Three seats, one business

Technician

Two things get treated as one thing, and the mistake costs people years. The first is how good you are at the work. The second is where in a business your value actually comes out.

The person hears the word technician and hears a rank, somewhere under manager, further under founder, and takes the label as a demotion. The word marks where the value gets made, nothing else, and a superb practitioner and a mediocre one sit in the same seat.

Where the Value Actually Gets Made

The technician is the one for whom the work is the doing of the craft: the value is produced by their hands, their voice, their presence, in the act itself, and if you take them out of the act no value is produced. The manager builds the structure that lets work happen without any specific person, the processes, the roles, the accountability, the machine that runs on whoever occupies the seat. The entrepreneur, who I call the founder, holds direction: what this becomes, where it goes, what it is for.

A master craftsman is a technician. A surgeon operating is a technician. A coach mid-session, producing a change in a client, is a technician in that moment, and a good one.

Skill does not move you out of the seat.

The Frame Where the Hand Is the Value

What keeps a person in the seat for years is not the craft. It is a frame that came in with the craft. A training that puts the hand first, everything manual before anything else, puts the work in deep, and it puts in a second thing in the same lessons without ever naming it: the artist frame, where the work and the maker are one object.

Take the maker out and there is no product. So the question of how this runs without me does not arrive as a hard question. It does not arrive.

A person can run a good year on craft, price it, deliver it, and never once need the word. The frame is doing what it was installed to do. It keeps the hand as the value, and anything that lifts the hand out of the act reads as damage to the work, so the years go into deepening the craft and nothing gets built under it.

Argument does not reach a frame. What I saw, I saw from inside it, and it was already read by the time it got to me. Mine came apart from outside it, on a plain remark from a person who was not standing in it.

The Seizure and the Assumption Under It

There is a second move that scrambles the seats, and it has a name. A skilled technician concludes that because they can do the work, they can build a business that does the work. That is the entrepreneurial seizure.

The error sitting inside it has its own name, the fatal assumption: knowing the craft is not knowing the business of the craft. They are two bodies of knowledge with almost no overlap, and the second is not harder than the first, it is a separate discipline the person has never practised, demanded at the exact moment the craft income comes under pressure.

I do the work well, so I can own the thing that does the work.

It gets said in every craft with the noun swapped. The vocabulary changes to CEO. The seat does not.

Not the market. Not the product. The hands, and whether anything in the business produces value with those hands out of the act.

Shelf

The frame is Gerber’s. His subject is why small businesses fail, and his answer is that the person who starts one is not one person doing one job. Three kinds of work have to be done, one body has to do all three, and the three pull against each other, so the business gets run by whichever one the owner is most comfortable being.

I did not stumble on him. I picked him, for three reasons that still hold. He built the frame for exactly this case, one skilled person turning into a business, and not for a corporate team.

He sorts people by their relationship to the work rather than by personality or by skill level, so the frame cannot tell me what anybody is like, only where their value comes out. And he carries the fatal assumption inside it as a working diagnosis rather than as a description, which fits most of the people I have watched try this.

It has a hole in it, and I found the hole by using it wrong. There is a function I kept needing a seat for and he has none: the room work. Getting into rooms, being trusted once inside them, holding access, moving a person who had no reason that morning to move.

In his three that is not the craft, not the structure, not the direction, and I spent a long time filing it under one of them and watching it slide back out. That is why I read the others. Adizes had already put the seat in.

He runs four roles instead of three, and the four letters are the model’s name, PAEI. The producer delivers the result. The administrator builds the systems and runs them.

The entrepreneur generates direction and change. The integrator builds the relational fabric, aligns people, makes a room cohere. Two of his claims matter for the seat question and both are stronger than they look.

First, nobody holds all four at strength. His whole argument is that the complete manager does not exist and the roles have to be assembled across people rather than found inside one. Second, he names what each role does when it is held alone, and for the entrepreneur held alone the name is the arsonist: direction with no administrator under it, so a thing gets lit, the carrying of it belongs to somebody else, and by the time anyone works that out the one who lit it is standing at the next one.

Belbin came at it from the other side, by measurement. He ran management teams against each other in a business game at Henley and recorded which combinations won, and nine roles fell out of the watching rather than out of a theory. Two of his findings stuck.

