Buyers Can Model the Revenue. They Can’t Model Who Stays.
Every metric in a services business is produced by a person. Why people risk is the part of a deal buyers worry about most and discuss least.
After the first of these pieces went out, a former colleague who once worked for me wrote privately. She agreed with most of it, and then she corrected it. Every number I had described, she pointed out, is produced by a person. Wallet share grows because someone earned the trust. A delivery model only changes if someone leads it. She was right, and she had put her finger on the part of the argument buyers worry about most and discuss least.
A buyer can model the revenue. What a buyer cannot model is whether the people who produce it will still be there in year two.
What buyers test on people
Management quality is priced. So is its absence. Every pack has a leadership page with confident photographs and a paragraph on culture, and buyers read past it quickly, because what they want to know is not on the page.
They want to know how deep the bench is behind the top team. They want to know whether the client relationships belong to the firm or to a handful of individuals, which comes down to one question: who does the client call, and what happens when that person leaves. And they want to know who the offices listen to, because that is who will land the plan.
Every metric has a person behind it
I wrote earlier that logo retention is gravity and the verdict is in the wallet. That was about the statistics. Behind every statistic is a relationship, and the relationship is personal before it is institutional.
The largest account I ever managed became the group's largest account, and it started as one relationship with one person, a head of expansion, for one service. That is where most accounts stay. What changed this one was not the relationship but what we built around it.
Their expansion ran in layers: one function first, then the next, then the next, in every new market. I mapped the layers and the people who ran each of them, and made sure we were the firm every one of those people called, not just the one who had called us first. Then I put the account on paper: a plan for every layer, and a global P&L for the account as a whole, so it could be reviewed, resourced and grown like a business rather than protected like a friendship.
That is the whole difference. A relationship is worth what the buyer thinks two people will do next. An account with a map, a plan and a P&L is a business. Making a relationship institutional without making it impersonal is one of the hardest things a services business does, and one of the most valuable.
The office below critical mass
The problem the leadership page never describes is arithmetic. A services office needs roughly 25 to 30 people before anyone in it has a backup. Below that, every delivery person is a single point of failure, however junior. In a ten-person office, one absence is a tenth of the capacity and very often all of one client.
I ran footprints with quite a few offices below that line, and I saw the same failure more than once. It was rarely resignation. It was someone burning out in the seasonal peak, or going on sick leave in the middle of it, and the work they carried had nowhere to go. That created two problems at once. One was the well-being of the person, and of the two or three colleagues absorbing the load. The other was that you cannot hire a temporary specialist in a regulated function at a week's notice. The work was late, the client noticed, and the office spent the next year rebuilding trust it had taken five to earn.
The only durable answer I found was to get the office over the line. Where a local business was available, we acquired it. Where it was not, we grew the office as fast as the market allowed, because every month below critical mass was a month of exposure. That is a slow and expensive fix.
I did not always see it coming. The monthly pack showed utilisation, and utilisation above target reads as good news right up until the person producing it stops. Looking back, high utilisation in a small office was the warning, not the result.
This is where AI changes the people story, and it is the part of the AI conversation I hear least. Everyone talks about headcount. The more important effect is that AI lowers the critical mass. If the routine work in a twelve-person office can be carried by the platform, that office starts to behave like a thirty-person office: the people in it are freed for the work only they can do, and the single points of failure become fewer and better protected. That is a resilience story. For the offices where the last mile actually gets walked, it is the most important thing AI will do.
Fewer, better, harder to keep
The labour arbitrage era optimised for many adequate people doing standardised work. That was the right design for its time, and it built the businesses most of us ran. It is not the design for what comes next.
As processing moves to machines, the work that remains is judgement, exceptions, relationships and local knowledge. That is the work that needs the best people rather than the cheapest. AI raises the floor of what a person in the business has to be, and it does so at the same time as it reduces how many of them there are. Technology takes out headcount and raises the importance of every remaining head.
I wrote in the first piece that a cost advantage built on labour arbitrage is a melting asset. This is the people version of the same problem, and I say it as someone who built those centres and would build them again. They were designed to hold thousands of adequate people cheaply, and every underwriting model still assumes them. The next model needs a few hundred excellent people, and nobody has yet worked out how to pay, promote and keep them inside a cost structure built for the opposite. The centre that was the sector's biggest cost advantage is becoming its biggest people risk. Finding, keeping and leading the people who remain becomes most of the job.
My former colleague put it in one line. AI makes people more important, not less.
What the leaders I rated did
They hired and promoted for judgement rather than throughput. This sounds obvious and is rarely done, because throughput is measurable and judgement is not.
They made relationships institutional without making them impersonal: a second name the client knew, a plan for every layer, and a P&L with the client's name on it rather than the salesperson's.
They treated retention as a commercial metric. Attrition in the teams serving the top accounts is the earliest warning of wallet share loss I know, and it arrives a year before the revenue does.
And they showed up, because the person the office listens to has to be in the office.
The conversations people remember
I spent a great deal of my career on the numbers. Targets, margins, utilisation, the monthly pack. When my former colleague wrote to me, what she remembered from our time working together was none of that. She remembered the conversations about culture, about who to keep and who to develop, about what kind of place we were trying to build. Looking back, those conversations were the work. The numbers came out of them.
That is the part the buyer cannot model. The buyer asks what the offices do differently. The better question is who is still there to do it.
Originally published on LinkedIn, 9 September 2026.
