How to Re-Earn a Customer Who Left for a Competitor
Why this matters
A dormant customer has nobody. A customer who left for a competitor has someone, and that someone is showing up, doing the work, and building a relationship of their own every month you wait. That makes this a different job from reactivation: you are not filling a vacancy, you are asking a person to switch away from something that is currently working well enough. Shops get this wrong by treating it as a discount problem, and a discount-led return sets the terms of every job after it. The whole approach turns on one reframe: you are not asking for the account back, you are asking for a test.
Step 1: Confirm they actually left, and to whom
Before spending any effort, separate three things that look identical from your side of the counter. A customer who went quiet may have left, may be dormant with no replacement, or may simply have stopped needing what you do because a property sold, a system was replaced, or a business shrank.
You find out by asking, and the question that works is flat and unembarrassed: "Have you got someone else handling this now?" Nobody is offended by it and almost everybody answers honestly. If the answer is no, you are running a reactivation and the approach in this article is over-engineered for the situation. If it is yes, keep going.
Ask who, if they will say. Knowing whether you lost to a large operation, a one-truck shop, or the tenant's brother-in-law changes what you are competing against and what you should not bother trying to match.
Step 2: Get past the polite reason
The first reason you are given is almost always price, and it is almost always a rationalization. Price is the socially safe answer because it blames nobody. The real reason is usually a behavior, and there are three questions that surface it.
"When you switched, was there a specific visit you remember, or did it build up over time?" This separates an event from a drift, and the two need completely different repairs. An event has a date and can be addressed directly. A drift means a pattern, and a pattern means a process, not an apology.
"What does the shop you use now do that we did not?" This is the highest-yield question in the whole process, and it works because people will describe a competitor's behavior in far more concrete detail than they will describe your failure. Criticizing you costs them something socially. Praising someone else costs them nothing, and it hands you the same information.
"If you had called me about it at the time, what would you have wanted me to do?" This gets you the fix in their words rather than your interpretation of the complaint, and it frequently reveals that what they wanted was much smaller than what you would have offered.
Step 3: Fix the thing before you make contact
This ordering is not optional and it is where most attempts fail. If you call with an apology and a promise, you have spent the one contact you get on an intention. If the promise then does not hold on the test job, you have confirmed their decision and there is no third attempt.
Build the fix first, in a form somebody else could execute if you were out sick. A policy change, a named person, a checklist item, a field on the record. Then make the call, and describe the fix in the past tense.
Step 4: Make one contact, and make it a statement
One contact. Not a sequence, not a drip campaign. A customer who left and is being courted by a series of messages is being pursued, and pursuit is the posture that guarantees a no.
The call is a statement rather than a question, because a question invites them to relitigate a decision they have already made and defended to themselves. The shape:
- Acknowledge the situation plainly and without wounded tone.
- Name what you got wrong, specifically, with evidence from your own records.
- Describe what you changed, in the past tense, as something already done.
- Explicitly do not ask for the account.
- Ask for one job.
Step 5: Do not lead with price, and do not discount the re-entry
The temptation is enormous and the logic against it is simple. If you buy the return with a discount, the discount becomes the reference price. Every subsequent invoice is measured against it, and the first one at your real rate is a price increase on a customer you just recovered. You will have re-earned them once and then given them a fresh reason to leave.
If price genuinely was the reason - and it sometimes is, particularly against a shop with a structurally lower cost base - the honest move is to say what your price buys and let them decide. "We are not going to be the cheapest and I am not going to pretend otherwise. What you get for it is the same tech, a written report, and a real answer on the phone at 8pm." A customer who leaves on price after that conversation was never yours to keep, and you have saved yourself a year of unprofitable work finding out.
Step 6: Ask for one job, scoped small
Asking for the account forces them to declare their current shop a failure, in front of you, on a phone call. Very few people will do that. Asking for one job costs them almost nothing and does not require them to fire anyone.
Take a small, low-stakes job if that is what is offered. A test job you execute perfectly is worth more than a large job you win by pressure and then have to deliver under a suspicion you have not yet cleared.
Step 7: Over-execute the test, and name what you did
The test job is not the moment for your standard service. It is the moment for your best, and specifically for the visible version of the thing you said you fixed. If you promised continuity of personnel, the tech who shows up is the named one, and they mention it. If you promised documentation, the report is in their hands before the truck leaves.
Name it once, without fishing for credit. "You will get me on this account from here out, and everything I find goes in the file so the next visit does not start from zero." One sentence. Saying it twice turns a delivered promise into a performance.