He called the first the allowable weakness: each role carries a cost bolted to its strength, and that cost is the price of the strength rather than a fault to be corrected. One of the nine is the resource investigator, the one who goes out, develops contacts, brings the outside in, and the weakness allowed to that role, in Belbin’s own phrasing, is losing interest once the initial enthusiasm has passed. The second finding is that two people holding one role inside a team compete destructively, so a duplicated seat is worse than an empty seat.

What the field did with him afterwards is a questionnaire people fill in about themselves, which is the reverse of how he got the roles in the first place, and worth knowing before somebody hands you the test.

Wasserman worked from a dataset of thousands of founders and put a number on where founding teams break. Around two thirds of the failures in high-potential ventures trace back inside the team rather than out to the market or the product, and the thing that breaks is one mechanism: the psychological contract, the version of the deal each person carries in the head and never converts into a stated one. Each founder holds a different unwritten version, everybody proceeds, and it holds until one of them reads a breach in it.

That is the terms step skipped, and it does not arrive looking like a terms problem. It arrives looking like a person becoming difficult.

Then there is the data on what founding actually pays, and it is the part almost nobody quotes. The administrative census of employer businesses is not a survey, it is every one of them from birth, and the shape barely moves across decades or across booms and slumps: about a fifth gone inside a year, about half inside five, roughly two thirds inside ten. A failure rate that steady is structural, not cyclical.

Hamilton, in the Journal of Political Economy, took the other question and compared self-employed people against wage employees of comparable ability over ten years, and found the median self-employed person earned around a third less than the same person would have earned on a wage. Astebro sharpened the shape afterwards: the returns are lottery-shaped, a low middle and a long tail, and entry carries on anyway. The reading the field settled on is that people are buying autonomy and paying for it in the wage they gave up, and most of them never say out loud that this is the transaction they made.

I read that and the shared-office floors I have worked on stopped looking to me like businesses being built badly. The day-and-night effort was the autonomy being paid for, in the effort itself.

The last two say why the payment keeps getting made. Cooper, Woo and Dunkelberg surveyed founders at the start rather than at the end: four in five put their own odds of success at seven out of ten or better, a third said certain, and the same people rated other businesses identical to theirs far lower. Camerer and Lovallo then reproduced the mechanism in the lab and gave it a name, reference group neglect: a person enters a competition on their absolute read of themselves and does not price the fact that everyone else entering was selected by the same confidence.

Lazear came from economics and asked who founds at all, and found the founder is a jack of all trades, that a variety of prior roles predicts founding while specialists become employees, and that the standard defect of the profile is a business that caps at the generalist’s own hours until one specific module gets built. Shane assembled the government data nobody wanted, in The Illusions of Entrepreneurship, and what it says about the ordinary founder is flat: a crowded industry, a low barrier, nothing to tell him apart from the others in it, undercapitalised, no intention to grow. Which industry a person entered predicted survival more reliably than anything about the person who entered it.

Five frames on the shelf, and I kept the one that cannot tell me what anybody is like.

Gerber sorts by relationship to the work, so he cannot psychologise anybody, and the fourth seat is a stretch I do not need. What the frames could not give me and the numbers did is the line under all of them: the entry test selects for confidence, and it does not test for seat coverage at any point. Nobody is asked on the way in which of the three seats they hold and who holds the other two.

That puts the failure rate somewhere other than the market: seat coverage in a population, which is low, meeting an entry process that never measures it.

What it cost me to find that true is a long stretch of reading one arrangement as ambiguous when it was my frame that had no slot for what was being supplied inside it. The reading wobbled every time. I took the wobble as the arrangement being hard to read, and it was the frame’s gap showing.

A frame with no seat for a function hands you back something that will not resolve, and you go on staring at the wrong end of it.

Manager

The manager seat ran through the whole thing, and it is the one nobody claims, because there is no story in it. Nobody leaves a job to become the structure.

The person who works for a living holds one seat well, mostly. A craft, a domain, a relationship function. Employment is the technology that supplies the other seats to a one-seat holder, and underneath the salary that is what a job is:

• somebody else fixes the want, so nobody at the desk has to find out what is worth building
• somebody else writes the terms, so they exist as stated things and not as guesses carried in a head
• somebody else does the arithmetic on what it costs and what is left
• somebody else reads the position, where the thing stands in its field and against who else is standing there
• and the person at the desk supplies action inside that frame

Founding removes the supplied structure, all of it, on one day. The two seats the person never held do not get occupied when they leave. They go empty, and the person keeps doing the one thing they know how to do, harder, for years.

