a 380,000 XPF job at 25 % margin | 1 quote in 238 · 0.42 % |
| Nautical operator — 60 enquiries a month, 4 min per enquiry, staff time at 2,000 XPF an hour | 133 XPF | 27,000 XPF — a 45,000 XPF booking at 60 % margin | 1 in 203 · 0.49 % |
| Professional services, Auckland — 8 proposals a month, 15 min each, chargeable time at NZ$90 an hour | NZ$22.50 | NZ$5,225 — a NZ$9,500 engagement at 55 % margin | 1 in 232 · 0.43 % |
Three trades, two currencies, job values two orders of magnitude apart — and the break-even lands between one quote in two hundred and one in two hundred and forty every time. That is not a coincidence, it is the point of the exercise: cost and value both scale with the size of the business, so the ratio survives the change of scale almost intact. For the trade firm the entire year of the habit costs twenty-four hours — one recovered job pays for all of it, and roughly nine more besides.
Which settles the argument unusually. You need not believe the 21 per cent, or any published figure at all: following up fails to pay only if it recovers nothing whatsoever, only if every client who went silent was always going to stay silent. Nobody who has kept a job book believes that.
If the maths were the obstacle this would be a maths problem. It is not. The three real reasons are structural.
None of that is repaired by resolving to try harder, which is why the resolution collapses in the second week. It is repaired by moving the trigger out of a person and into a system far smaller than the word suggests.
Two failure modes, and the second is the one that matters here. The first is the empty message — "just checking in", "bumping this up your inbox". It adds nothing and teaches the client that your messages can be skipped without consequence. A follow-up has to carry one thing the first did not: a lead time that has changed, a supplier price that has held, a comparable job finished nearby.
The second is specific to a market this size. In Nouméa or Port Vila you will meet this person again, at a trade counter or a school gate, and a sequence that reads as bulk mail does damage no recovered job repays. That is the honest argument for a stopping rule: two follow-ups, then nothing until they come back to you. Not because a third message never works, but because the cost of becoming the firm that will not stop is paid somewhere the numbers cannot show you.
Twice is the defensible answer for most small firms: once around day four, once around day twelve, then stop until the client comes back. There is no dataset supporting a specific number, so the interval is a judgement rather than a finding. What the arithmetic does support is that two well-spaced messages cost so little against the contribution of one won job that they pay for themselves at a recovery rate below half a per cent. Beyond two, the gain becomes speculative while the relational cost in a small market becomes real.
Very little, and it is worth being clear about that. The most quoted claim — that 80 per cent of sales need five follow-ups — traces to Marketing Donut, which attributes it to unnamed studies by unnamed research companies over unstated years. Nothing there can be checked. The one substantial published dataset is Yesware’s analysis of over 500,000 emails in the first quarter of 2014, which found that 70 per cent of unanswered threads stopped after a single attempt and that second messages drew replies about 21 per cent of the time. That is vendor telemetry from a self-selected population, twelve years old, and it measures replies rather than won work.
The reminder should be; the message should not be sent unread. The part worth handing to a machine is the part people fail at — noticing that a quote has gone quiet, producing the list, drafting the text and putting it in front of someone at the right moment. The part to keep is the last thirty seconds, where a person adds the detail that makes the message specific to that client and that job. Sequences that send themselves on warm leads are how a small firm acquires a reputation for bulk mail, which costs more than the jobs recovered.
Produce a list of every quote sent in the last ninety days with its date and status. Most owners discover this takes far longer than expected, and that discovery is the finding: the follow-up is not being skipped through lack of will but because the list does not exist and cannot be reconstructed on a Friday afternoon. Fix the record first, then the trigger, then the wording. Doing it in the other order produces well-written messages that never get sent.
Kanaky Tech is an AI automation agency working across New Zealand and the Pacific. Start with a free AI opportunity audit: we map how your business actually runs, rank what is worth automating, and give you a clear scope before anything is built — no obligation.