Your Dormant Customer List Is Worth $12K a Year

There are a few hundred people in your scheduling software who paid you once and never called back. Nobody fired you — the water heater stopped leaking, the problem went away, and your name faded out of their phone.

Here is what that list is worth. 480 past customers with no booking in 12 months, a $640 average ticket, and a deliberately stingy 4% response rate is $12,160 in work you already know how to do. No ad spend, no new lead source, no discount required.

The reason to do it in September is that the alternative gets expensive in January. In about ninety days the weather is going to hand a few hundred households in your service area an emergency, and they will call whoever they can name that morning. Searches for “frozen pipe repair” run 609% higher in January than in August, according to WebFX’s analysis of a year of U.S. home-service search volume in its seasonal search trends report.

In January you buy those calls at auction. Searchlight Digital tracked $6.72M in Google Local Services Ads spend across 888 contractors and 126,650 leads for its 2026 cost-per-lead benchmark: plumbing leads ran $57 each in February 2026, and only about 19% of LSA leads become paying customers. That is roughly $300 to buy one customer. Right now you can reach a customer who already paid you once for the price of a text message.

Ten minutes gets you the same figure for your own list.

Nobody left. The reason just ran out.

Owners assume a customer who never came back was unhappy about something, so they leave the bear alone. Almost none of them were unhappy. The water heater got flushed, the drain got cleared, the problem stopped being a problem, and the next reason to call you never arrived on a day when your name happened to be in front of them. Two years later they need a plumber, they Google one, and they do not remember that you are already in their phone.

The best data on this comes from dentistry, which tracks it more closely than the trades do. Levin Group’s 30-year ongoing study of top-producing practices, reported by Dental Products Report, found that top 10% practices lose about 7-9% of their patient base a year while most practices lose 12-15%. The difference is not clinical skill. It is whether somebody owns the follow-up.

That is dental data, not trade data, but the mechanism travels. A 12% annual leak on a list nobody works compounds into hundreds of people who used to be yours.

Run it on your own list

Those numbers up top came from a six-truck plumbing and drain shop: roughly 1,400 customer records going back to 2019, 480 of them with no booking in over 12 months, and a $640 average ticket on a water heater flush, a winterization, or a drain job.

Now the response rate, and I am going to be stingy with it on purpose. Relentless Digital’s database reactivation FAQ says typical campaigns produce a 3-8% reactivation rate, and its example runs a 3,000-person database at 5% for 150 jobs at a $2,500 ticket, which comes to $375,000. That is a vendor illustrating a vendor’s product on a ticket most plumbing shops do not have. Apply the same test you would use on any AI vendor’s ROI slide and take the bottom of their range with your real ticket.

Four percent of 480 is 19 jobs. Nineteen times $640 is $12,160.

Channel matters more than the copy does. Winback Engine, which runs reactivation campaigns in fitness, wellness, beauty, and franchise verticals, reports email at 2-5%, SMS at 5-10%, and phone calls at 25-40%. Those are not home-services numbers, but the ordering holds everywhere anyone has measured it: the more human the channel, the better it works. Their other finding should move your calendar. Results degrade sharply past 12 months of inactivity, and a list sitting at 18 months is largely stale. Your 480 names are not getting more valuable while you think about it.

Fall is the cheapest excuse you will ever have

A reactivation text with no reason attached is a promotion, and people delete promotions. A reactivation text tied to the weather is a heads-up, and people answer those.

The frozen-pipe spike is not an outlier. The same WebFX data shows “water heater repair” peaking in December and furnace repair climbing 137% in January, and Google Ads bids climb right along with them. That is the part that costs you money.

Whatever your trade, the service your customer will need in Q4 or Q1 is knowable in September — and September is when the message is still free.

What the AI is actually doing

Small job, done well. You are not buying a reactivation platform.

Export your dormant list with the two columns that matter: last service date and last service type. Paste thirty rows into ChatGPT or Claude and ask for a text under 300 characters per service type, naming the seasonal service, referencing roughly when you were last out, and offering two scheduling windows. You get back four or five variants in fifteen seconds. Fix the two that sound like a robot wrote them, then mail-merge from your software.

That is fifteen minutes of AI work replacing the afternoon of writing that is the real reason this never gets done. The judgment stays yours: who gets a text, who gets a phone call, and what you are willing to promise.

Here is the shape that works. “Hi Dana, this is Ray at Bridgeview Plumbing. We replaced your water heater back in March 2024. We’re booking winterizations before the first freeze and I have Thursday morning or Friday afternoon open. Want one? Reply STOP to opt out.”

Specific. Dated. Two options. No coupon.

Three ways to break this

Texting people who never agreed to be texted. A past customer’s phone number is not blanket permission for a marketing text. Under the TCPA (the federal law covering marketing texts and calls), promotional messages generally need express written consent, and the rules treat a service reminder differently from a promo blast. ActiveProspect’s guide to TCPA text rules is a readable summary. Include an opt-out on every message, honor it immediately, and ask your software vendor what consent they captured at booking. Email is the safer first move if you are unsure.

Sending all 480 on Monday morning. Nineteen jobs is great. Nineteen jobs arriving as sixty inbound replies while your one dispatcher is already fielding normal calls is how you end up with the same problem as the calls nobody answers after five, except you paid to create it. Send in batches of 100 across two weeks.

Leading with a discount. You do not have a price objection. You have a memory problem. Offer a date, not 15% off, and you keep your margin and your positioning.

What to do this week

Three steps, in order. Each one is a decision you can settle today.

1. Pull the list, get your own number, and settle consent. Jobber’s Client Re-Engagement Report already shows who has not booked in a set period. On Housecall Pro, go to the Marketing Center and build a campaign audience filtered on last job date. Anything else, sort by last invoice date. Count the names, multiply by 4%, multiply by your average ticket. While you are in there, ask your software vendor what consent they captured at booking — if the answer is vague, email goes first and text waits.
2. Name the service the season is about to sell for you. One message per service type, not a newsletter. Draft it with AI tonight, edit it yourself, save it as a template you reuse every October.
3. Decide who reads the replies, then send the first 100. Oldest lapse last, because the 12-to-18-month group still remembers you. Track two numbers only: replies and booked jobs.

Send the next hundred each Friday until the list is done, then put a recurring item on the first week of every September and March so nothing on it ever ages past six months, the same way auto-renew keeps a maintenance plan from quietly expiring. January is going to arrive on schedule either way. The only part you control is whether these 480 people already have your name in their phone when it does.