Step 8: The three-job rule
The relationship is not re-earned at job one. On job one, they are testing whether the fix was real. On job two, they are testing whether it survives you being busy, which is the test the first job cannot run. Job three is the first one where they are not testing anything, and that is where a relationship exists again.
Do not ask for more scope, a bigger portion of their work, or a maintenance agreement before job three. The ask that would have been natural after a long relationship reads as opportunism after one successful test.
Step 9: If the answer is no
One sentence, then out. "Understood. If anything changes, you know where I am." Then put a calendar note for twelve months out with one line about why they left, so whoever picks it up has context instead of a cold record. No further contact in between, and no adding them to a campaign list. The twelve-month gap is what makes the eventual second contact a fresh conversation rather than the continuation of a pursuit.
Worked example: one commercial account, followed through
A small multi-tenant building owner had used a shop for 8 tickets over 3 years, a pace of about 2.7 tickets a year. Then nothing for 14 months.
Step 1 confirmed it: yes, someone else had it now.
Step 2 produced "your prices went up" on the first pass. The second question, about what the new shop did differently, produced the real answer in one sentence: "Their guy knows the building." Pulling the history confirmed it. Across the 8 tickets the shop had sent 6 different techs, and on the last 4 tickets it had sent 4 different people. The owner had re-explained the layout, the tenant access rules, and which panel served which unit on essentially every visit, and on the last one a tech had disconnected something a previous tech had labeled and left.
Price was real but secondary. The shop had raised rates in that period, and against a background of having to teach a stranger the building every time, a rate increase is what converts irritation into a decision.
Step 3 came before any contact. The shop assigned a named lead tech to the account permanently, with one named backup, and built a two-page site file: unit-to-panel mapping, tenant access contacts, known quirks, prior work with dates, and the verified locations of the main shutoffs and the disconnect serving each unit. That last item is worth building on every commercial site regardless of any of this, because on an after-hours call the difference between knowing where the shutoff is and searching for it is the difference between minutes and real damage.
Step 4 was one call, roughly: "I know you have had someone else on the building for about a year. I looked at our history. We sent six different techs across eight visits, and the last four visits were four different people. You had to teach your building to a stranger every time, and one of them undid another one's work. That was ours. There is one tech on this account now with one named backup, and there is a site file so it does not reset. I am not asking for the building back. If there is one job this year you would be comfortable trying us on, I would like that one."
Step 6 got a small job: a routine service call on a single unit.
Steps 7 and 8 ran over the following 18 months, in which the account produced 3 tickets, a pace of about 2 a year against the prior 2.7 a year, roughly 74% of the old pace. The competitor kept a share of the work and still has it.
That partial result is the realistic one, and calling it a failure is how shops abandon a method that works. Measured against the alternative - the account at zero - three tickets in 18 months on about 6 hours of total owner and office time (two hours building the site file, one hour across discovery and the call, three hours of owner attention spread over the three visits) is a good return. The shop's own estimate was that acquiring a comparable commercial account through advertising took several times that effort with a much lower hit rate, which is a judgment about their market rather than a general rule, but it is the comparison worth making in yours.
The site file also turned out to be the most transferable part. Once built for one account, the same two-page format went onto every commercial site the shop served, which fixed the underlying problem for customers who had not left yet.
What changes the answer
If they left over a safety or workmanship failure, this process is too light. A customer who left because your work created a hazard needs the failure investigated, documented, and disclosed to them in writing, and any re-entry conversation comes after that, not instead of it.
If you fired them, none of this applies. Reactivating a customer you deliberately ended is a separate decision about whether the conditions that made them unworkable have changed, and it is not a competitive-displacement problem.
If they left as part of a formal vendor rebid, the individual relationship is not the lever. Getting on the next bid list is, and the timeline is the bid cycle rather than your test job.
If your shop is at capacity, do not run this. A re-entry test job you cannot schedule for three weeks fails on the exact dimension you are being tested on, and you will have used your one contact to prove their decision was correct.
How to verify you got this right
At job three, ask one question: "Anything I should be doing differently on this building?" A real answer means the relationship is live. "No, all good" from a customer who left you once means they are still being polite, and you have work left.
The other check is on your own end. Six months after the fix in Step 3, verify the fix is still running - that the named tech is still the one going, that the site file is still being updated. A repair that decays back to the original behavior does not just lose the customer again, it loses them permanently, because the second departure carries evidence that your promises do not hold.
References
- Trade-standard practice for site documentation and account continuity on commercial service work
- U.S. Small Business Administration (SBA), customer retention and competitive positioning guidance for small firms
- See related: The Lost Customer Win-Back Checklist, When a Customer Cannot Be Won Back and That's Okay, Customer Relationship Recovery, Deciding Whether to Take Back a Customer You Let Go