They are not a bad hire, they are in a bad seat

Your head of sales is discounting too hard. You have discussed it three times. You are starting to think you hired wrong.

Before you act on that, answer one question. Would a different person in the same chair, with the same information and the same targets, behave the same way? If yes, replacing them buys you a recruitment cycle and returns you to exactly here.

Why these three models

The decision is whether to change the person or the position. The features that fire are a behaviour that persists across occupants, an incentive structure nobody designed on purpose, and a trajectory that has never been plotted.

Three lenses. The archetype produces a complex answer about which recurring structure you are inside and what its named exit is. Rule design produces an equilibrium answer about what the seat is actually paying for. Behaviour modes produce a cycle answer about the shape of the problem, which decides whether it is a problem at all. The first and second disagree productively: one says the structure is generating this, the other says structures are designed and yours can be redesigned.

1. Bounded by the seat

Start with the observation that makes the rest of this article possible.

People behave rationally given the information, incentives, goals and pressures that reach their position. Not rationally in the abstract. Rationally given what that chair can see and what it is rewarded for. Meadows puts the consequence in the second person, and it is worth sitting with: if you become a manager, you will probably stop seeing labour as a deserving partner in production and start seeing it as a cost to be minimised; if you become a financier, you will probably overinvest during booms and underinvest during busts, along with all the other financiers.¹

She adds the line that closes off the comfortable escape. You might retain your memory of how things look from another angle and burst forth with innovations that transform the system, but it is distinctly unlikely.¹

So the diagnostic is not about character and it is testable. Has this behaviour appeared in more than one occupant of this seat? Does it appear in the equivalent seat at other companies? Would you do it, honestly, if you were compensated on this quarter’s closed revenue and could not see next year’s churn?

Where the answer is yes, this is a structure producing a behaviour, and the archetype has a name and a stated exit. Rules that produce the appearance of compliance while distorting the system are rule beating, and the way out is to redesign the rules so creativity runs toward their purpose rather than around their wording.¹

System traps, the archetype lens

• Assumes: recurring structures produce characteristic behaviours largely independent of who occupies them.
• Fits because: the behaviour has survived a conversation, and probably an occupant.
• Breaks when: the behaviour genuinely is idiosyncratic, which happens and which this lens will talk you out of seeing.
• Evidence: grade B. The archetypes are widely observed and individually plausible, and have not been formally tested as a set.
• Counteracts: diagnosing character when the seat explains the behaviour.
• May reinforce: excusing genuine underperformance as structural, which is the failure mode of everything in this article.

2. What the seat is actually paying for

The second lens tells you what to change, and it is more specific than “fix the incentives.”

Start from the behaviour you want and work backwards to the rules that would make it the self-interested choice. That is the whole discipline. Applied to a seat, it decomposes into three questions that are worth asking separately, because they fail separately.

What can this seat see? A salesperson who cannot see churn is not ignoring churn. It is not in their world. Adding the number to their dashboard is an information change and it costs nothing.

What is this seat paid for, precisely? Not what the role description says. What the compensation formula computes. If it computes closed revenue this quarter, it is buying discounting, and it will keep buying it after you replace the person.

What can this seat decide without asking? Authority limits generate as much behaviour as incentives do. A person who must escalate every non-standard deal will standardise aggressively, which looks like rigidity and is actually a rational response to a bottleneck.

The trap to avoid is the obvious repair. Adding a churn penalty to a revenue formula usually produces a new distortion rather than a fix, because the seat now optimises a two-term expression and the second term is easier to game than the first. Systems produce exactly and only what you ask them to produce.¹

Mechanism design, the seat lens

• Assumes: you can change the rules, and behaviour follows self-interest under whatever rules exist.
• Fits because: the seat’s information and payoff are yours to set.
• Breaks when: the binding incentive is not one you control, such as what the role is worth on the external market.
• Evidence: grade A as theory. Implementations fail on enforcement and on unmeasured terms far more often than on design.
• Counteracts: exhortation as a management tool.
• May reinforce: over-engineering compensation, which adds gameable terms faster than it removes them.

3. Name the shape before you blame anyone

The third lens runs first in practice and is almost always skipped.

Plot the behaviour over time before diagnosing it.² Discounting that has grown steadily for six quarters is a different object from discounting that spiked last quarter, and they have different causes. Steady growth points at a structure with no balancing loop. A spike points at an event. Oscillation points at a delay, usually between a target being set and its consequences being visible. A slow drift downward in your standard for what counts as an acceptable discount is drift to low performance, and its exit is to anchor the standard to the best past result rather than to the trailing average.¹

There is a second reason to plot before diagnosing, and it is about you rather than them. Drawing the behaviour over time is the first step of any honest modelling process, and it is skipped precisely because everyone in the room already believes they know the story.² The plot is cheap and it regularly contradicts the account that three people had already agreed on before the meeting started.

This matters here specifically because the seat diagnosis is seductive. It is more sophisticated than blaming a person and it will fit almost any trajectory if you do not check. Naming the shape first is what stops a structural story being applied to a one-off.

