Introduction — Forty-One Thousand People at Nine in the Morning
A dead discount code, an imported list, and an open rate that rose for the wrong reason.
Wren managed email for five clients. Newsletters, promotions, a couple of welcome sequences, and a monthly report each. It was going well enough that a sixth was waiting.
The Black Friday campaign went out at nine in the morning to forty-one thousand people. The discount code in it was the previous year's, deactivated in September. By nine past nine the first replies had arrived. By half past, the client's phone was going and somebody in their support inbox was typing the same explanation for the eleventh time. There is no recall. The email had already been delivered, opened, forwarded and screenshotted, and the only available action was to send another one — which meant the day's volume doubled, to a list that had already been mailed twice that week.
That was the visible failure. Three others had been running quietly for months.
The first was a list. One client had imported around nine thousand addresses from an old CRM at the start of the arrangement — records collected over six years for various reasons, some of them consent to be contacted about something else entirely. Nobody had asked where the addresses came from, because they were already in the platform when Wren arrived and importing them felt like a technical step rather than a decision. The complaint rate on that client's sends had been climbing all autumn. In November the sending domain was throttled by one large mailbox provider, and because Wren had set every client up on the same shared sending infrastructure — sensibly, cheaply, at the start — the throttling degraded delivery for all five.
The second was an automation. A welcome sequence built in month two contained a line about a service the business had stopped offering in the spring. It had sent, unchanged, to every new subscriber for seven months. Nobody had audited it, because automations are the part of the system that is supposed to look after itself.
The third was a number. One client's open rate had risen from roughly 22% to roughly 38% across the year, and they were delighted. Wren had put it in every report because it was what the platform reported. It was also true that a large share of that client's audience used a mail app that fetches images automatically, which registers as an open whether or not a human ever looked at it. The rate had improved because the measurement had changed. Meanwhile the click rate — inflated in its own way by corporate security scanners following every link in every message — was being read as evidence that the content was working.
By December, Wren was doing the work of a demanding job in evenings, on fees set before any of this was understood, with no way to explain most of it. There was no pre-send checklist, so the code was never checked. There was no consent record, so nobody could say what the nine thousand addresses had agreed to. There was no automation audit, so a sentence stayed wrong for seven months. And there was no baseline with definitions attached, so a measurement artefact looked exactly like success.
Nobody behaved badly. The code had been valid when it was written. The import was routine. The welcome sequence had been correct on the day it was built. The open rate was real.
Every part of it was ordinary, and all of it was expensive.
What this book is about
This book is about the four things underneath that story, because every one of them is a property of the trade rather than a mistake anybody made.
The permission is the asset. A list is not a possession. It is a set of individual agreements, each one specific to what the person actually said yes to, each one revocable, and all of them decaying with time. You cannot buy it, you cannot infer it, and you cannot inherit it without asking where it came from. Everything in this trade — deliverability, revenue, the legal position, the ethics — rests on that one thing, which is why it is Chapter 5 rather than an appendix.
A send is simultaneous and irreversible. There is no draft state, no staged rollout by default, no undo. The mistake does not reach one person and then get caught; it reaches everybody at the same moment, and the only remedy is another send, which costs more of the same asset. This is the reason the pre-send checklist in this book is short, fixed and never skipped.
Deliverability is a reputation you can only lose. Mailbox providers decide whether your client's email reaches the inbox, using signals you cannot see, with no appeal and no explanation. That reputation is built over months and can be damaged by one send to a bad list — and if your clients share sending infrastructure, one client's mistake becomes everybody's problem.
The metrics are contaminated. Opens are inflated by privacy features that fetch images on the subscriber's behalf. Clicks are inflated by security systems that visit every link before a human sees it. The number every client asks about first is the least reliable one available, and the improvement they are celebrating may be a measurement artefact.
What this book will give you
A consent and provenance record, so "where did this list come from?" has an answer before it becomes a question. A fourteen-point pre-send checklist that takes minutes and has never once been the thing that was worth skipping. A seed list and a test-send routine that catch what a proofread does not. A claims and offers register, so a discount code has a start date, an end date and somebody's name against it. A quarterly automation audit, because the sequences running unattended are where a wrong sentence lives longest. A list hygiene routine and a sunsetting policy — because removing people from a list reliably raises revenue, which is the most counterintuitive thing in this trade. And a report that shows delivered rather than sent, puts every rate beside its denominator, and states plainly which numbers are contaminated and by what.
The economic unit is revenue per campaign — what a build, a check, a send and its follow-up actually return against what they cost.
The leading indicator is build hours per campaign, which should fall steeply once the template system exists and does not fall at all without one.
The master variable is the engaged share of the list — the proportion who have opened or clicked in a recent window. It drives deliverability, it drives revenue, and it is the number that makes sunsetting profitable rather than frightening.
And the risk indicator is the complaint rate, tracked cumulatively, because it is the figure that ends sending programmes and it gives very little warning.
Who this is for
Somebody who has inherited a list and never asked where it came from. Somebody who sends without a checklist because nothing has gone wrong yet. Somebody with automations running that they have not read in six months. And anybody who has reported an open rate going up without checking whether anything else changed.
There is one more person in this, and they never asked to be here: the subscriber. They gave an address to somebody, once, for a reason, and everything in this book is ultimately about whether that reason is still being honoured. Chapter 5, Chapter 9 and Chapter 15 exist for them.
©2026 James Henderson / https://localhandyman.work