Capital
You buy a balance sheet and a customer list. You also inherit forty people's settled assumptions about how things are done here.
The purchase agreement lists what changed hands. The repositories, the contracts, the receivables, the name over the door. It is a long document, it is thorough, and it does not mention a single thing that will determine whether this works.
Because what actually transferred is a set of agreements nobody wrote down.
How a release gets approved. Which customer gets called back first. What the word urgent means here, as distinct from what it means anywhere else. Who you check with before promising a date, regardless of what the org chart says about reporting lines. None of it is in the data room. All of it is running the company.
Every company has two structures. The org chart is real enough — it sets pay, titles, and who runs reviews. The decision map is different: it is who people actually ask before they do something.
These overlap less than you would expect. In one business I know well, nearly every operational question routed through a support lead two levels below anyone with authority. Eleven years in the seat. She knew which customers would wait and which would call the founder at home, which meant she was functionally the company's risk department, and she appeared on no document anywhere as anything but support.
You find the decision map by watching, not asking. Ask who decides and people will describe the org chart, honestly, because that is the version everyone has agreed to believe. Watch where the questions actually go for three weeks and you will have something closer to the truth.
The org chart tells you who is accountable. The decision map tells you what breaks when that person leaves.
"We tried that already."
New owners hear this constantly, and it produces one of two reactions, both wrong.
The first is deference. You are new, they have been here twelve years, they say it did not work — so you drop it. Do that enough and you have paid a premium for a museum: all the constraints of the previous owner and none of the relationships.
The second is dismissal. They tried it with worse tooling, in a different market, before the platform existed. You are not them. So you proceed — and about a third of the time you find the original reason is still sitting there, and that it was never really about tooling at all.
The useful response is neither. It is a question: what exactly happened?
Not did it work, which asks for a verdict. What happened. Who did it, when, what specifically went wrong, and what was true at the time. Half the time the constraint has genuinely dissolved — the library is maintained now, the customer who objected churned, the engineer who quietly resisted it left in 2021. The other half it is structural and permanent, and you have just saved yourself a year.
Either way you now own the reasoning rather than the conclusion. That is the difference between inheriting a company and being inherited by one.
There is an old principle about not clearing a fence from a field until you know why it was put there. It is usually quoted at people moving too fast, which is fair, but it gets misread as an argument for leaving fences alone.
It is not. It is an argument for finding out — and then, quite often, taking the fence down.
Software businesses of this age are dense with them. A pricing tier grandfathered in 2013 that three customers are still on. A nightly job that must finish before four in the morning, and nobody remembers what depends on that. A feature built for one hospital in 2015, still shipping to everyone, still carrying its own branch of logic through every refactor since. A retry loop added after a bad outage that now papers over a fault worth fixing properly.
None of these were irrational. Each was a sensible answer to something real. The problem is that the something is gone and the answer is still here, drawing salary and slowing every release.
My rule, imperfectly kept: change nothing in the first ninety days that I cannot explain the original reason for. Not change nothing — change nothing I do not understand yet. It is slower than most plans allow. It has never cost me as much as the alternative.
This is also, incidentally, the entire argument for treating technical debt as an ownership question rather than an engineering one. Debt you understand is a schedule. Debt you inherited and never interrogated is a fence.
There is a version of this person in every company you will ever buy. Long-tenured, technically indispensable, visibly unenthusiastic about the new owner. Everyone flags him during diligence. He is described as resistant to change.
He is almost never resistant to change. He is protecting something.
Usually one of three things: a way of working he believes is the only reason quality holds, a person he is shielding from consequences he thinks unfair, or a promise the founder made him that he assumes you will not honour. Sometimes it is simply that he has watched the industry do this to people like him before.
Ask what he is worried about. Then — this is the part that matters — fix one of the answers inside a month. Not all of them. One, visibly, early. He is the most efficient piece of communication infrastructure in the building, and what he says about you in week six will still be operative in year three.
You cannot buy that. It was not in the agreement. But you did inherit it, and it is currently deciding whether the plan you paid for is going to happen.
I have come round to a view that would have sounded soft to me some years ago.
The financial statements are not the business. They are a record of decisions the business made, at a resolution too coarse to show you any of them. Gross margin is what happened when forty people made judgement calls about what to build and what to decline. Net revenue retention is what happened when someone decided which ticket to answer first.
You bought the judgement calls. The statements only told you how they had been going.
Which means the first hundred days are not really about strategy. They are about learning the ten or fifteen unwritten agreements that generate the numbers — who decides, what gets protected, which fences still have live reasons behind them — before adjusting any of it.
Get that wrong, and you spend year two rebuilding something you removed in month three without knowing it was load-bearing.
Get it right, and you usually find the business already knew what it needed. It was waiting for somebody to ask.
— Ahmad
Talk through the decision with me.
I set aside a small number of 30-minute founder conversations each month. We'll use the time to understand where the business is today, pressure-test one important question, and identify what deserves attention next.
One operating framework, every Thursday. Free, and you can unsubscribe anytime.