The Customer Who Quietly Stopped Calling
Why this matters
A customer who complains is a customer you still have. The dangerous ones leave without a word, keep paying the last invoice on time, and never appear on any report you look at, because your reports are built around work that happened rather than work that stopped happening. By the time an owner notices, the replacement shop has been on the property for months and has a relationship of its own. This is a diagnosis, and like any diagnosis it goes wrong in predictable ways. What follows is one account, followed from the first signal through three wrong reads to the actual cause.
The signal, and why it took five months to surface
The account was a nine-building residential portfolio run by a single property manager. Over the twelve months ending in March it produced 11 tickets, a steady pace working out to a median gap of about 32 days between calls. Then one ticket in April, and nothing.
Nobody noticed in May, June, or July. There was no complaint, no unpaid invoice, no argument. The account simply stopped appearing on the schedule, and a schedule only shows what is on it.
It surfaced in September when the office ran a gap report: every active customer sorted by days since last invoice, longest first. That account sat near the top at 147 days, which is about 4.6 times its own 32-day median gap. That ratio, days-since-last-ticket divided by that account's own historical median gap, is the signal to look for. An absolute number of days is useless across a mixed list, because a 147-day gap is alarming on a monthly account and completely normal on a once-a-year one.
The shop's own standing rule was to investigate at 3 times median gap, which for this account would have been 96 days, in July. The report was quarterly and it should have been monthly. That is the first lesson and it is about detection, not diagnosis: the cost of finding this late is that the replacement shop gets two extra months of unopposed relationship-building.
First read: it is seasonal
The immediate assumption was that this portfolio always went quiet over the summer. It is the comfortable read because it requires no action.
Ruled out in ten minutes. The same six-month window in each of the prior two years had produced 4 tickets and 5 tickets on this account. A quiet summer for this portfolio historically meant a reduced pace, not a stopped one. Zero was not seasonal.
The general form of this check: before you accept seasonality as an explanation, pull the same calendar window from at least two prior years on that specific account. A shop-wide seasonal dip is real and it does not prove that any individual account's silence is seasonal.
Second read: we lost on price
The second guess was a rate problem, and this one was half right in a way that nearly sent the diagnosis into a ditch.
The evidence against it: the last six invoices before the stop carried no disputes, no line-item questions, and all six were paid within terms. The shop had not changed its standard hourly or flat-rate pricing in that period.
The evidence that kept it alive: the shop had raised its after-hours rate in February. So the question became how exposed this account was to that rate, and the history answered it. 2 of the 11 tickets in the prior year were after-hours, so roughly 18% of the account's tickets touched the rate that changed. A change affecting under a fifth of an account's work does not usually produce a complete stop on the other four fifths.
That is the reasoning that mattered. The wrong version of this step is to find any plausible cause and stop. Price was plausible, and testing its actual exposure showed it could not carry the weight of the whole outcome. It stayed on the list as a possible contributor and came off the list as the explanation.
Third read: our contact left
The third guess was the most common real cause of a silent portfolio account, so it deserved a real check rather than a shrug. Managers move, and a new manager arrives with their own vendor list.
One phone call ruled it out. The same manager was still in the role and answered directly.
This is worth doing before anything else that costs effort, because if the contact had changed, every conclusion drawn from the old relationship would have been wrong, and the correct next move would have been introduction rather than repair.
What the call actually turned up
The manager was not hostile, not avoiding, and not aware there was anything to discuss. Asked directly whether someone else had been doing the work, the answer was yes, another shop had handled the last three jobs.
Asked why, the first answer was the one you almost always get: "nothing really, it just kind of happened." That answer is not a lie and it is not the end of the conversation. It means the departure was made of small things, none of which individually justified a phone call to complain.
The question that opened it up: "Can you tell me the last time we did something that annoyed you, even a small thing you did not think was worth mentioning?"
Two things came out.
One. The portfolio required a short written condition report emailed to a corporate address after every visit. Those had been going out reliably. They stopped in February.