The mechanism is short: the ordinary entrant is a one-seat holder who removed the structure that made their one seat productive.

Read the standard list of why businesses die and it is that sentence, itemised. No market need is the want never fixed, a want built out of the founder’s own head and never checked against anybody who would pay for it. Ran out of cash is the manager seat empty on the money axis, and the usual telling has the direction backwards: nothing in the business priced what the business was consuming while it was consuming it, and the end of the money is that arithmetic arriving all at once.

Team failure is the terms never stated. Outcompeted and mistimed is the position never read before committing. And working day and night for years with nothing viable coming out is action running with none of the four under it, motion that cannot converge because nothing was specified for it to converge on.

Coverage, absent, does the billing, and the market takes the name for it.

I have a manager layer and it is not human. I built it out of machines: a thing written once and left to run everywhere, so a change lands in every place at once without me standing in any of those places. And the definition that counts is narrower than the word suggests.

The manager seat does not require employees. It requires the business to run on systems instead of on the owner’s daily reinvention, and the function of the seat is converting fresh effort into reusable process.

The proof of it is dull. A client asks for direction material, and what would once have taken a day of thinking goes back inside five minutes: the position, the format, the delivery style, all saved, so the thing got recalled instead of thought. That is the seat, working.

It took years to build and it started compounding in front of a client, in a language I was not working in.

I delegated, completely, to the only workforce available to me. A person you hand work to has a state, and you carry that state while they carry the task. My history had priced that high enough that I could not buy it.

So years of manager instinct went into machines instead of into staff. Machines were the employees whose states I did not have to carry.

And the bottleneck sits exactly where the machine stops. The tacit element does not transfer. I can write the structure, the order, the checks, and the last pass still comes back to my own hands.

That gap is narrower than it felt for a long time: the human version of a function that already runs fluently on machines.

Everything but the last look hands over.

I check one thing first now, on any business somebody asks me about. Find which of the three seats is running empty, then find out whether the empty one is covered by a person, covered by something built, or covered by nothing. A seat covered by nothing sends no notice.

It bills in hours, and it takes them from whoever is standing nearest.

Founder

The founder seat is the one people take by vocabulary.

The founder who holds it needs the manager layer underneath, because direction with no execution structure has nowhere to land except back on the founder’s own hands. So the work comes home. Each thing they start arrives at their own hands to be done, and hands in the act is the technician seat, whatever the card says.

All the execution routes to the centre, and it routes there because there is no structure anywhere else that could hold it.

Claimed is not held.

The seat runs the other way too, and that one took me longer to see, because it does not look like anything. A seat can be held at full strength and pointed at somebody else’s objects. Mine was held early and at speed, want-driven, jobs changed and stacked and then converted into a company and a move abroad, all before the seat had a name I could use for it.

Then it went to work on other people’s ventures and stayed there. Clients got direction the day they asked for it. My own wants I left unstated, because stating them is the step I could not do, and a seat that cannot be aimed at a want of mine gets aimed at somebody else’s.

Nothing was empty. The seat was aimed outward and it delivered there, which is why it shows up in no audit anybody performs: it is producing, just not here.

The relationship function lives here too, once you refuse the fourth seat. Rooms, trust, access, the credibility that makes a person move.

In a business run by one founder that is narrative, and narrative is direction carried into a room by a body. That is founder work, and no department holds it.

Three seats, one business.

The claim I hold as settled: a founder-driven business has to contain all three to some degree or it cannot grow past that founder’s hands. Not equally, and not inside one person. Somewhere, held by somebody, or built into something that runs without anybody.

And the three do not arrive together. One gets held for life, one gets held early and then aimed at other people, one gets built quietly in a medium nobody counts as management, and a person can be running all three for years while answering no to the question of whether they hold any.

So the question I still carry is not the one I started with. By the definition that counts, systems instead of daily reinvention, the manager seat is covered, and what is left uncovered is a narrow strip at the end where the last look will not hand over. The founder seat is the open one.

It ran outward for years and it is turning now, late, onto my own objects, and the distance between those two directions is roughly the size of the business I did not build. All three were held the whole time. What the founder seat never got was an object of mine to work on, and the price of that is a company that does not exist.

What it does with one handed to it this late, I have not found out yet.