When the job stops fitting, or the shape of the team does

What it says

The old joke is that people rise until they're no good at it. This is the version with numbers behind it: 38,843 salespeople across 131 firms, and 1,553 promotions into management. How well somebody sold strongly predicted getting promoted — and it predicted their team doing worse afterwards. The same researchers ran it the other way round: rank candidates on how well they'd probably manage, and you'd have ended up with managers about 30% better. The best salesperson keeps being the obvious choice. Obvious is doing a lot of work in that sentence.

Where it comes from

Benson, Li and Shue, in the Quarterly Journal of Economics in 2019. The link goes to the free working paper. One honest limit, and it matters: this is sales, where what one person produces is unusually easy to count, so the effect is probably sharper there than wherever you work. It shows firms leaning too hard on how somebody is doing right now. It doesn't tell you what to lean on instead when the next job is one nobody can measure cleanly.

What to do with it

Ask what evidence you're promoting on, then ask how much of the new job that evidence is even about. Being the best at this job is evidence about this job. If the new one is mostly made of things nobody has ever watched this person do, you're guessing. Guessing is allowed, as long as the person being guessed about knows that's what this is. Guessing while everybody agrees it's obvious is the part that goes wrong.

When it's the wrong lens

When the step up really is the same work with more of it. Then how they're doing now is exactly the right evidence, and this card is a reason to hesitate over nothing.

What it says

Conway's own sentence is that any organisation designing a system is stuck producing a design that copies how that organisation talks to itself. Plainly: the joins in what you build turn up wherever the joins between your teams are. Two teams who barely speak will build two halves that fit badly, and they'll fit badly at exactly the point nobody talks about. So splitting a team is a decision about the product. You just made it while thinking about people.

Where it comes from

Mel Conway, 1968, in Datamation — the link goes to his own site. Worth knowing: Harvard Business Review turned the paper down for not proving its point, and that's a fair description of it. It's an argument, not a study. There's later work supporting it which we haven't read, so treat this as something to test against your own experience rather than as a finding. Sixty years of people quoting it isn't evidence either.

What to do with it

Before you reorganise, say out loud what the new shape makes easy and what it makes expensive. Draw the thing you want to build. Then draw the teams. Then see whether the lines land in the same places. If they don't, you'll be fighting your own structure for as long as it lasts, and the structure usually wins.

When it's the wrong lens

When one person is deciding whether to take a bigger job. This card is about structure, and it has nothing useful to say about anybody's next step.

What it says

Two ceilings, and mixing them up is the usual mistake. One is the structure — promotion isn't likely, usually because there's nothing above. The other is the work — the job has stopped holding anything this person hasn't already done. A review of 72 studies across forty years ties both to people being less satisfied, less committed, doing worse, and leaving. The thing sitting in the middle of all that was feeling unsupported. That's the half you can do something about.

Where it comes from

The idea goes back to Ference, Stoner and Warren in 1977; the link goes to the 2019 review in the Journal of Vocational Behavior. Two limits worth carrying with it. Almost all of it is people saying they feel stuck rather than anybody checking whether they are, and most of it was measured once rather than followed over time. And the review's own finding is the useful warning: this only bites for people who mind about being promoted. Plenty of people don't.

What to do with it

Work out which one it is by asking what would have to change. If the answer is a title or a level, it's the first kind, and sometimes there's genuinely nothing to do but say so. If the answer is that the work has gone stale, it's the second kind, and that one usually gets fixed without anybody being promoted at all. It's also the more common one. And it's the one people reach for a promotion to solve.

When it's the wrong lens

When somebody's perfectly happy where they are. A plateau is only a problem for somebody who wants to be climbing, and treating being content as a problem to fix is its own kind of insult.

The move

List what somebody in the new role actually spends a week doing — real activities, the dull ones included. Then mark the ones the person stepping into it has done before. Count what's left unmarked.

Why it works

It turns a vague feeling about whether somebody's ready into a short list of things nobody knows yet, and every one of those can be tested cheaply — a project, two weeks standing in, one of the activities borrowed for a month. It also makes it obvious when a promotion is a completely different job wearing the same ladder.

How it fails

Writing qualities instead of activities. Nothing checks "strategic thinking" or "leadership" against anything, so the list quietly agrees with whatever you already thought.

The move

Two questions: which parts of the last six months would you want more of, and which would you happily never do again? Ask them of yourself, or ask them out loud. Then look at whether the next step is made of the first list or the second.

Why it works

It's answerable. "Where do you see yourself in three years" asks somebody to perform being ambitious, because everybody knows the right answer is upwards. Ask what the work actually feels like and you get something true instead. Quite often the honest answer isn't upwards at all. It's sideways.

How it fails

Asking it after the decision is made and the role is already being advertised. Then it's a conversation nobody can act on, and everybody in it knows by the second question.

The move

Ask what would have to change for this to feel like it's going somewhere. If the answer is a level or a title, somebody has to say plainly what is and isn't available. If the answer is about the work, start there, because that one can usually be changed this week.

Why it works

Promotion is the only word most people have for a feeling with at least two causes behind it. Pull them apart and some of these conversations stop needing a vacancy to end well. Which is just as well. There usually isn't one.

How it fails

A hint standing in for the hard sentence, because the hint costs nothing today. It buys a few months and costs the relationship, and the trust goes here rather than in the later conversation where somebody finally admits it.