Two. In March, the shop missed a scheduled window on one building by most of a day and nobody called ahead. The tech showed up late, did the work, and left. The manager fielded a resident complaint about it and had nothing to say.
Neither of those is a firing offense on its own. Together, and unacknowledged, they were enough that when the next job came up the manager tried someone else, that shop showed up on time and sent the report, and the decision quietly made itself.
The cause behind the cause
The reports stopped in February because the office person who wrote them left the shop in January. Nobody else knew the obligation existed. It had never been on a ticket, a checklist, or a job template. It lived in one person's habit.
That is the actual failure, and it is more useful than "we lost a customer over paperwork." An account-specific obligation that exists only in a person's memory has a defined expiry date, and the date is the day that person leaves. Every shop has several of these right now: a gate code, a report, a preferred arrival window, a resident who must be notified first. If it is not attached to the customer record or the ticket, it is not a process.
The missed window in March mattered for a different reason. The missed window itself was survivable. The absence of a call about it was not. The manager's actual exposure was a resident complaint they could not answer, and a two-minute heads-up would have removed it entirely.
Underneath both: the shop had "no complaints on file" for this account and had read that as no problem. Absence of complaint is not evidence of satisfaction. It is most often evidence that the customer decided you were not worth a difficult conversation.
The repair
The shop did four things, in this order.
- Fixed the process before making contact. The condition report became a required field on the ticket for every building in that portfolio, so completion was gated on it rather than remembered. Making the promise before the mechanism exists is how a recovery fails twice.
- Owned the two specific things by name, in one call, without a discount attached. "Our reports stopped in February when we lost the person who sent them. That was on us and it is now built into the ticket so it cannot happen again. And in March we were most of a day late on one building and nobody called you. You had to answer for that and you should not have had to."
- Asked for one building, not the portfolio. A request to take nine buildings back forces the manager to declare the other shop a failure. One building is a low-cost trial for them and a real test for you.
- Set a check-in at 90 days with a date on it, so the follow-up was not left to the same memory that lost the reports.
Where it landed
At nine months, 3 of the 9 buildings were back. Those three produced 3 tickets over the nine months, an annualized pace of about 4 tickets a year across 3 buildings, or roughly 1.3 tickets per building per year. The account's historical pace had been 11 tickets a year across 9 buildings, about 1.2 tickets per building per year. So the recovered buildings were running slightly above the old per-building rate, which is the number that told the shop the relationship was genuinely working rather than being tolerated.
Full portfolio recovery did not happen and probably will not. That is a normal outcome. A partial win-back on a silent departure is a good result, and treating anything short of the whole account as failure is how shops talk themselves out of trying.
What would have changed the conclusion
If the manager had left. Everything above would be misdirected effort. The correct read on a portfolio that goes silent right after a contact change is that you have no relationship to repair, you have one to start, and the move is an introduction plus a case for being kept, not an apology for something the new person never experienced.
If procurement had rebid the portfolio. Some portfolios move to a formal vendor list on a fixed cycle. If the silence lines up with a rebid you were not invited to, the cause is upstream of anything you did on a job, and the repair is getting on the next bid list rather than fixing reports. You find this out by asking whether the vendor list changed, which is a different question from asking whether they were unhappy.
If the last invoice had been disputed or unpaid. Then price and billing move to the front of the queue instead of being ruled out, and the recovery conversation has to resolve the money before it can address anything else.
If the gap had been 2 times median rather than 4.6. At 2 times, on an account with a 32-day median gap, you are looking at roughly 64 days, which for a nine-building portfolio is within normal noise. Investigating every account at 2 times median generates enough false alarms that the office stops running the report. The 3-times threshold exists to keep the report credible.
References
- Trade-standard practice for account review cadence and vendor-of-record management in property portfolios
- U.S. Small Business Administration (SBA), customer retention guidance for small service firms
- See related: The Lost Customer Win-Back Checklist, When a Customer Cannot Be Won Back and That's Okay, Customer Churn Reasons + Recovery Strategy