At Quique's company, a distributor of professional cleaning products in Valladolid, northern Spain, clients didn't leave with a slammed door. They left in silence. And the ones who leave in silence are the worst, because you don't find out until much later, when there's nothing left to do.
The case that opened his eyes was a chain of care homes, an eight-year client, one of the good ones, the kind that ordered every fortnight without fail. One day, in a meeting about the numbers, someone asked about them. Quique looked. They hadn't placed an order in four months. Four months. Nobody had noticed, because nobody watched who stopped ordering; they watched who ordered. The care home hadn't complained about anything. Quite simply, a competitor had made them an offer, they'd started trying it, it had gone well, and they'd switched. Without a sound.
Quique called. Too late. The man at the care home was kind, almost apologetic: "It's just that another supplier came in, we tried it, and we were comfortable. If you'd called us at the time, maybe…" That "maybe" hurt Quique more than the loss. Because it meant it could have been avoided, and it wasn't avoided for not looking.
What a client does before leaving
A client who leaves almost never leaves all at once. They give warning, even if they don't mean to. The thing is they warn with facts, not with words, and the facts have to be read.
The care home had given warning. First it stretched the gap between orders: from every fortnight to every twenty days, then every month. Then it lowered the amount: it started ordering only some things, the ones the new supplier didn't yet stock. Later it stopped ordering the good ranges, the higher-margin ones, and kept only to the basics. Each of those steps was a signal. Lined up, they told a very clear story: this client is leaving. But nobody lined them up, because they lived loose, one in each order, lost among the hundreds of orders from everyone else.
What for a person is impossible to see — the slow change in one client's behaviour among hundreds — is exactly what data does best, if someone asks it to look.
What he asked for and what he needed
Quique called me asking for "a system to win new clients". He'd lost a few and wanted to replace them. I understood, but I suggested looking at the other side first.
—Before spending to bring new clients in the front door, let's close the back one, the one they're leaving by without your noticing. Keeping the ones you already have costs a tenth of what it costs to win a new one. And the ones you already have, you also know.
What we built wasn't anything exotic. We took his order history — which he already had, in his usual system — and put a watchman on top of it. A system that looks, client by client, at their normal buying rhythm: how often they order, how much, what. And that flags when a client steps out of their own rhythm. Not the average rhythm of everyone: their own. Because a client who orders monthly isn't a problem if they always ordered monthly; the problem is the one who used to order weekly and now orders monthly.
A warning in time, not a year-end report
The key to this isn't the analysis. It's the timing. A year-end report telling you "you've lost these twelve clients" is useless, because you've already lost them. What's useful is a warning at the moment the client starts to leave, when there's still time for them to stay.
That's what we gave Quique's system: every Monday, a short list. Not of all clients, which would be noise, but only of those who that week had given a signal of cooling off. "These five clients have changed their pattern. Take a look." Five names. One call each. Fifteen minutes of a rep's time.
And the important part: the call wasn't to sell. It was to ask. "How's everything? Is the service working? Is there anything we could do better?" Sometimes there was nothing and the client had just had a slow patch. But other times, in that call, out came what was happening — a price, a delay, a competitor circling — in time to do something.
The ones who stayed
In the first quarter, the Monday list flagged two clients cooling off the way the care home had. One was leaving over a delivery-times problem nobody had escalated. It was fixed. The other had a competitor's offer on the table; Quique matched part of it and improved something else, and they stayed. Two clients who, with the old method, would have left in silence and been discovered four months later, with a "maybe".
Quique told me something a few months later that sums up the change well.
—Before, I found out a client had left when they'd already left. Now I find out they're thinking of leaving while I can still do something.
I told him that sentence applies to almost everything in a company. That most problems don't warn you with an alarm: they warn with a small, slow change that's in the data long before it blows up. The work isn't having more data. It's putting someone — or something — to watch the data you already have, and to speak up in time.
Because clients, like almost everything, rarely leave suddenly. They leave slowly. And slowly means there's time, if someone is watching.
Closing the back door
Almost every company watches who buys and hardly any watch who has stopped buying. But keeping a client you already have costs a fraction of winning a new one, and the one who leaves in silence is always discovered too late.
Clients warn you before they leave, but with facts — orders further apart, lower amounts, less of the high range — not with words. Those signals are already in your history. Putting a watchman that lines them up and warns you in time turns a silent loss into a timely call.
Would you know today which clients are starting to leave?
If you only find out a client has left once they've already gone, there's room to spot it sooner with the data you already have. We can look at it together.
Let's talk