The Customers You Already Have

Why aftercare might be the most profitable place you're not using AI

Last month, this column was about measuring the return on AI. Pick one task, time it, put a number on it. This month, a place where the return is hiding in plain sight. Not in new leads. In the customers you've already got.

The glazing industry has always been geared towards the next job. Almost every marketing conversation I have is about new enquiries: where they come from, what they cost, how to convert more of them. Meanwhile, nearly every installer is sitting on something more valuable and barely touched. A list of past customers. People who chose you, paid you, and would recognise your name if you got in touch.

Most of those customers hear from a business exactly twice. Once around the installation, and once if something goes wrong. Between those two moments, which might be ten years apart, silence.

Part of the reason is practical. Aftercare information is usually scattered. What was fitted, when, what the warranty covers, whether anything has been serviced since. It lives across job files, old emails and someone's memory. Pulling it into shape always felt like a project for a quiet January that never arrives.

That's exactly the kind of job AI has made cheap. Take your completed jobs from the last few years, whether that's an export from your CRM or a messy folder of job records, and ask ChatGPT, Claude or Copilot to build a structured aftercare register from it. What was installed, when, warranty end dates, and a sensible contact schedule for each customer. What used to be a winter admin project becomes an afternoon.

Once that register exists, options open up. A one-year check-in asking how everything is performing. A note when a warranty passes its halfway point. A reminder that hinges and gaskets last longer with an occasional service. None of it salesy. All of it drafted by AI in minutes, read by a human, with one personal line added before it goes.

Picture how that plays out. A customer had windows fitted in 2023. Two years on, they get a short message: hope the windows are still performing well, here's a reminder of what your warranty covers, and if any handles or hinges feel stiff, a service visit is straightforward to arrange. Nine customers file it away with a good impression. The tenth replies, because the back door has been catching for months and they didn't want to make a fuss. That's a service visit, possibly a chargeable one, a customer who feels looked after, and a five-star review that mentions how good you were years after the install. All from a message AI drafted in seconds.

Here's why it's worth the effort. A past customer costs nothing to reach and already trusts you. And glazing buying rarely ends at one contract. Doors follow windows. The second phase follows the first. The conservatory conversation happens three years later. Add the neighbours and family who ask "who did yours?" and the quiet economics of aftercare start to beat most paid marketing.

Two words of honesty, because they matter. First, data protection. Contacting customers about their own installation and warranty is service. Blanket marketing to an old database is a different thing, and consent rules apply. If in doubt, check your lawful basis before you press send. Second, don't fully automate this. AI can build the register and draft the messages, but a person should read every one before it goes out. An aftercare message with the wrong product in it does more harm than no message at all.

Starting is simpler than it sounds. You don't need the full register on day one. Take your last fifty completed jobs, get them into a spreadsheet, and ask AI to draft a check-in message for each based on what was fitted and when. Read them, personalise them, send a handful a week. If the responses justify it, work backwards through the rest of the database. If they don't, you've spent an afternoon finding that out, which is precisely the kind of small, measured experiment this column keeps coming back to.

There's a measurement angle too, to connect back to last month. Aftercare is one of the easiest AI uses to measure. Count the replies. Count the service visits booked. Count the quotes that start with a check-in message. Within three months you'll know exactly what it's worth to you.

So try this. Ask yourself what your last hundred customers have heard from you since handover. If the honest answer is nothing, you've found this month's project. And for once, it doesn't start with finding a single new lead.