Losing a Long-Standing Customer to a Competitor

Why this matters

A customer you have had for years stops calling, and you find out they are using somebody else. The shop's account of it forms within a day and it is almost always the same account: they went cheap. That story is comfortable, it requires nothing of anyone, and it is wrong more often than it is right.

The cost of getting it wrong is not the one account. It is that the actual cause is still running, doing the same thing to customers who have not left yet, and you have filed the whole event under a cause you cannot fix. A single loss is a free diagnostic on a system you cannot otherwise see from the inside, and the entire value of it is in one phone call most owners never make.

The call nobody makes

It does not get made because the anticipated version of it is humiliating: you calling a customer who left, to ask them to explain themselves, while they suspect you are about to sell.

That version is avoidable, and the thing that makes it avoidable is taking the sale off the table out loud, in the first two sentences. The call that works sounds like this:

"I am not calling to try to win the work back. I would just like to know what we got wrong, because if we are doing it to other people I would rather know."

Almost everyone answers that. They answer it because you have removed the reason to be polite, and because most customers who leave a long-standing supplier feel slightly bad about it and would like to say why. What you must then do is the hard part: do not pitch, do not defend, do not explain the circumstances behind whatever they name. The moment you explain, the conversation converts into the one they were dreading and the rest of the information stops.

Make the call yourself if you are the owner. Not the person who managed the account, who cannot ask neutrally, and not an office survey, which gets the polite answer.

The one question that gets past "price"

The first answer is usually price, because price is the socially easy answer that blames nobody. Take it seriously and then test it with one follow-up:

"If we had been the same number, would you still have moved?"

The whole card is in that question. A yes means it was never price, and almost nobody will keep insisting it was after saying yes - at which point "so what was it?" gets you the real answer. A no, given without hesitation, means price genuinely decided it, and that is a different problem with a different fix.

Listen for the hesitation as much as the word. "Well... probably, yeah" is a yes wearing a no's clothes.

What the answer usually turns out to be

Six causes cover nearly all of it. Each has a tell you can find in your own records after the call, which is how you confirm what you were told rather than trusting one person's memory.

The fourth row is the one that surprises owners. A job that went badly and was fixed technically is not resolved: the repair closed the ticket, and nobody ever went back and closed the relationship. Customers leave over that months later, and the record shows a successful repair, which is why the internal story never includes it.

The case: eleven years, a better number, and nineteen days

A shop loses a customer of eleven years, a property owner with three buildings. Word comes back through a supplier that the customer "got a better number."

The owner makes the call, opens with the sentence above, and gets price as the first answer. Then the follow-up: if we had been the same number, would you still have moved? The customer says yes, immediately, and then gives the real account without being asked twice.

What it actually was. Over the previous 18 months their maintenance visits had been covered by four different technicians. The last one did not know the building, spent a morning on something the regular tech did in twenty minutes, and a tenant complained about it. Separately, the customer had asked for a quote on a planned replacement and it took 19 days to arrive, by which time they had asked somebody else and that shop had turned it around in three.

Neither of those is price. Both are in the shop's own records and neither had ever been looked at.

The check against the records. Four distinct techs on the account in 18 months, confirmed. Quote sent 19 days after the site visit, confirmed. The tenant complaint appears in the notes with a closed ticket and no follow-up contact logged after it, which is row four of the table above sitting in plain sight.

Then the owner widened it, and this is the step that converts one loss into a repair. Six customers had gone in the two quarters just closed. Four of the six were multi-building property customers, the same customer type. Of those four, three had a quote outstanding longer than 14 days before they left.

What the numbers said, with their bases named. The shop's median quote turnaround across all quotes issued in those two quarters was 11 days, against a stated intention of two. On the narrower base of quotes to property customers only, the median was 19 days. Those are two different populations and the second is the one that matters here, because property customers are the ones comparing suppliers on a schedule rather than in an emergency.

The fix, and what it cost. A 48-hour turnaround standard on any quote above a defined size, with one named person responsible for issuing it rather than whichever estimator had a gap, and the clock started at the site visit rather than at whenever the file reached a desk. Cost: nothing but a rule and a measurement. It was not found by the loss of one customer. It was found by asking one customer why and then counting.

Noise or signal

Not every loss deserves this. Treating all of them as signal produces a shop that reorganises itself every time somebody moves house.

One loss is noise. Act on the call itself, not on the shop.

The pattern worth acting on is three or more losses inside two consecutive quarters that share a customer type or share a cause. Either one qualifies on its own.

Run the case above against that rule. Six losses in two quarters, of which four share a customer type and three share a cause, so it clears the bar twice over and the cause-based cluster is the tighter of the two. Had the six been six different customer types with six different reasons, the correct reading would be ordinary turnover at a shop of that size and the correct action would be none.

Two qualifiers on the rule. Seasonality is not a cause, so compare against the same two quarters last year before calling a cluster. And a cluster of the same customer type with different causes is a segment problem rather than a process problem, which usually means a competitor has aimed at that segment - see related: Being the Incumbent When a Challenger Arrives.

None of that triage is possible without a record, and this is the part that has to exist before the loss happens. Keep a loss log with five fields and nothing else, because a longer one does not get filled in: the date, the customer type, the stated reason, the reason the counterfactual question produced, and one line on what the records showed. Two minutes per loss, written the day of the call while you still have the words.

The log is worth its space for one reason that only shows up later: the third loss is where the pattern becomes visible, and by then nobody remembers the first two accurately. A shop working from memory reconstructs its losses as the story it already believes, which is the price story, and then finds the pattern that story predicts.

Whether to chase the return

Not in the same call. Asking for the work in the conversation you opened by saying you were not asking for the work costs you the credibility of every sentence before it, and word of it travels.

Chase it later, and only when two things are true: the cause is fixed, and you can name what changed in one sentence without it sounding like marketing. "We put a 48-hour rule on quotes and I own it now" is a reason to try again. "We would love another opportunity" is not. The approach itself is a separate job with its own method - see related: How to Re-Earn a Customer Who Left for a Competitor.

Be honest about the odds. A customer who left over a fixable process fault and has now had six months of adequate service from someone else is not waiting for you, and the realistic ask is a single test job rather than the account back.

The losses you should let go

Three cases where the right answer is to close the file and spend the effort on the cause instead.

Price genuinely decided it and the gap is structural. If they said no to the counterfactual question without hesitating, and the other shop's number is below what your cost structure supports, chasing it means buying the account. Whether their number is sustainable is a separate question with observable answers - see related: Competing Against a Shop That Does Not Know Its Own Costs.

They were price-led from the start. A customer won on price eleven years ago has been available to the next cheaper number for eleven years, and the tenure was luck rather than loyalty.

The relationship was with a person who now works there. When a tech leaves and a customer follows, you are not competing with a shop, you are competing with a relationship you allowed to belong to one employee. The account may be gone and the lesson is structural - see related: A Competitor Is Recruiting Your Technicians.

In all three, the loss is finished and the useful work is upstream of it. That is the actual discipline of this card: the call is not a retention tactic, it is an inspection, and most of what it finds is not about the customer who left.

References

• See related: How to Re-Earn a Customer Who Left for a Competitor, Being the Incumbent When a Challenger Arrives, Competing Against a Shop That Does Not Know Its Own Costs, A Competitor Is Recruiting Your Technicians
• See related: Knowing Which Competitor You Are Actually Competing With, The Callback That Prevents a Bad Review
• Trade-standard practice for quote turnaround measurement and account-continuity tracking in residential and light commercial field service