Behaviour modes, the shape lens

• Assumes: a trajectory belongs to a small set of shapes, each implying a different structure.
• Fits because: you have the data and have been arguing from impressions.
• Breaks when: the series is short or noisy, which at small volumes it usually is.
• Evidence: grade B. Descriptive and useful, with no strong predictive claim attached.
• Counteracts: explaining a pattern before establishing that it is one.
• May reinforce: reading structure into three noisy points.

The levers, cheapest first

• Plot it. Ten minutes. If the behaviour is a spike rather than a trend, stop reading about seats and find the event.
• Run the substitution test. Ask whether the previous occupant did this, and whether the equivalent person at a company you respect does it. Two answers, no cost, and it usually settles the question.
• Change what the seat can see before you change what it is paid. Information is reversible and cheap. Compensation is neither.
• Read the formula, not the job description. Whatever it computes is what you are buying, and the gap between the two is where the frustration lives.
• Raise the authority limit before adding a rule. Much of what looks like poor judgement is a rational response to needing permission.
• Then replace, if the substitution test comes back negative. Sometimes it does. This article is not an argument against ever changing a person.

What to do this month

Do now, sized at one hour, effect immediate. Write down what the seat can see, what its formula computes, and what it can decide alone. Three lines. Reversible, free, and dominant across every scenario about whose fault the behaviour is.

Hedge, where the premium is the whole loss, live before the next cycle. Add the one number the seat currently cannot see to its regular view. If the behaviour was never informational you have added a row to a dashboard, and that is the entire downside.

Defer and trigger, size fixed now. Do not redesign compensation this quarter. Pre-commit the trigger: if the behaviour persists a full quarter after the information change, the formula gets opened, and it gets opened with a named list of what it must not reward. Fix that list now, while you are not annoyed, because a compensation change designed during a bad quarter encodes the bad quarter.

Note the arrivals. Information changes land in weeks. Compensation changes land at the next period boundary and then take a quarter to show, which is long enough for you to conclude they failed.

What usually happens next

Run the break test first. Has a rule changed, has an actor entered or left, has a measurement become a target? If you started reporting this behaviour to the board, expect it to improve in the report before it improves in reality, and the historical series to stop describing the same process.

If nothing broke, the pattern is reliable and slightly bleak. The person is replaced, the successor is given a month of goodwill, and the behaviour reappears in the second quarter. The usual conclusion drawn is that the market for that role is bad. The usual truth is that the seat was never opened.

Subtract the counterfactual before crediting a replacement. A new occupant who arrives alongside a new target, a new dashboard and a fresh quarter is four changes, and the person is the most visible and least likely of them.

What this ensemble cannot see

All three lenses treat the seat as the unit and the occupant as interchangeable within it. Neither is quite true.

People are not interchangeable, and the substitution test is a thought experiment answered by the person running it, which makes it exactly as reliable as their honesty about their own hiring. A founder who does not want to fire someone will find the seat explanation persuasive. A founder who does will find the character explanation persuasive. Nothing in this framework corrects for which of those you are.

There is also a limit on the archetype library. The traps are widely observed and have not been tested as a set, and they fit almost any organisational story once you know them. That flexibility is what makes them useful for generating hypotheses and what makes them poor evidence for any particular one.

And one property none of these models contains: a person who has been managed as though the problem were their character will eventually behave as though it is, whatever the seat does. By the time you run this analysis, three conversations may already have made the structural fix insufficient.

The one action that survives the ignorance: before the next conversation about this person, write down the three seat facts and show them to someone who does not know the individual. Ask what they would predict about the occupant’s behaviour. If they predict what you are seeing, you have your answer, and you got it from someone with no stake in it.

Who has to move

This changes nothing unless the manager who is frustrated reads it, and frustration is the state in which the character explanation is most available and the structural one least. The cheapest first test is the three seat facts, written down, before the next performance conversation rather than after it. If the facts explain the behaviour, the conversation is a different one, and it is a great deal cheaper than a recruitment cycle.

Sources and notes

1. Donella H. Meadows, Thinking in Systems: A Primer, Chelsea Green. Bounded rationality, including the passage on being lifted into another position and the observation that retaining an outside perspective is distinctly unlikely, is chapter 4. The system traps, each stated with its way out, are chapter 5: the rule-beating exit quoted here, drift to low performance and its anchoring remedy, seeking the wrong goal, and the observation that systems have a terrible tendency to produce exactly and only what you ask them to produce. The twelve leverage points, in which information flows and rules sit above every physical parameter, are chapter 6. Note: Meadows states in the original leverage-points article that the list arose from personal experience and insight rather than rigorous empirical testing, and this piece uses it accordingly, as a search order rather than a measurement.
2. John D. Sterman, Business Dynamics: Systems Thinking and Modeling for a Complex World, McGraw-Hill. The fundamental modes of dynamic behaviour used in section 3, and the practice of drawing the reference mode before building any explanation, are chapters 4 and 3 respectively. The point that unanticipated effects signal a narrow model boundary rather than a property of the world, which underlies the caution about compensation repairs in section 2, is chapter 1.

A note on the risk this article carries. Everything above can be used to avoid a decision. “It is the seat” is a comfortable conclusion, available for any behaviour, and it postpones a conversation that is unpleasant. The substitution test exists to make the claim falsifiable, and it only works if you answer it before you know which answer you want.

Joshua Agonya Pi’Rwot, Founder.