Your best lead is somebody who already paid you once

A heating engineer came to us wanting to spend a couple of thousand a month on ads. He had done the sums, he had a landing page in mind, he was ready to go.

Somewhere in the first conversation I asked how many customers he had done work for, ever. He thought about it and said probably fourteen hundred. And when had he last contacted any of them. Never, was the honest answer, other than the invoice. A few of them had rung him again over the years, on their own initiative, when something else broke.

So there is a list of fourteen hundred people who have let this man into their homes, seen his work, paid him money, and formed an opinion. Some of them are on boilers he serviced six years ago and has not seen since. Some of them have moved and got a new house with new problems. All of them are, right now, a warmer prospect than any stranger who clicks an ad, and reaching them costs the price of writing something.

We did that first. Not because ads are bad, but because it is very hard to justify buying attention while you are sitting on attention you already earned and never used.

The list you forgot you had

Almost everybody has one and almost nobody thinks of it as a list, because it is not in one place. It is in the invoicing software. It is in an old spreadsheet. It is in the phone's contacts, half of them saved as Dave Boiler or Mrs P Kitchen. It is in a WhatsApp history. It is in a shoebox of job sheets, in a couple of cases I have seen.

The first job is just assembling it, and being realistic about what state it is in. Duplicates, dead numbers, people who moved, one entry that is just a mobile number and the word urgent. That is normal. You do not need it to be clean, you need it to exist.

The second job is the interesting one: working out what you actually know about each person. Not demographics, nothing clever. What did you do for them, when, and what does that imply about now. A boiler installed five years ago is a service conversation. A bathroom done last spring is not, but the same customer might be thinking about the kitchen. A landlord you did one flat for probably has other flats. That is the whole segmentation strategy and it is not sophisticated, it is just paying attention to your own records.

Why a blast wrecks it

The instinct is to write one message and send it to fourteen hundred people. Please do not, and if we are doing this together we will refuse to.

A message to everybody has to be about nothing in particular, so it ends up being an offer. Ten per cent off, this month only. Which does three things, all bad. It reaches the people who did not need anything and teaches them that you are now a business that sends offers, so the next one gets deleted unread. It discounts work you would have got at full price from the people who did need something. And it burns the one asset that makes this whole approach work, which is that a message from you is currently a rare and personal thing.

Fourteen hundred is not a big number in advertising terms. It is a huge number of individual relationships. Treating it as an audience is the mistake.

So what does a good one look like

Specific to what you did, honest about why you are writing, and asking a real question. Something in this shape:

Hi [NAME], it is [NAME] from [BUSINESS]. I fitted your boiler back in [MONTH, YEAR] and I am going through old jobs, checking in on anything I have not seen for a while. It is coming up for its fifth winter, so it is worth a look at this point. Nothing urgent, and no obligation. Do you want me to put you in for a service in the next few weeks, or shall I leave you be?

Read that again and notice how much of it is doing nothing salesy. It says who it is and proves it with a fact only the real person would know. It gives a genuine reason for the timing. It contains no offer, no discount, and no urgency that was invented for the message. And it explicitly offers the option of leave you be, which is the part that makes it acceptable to receive.

That last line is not politeness for its own sake. It is the difference between a message from a tradesperson you remember and a marketing message, and everybody on the receiving end can tell which is which within about four words.

The rules we hold ourselves to

This is the system with the most obvious potential to be obnoxious, so the constraints matter more here than anywhere else. These are not legal minimums, they are what we will and will not build.

• Written individually, from your name, referencing the actual job. If we cannot say something specific to that person, they do not get a message.
• No invented urgency and no manufactured discount. If there is a real reason the timing matters, say it. If there is not, the reason is that you are checking in, and that is enough.
• One message. If they do not reply, that is an answer, and the list does not get worked again for a long time. A year, typically, and only if there is a genuine new reason.
• Stop means stop, immediately and permanently, and it is honoured across everything rather than just that campaign.
• Nobody who complained about their job gets a cheerful check in. That should be obvious and it needs a rule anyway, because the export does not know.
• It goes out in batches slowly, because if somebody replies you have to be able to actually answer them. Fourteen hundred messages on a Monday morning is not a campaign, it is a self inflicted crisis.

What to expect, and what not to

I am not going to give you a response rate, because it depends entirely on your trade, how long ago the work was, and whether the reason for contact is real. Somebody with a five year old boiler has an actual reason to reply. Somebody whose loft you insulated does not need it doing again and never will.

What is reliably true is the ordering. A message to a past customer outperforms an ad to a stranger by a distance, on both response and on how the conversation goes, because they are not deciding whether to trust you. They already did that, in their kitchen, and it worked out.

The thing that surprises people is the second and third order stuff. A decent number of replies are not yes or no, they are somebody remembering they meant to ask you about something else entirely. And a few are just nice, people saying thanks for checking, which sounds like it is worth nothing and is in fact exactly the thing that gets you recommended at a barbecue in June.

When not to bother

• Genuinely one time work. If what you do is done once per property and lasts thirty years, there is nothing to reactivate. Ask for referrals instead, which is a different conversation with different rules.
• A list under about a hundred names. At that size, put the phone in your hand and ring them. It will work better, you will enjoy it more than you expect, and there is no system to build.
• If the record of what you did for whom is genuinely gone. The specificity is the whole mechanism, and without it you are writing a blast, which we have already established is a bad idea.
• If you are already at capacity. Waking up demand you cannot service is a way of teaching your best customers that you are unavailable.

If you do run this and it works, the immediate problem becomes handling the replies properly, which is the first response piece. It is a genuinely stupid way to lose the work: you finally write to fourteen hundred people who like you, and then the answers land in an inbox nobody is watching.