Role fit: what the role needs, not who impresses

Fit is not a quality a person has — it is the distance between a candidate and a role you wrote down first.

The idea

Most bad hiring arguments are arguments about a role nobody wrote down. One person is comparing the candidate to the best interview they saw this month; another is comparing them to the person who just left. Neither comparison is the job.

So write the role: the two or three things this person must deliver in year one, the three requirements that are genuinely must-have, and what the team can already do without them. Everything else is a preference or something learnable in a quarter.

Then the becomes one sentence you can say out loud: here is the requirement I am trading away, and here is why.

Build the scorecard, then test it

Eight requirements came out of a twenty-minute intake call. Sort them. The must-have column is capped at three, so something has to be cut — and four candidates, whose evidence never changes, are re-judged every time you move a line.

Req: senior analyst, billing and payments.

Year one, two outcomes: design and run the pricing experiment programme; own a billing metric layer finance will sign off on.

The team today: four strong SQL modellers, solid BI tooling, deep payments knowledge, plenty of mentoring. Nobody who has designed a clean experiment, and nobody who can hold a room with finance.

intake wish list

everything the hiring manager said out loud. Not a role profile.

must-have

0 / 3

a miss here ends the loop

preference

scored, never disqualifying

teachable

learnable in a quarter, here

Nothing is sorted yet. Ten things from an intake call is a wish list, not a role — press step to watch it become a card, or move any requirement yourself.

Marks read left to right in card order: filled = strong evidence, ring = some, dashed = none. A warm ring marks the single requirement that decided the verdict.

How it works

Write two or three year-one outcomes first. If you cannot name what this person will have delivered twelve months in, the req is not ready and no candidate can be judged against it.
Take the intake list, then cut must-haves to three. Ten must-haves means every candidate fails something, so the trades get made privately in the debrief instead of publicly on the card. Drafting the cut is the interviewer’s job; confirming or swapping it is the hiring manager’s.
Sort the remainder into preference and teachable. Teachable means learnable in a quarter with the support this team actually has — not “smart people figure it out.”
Write down what the team is already strong at. A hire fills a gap in a team; it does not maximise an individual score.
Attach evidence to each must-have before the pipeline opens. If you cannot name the question that would fail it, it is not a must-have.
Run every candidate against the card, not against each other. Same evidence, same columns, verdict recomputed.
Say the trade out loud. “We are trading X for Y” is the only sentence that survives a week.

card (billing analytics)   must-have  influence without authority
                                      experiment design
                                      stays past two years
                           preference advanced SQL · BI tooling · mentors juniors
                           teachable  payments domain · python

candidate: Priya   (evidence fixed — nothing about her changes below)
  must  influence            strong   pass
  must  experiment design    strong   pass
  must  stays past two yrs   none     FAIL   3 roles in 6 yrs; names management as next step
  preference score           6 / 6    (strong = 2, some = 1, none = 0)
  verdict  no-hire — decided by ‘stays past two years’

move that one line from must-have to preference:
  must  influence            strong   pass
  must  experiment design    strong   pass
  preference score           6 / 8    (stays: none = 0 of 2)
  verdict  hire — decided by ‘experiment design’, the gap the team has

Same person, same interview, opposite outcome. That is not a flaw in the method — it is the method showing you exactly which requirement is carrying the decision, so you can defend it or drop it.

When to use it

Situation	What the card does — and its limit
A vague opening (“we need another analyst”)	Forces year-one outcomes before sourcing. Limit: if the manager genuinely doesn’t know the outcomes, the card can’t invent them — that’s a signal to pause the req.
Debrief between two interviewers	Converts “I liked them more” into a named requirement. Limit: it only settles disagreements about the role, not about whether the evidence was real.
Overqualified, switcher, or multi-move profiles	Turns a vibe into a testable question. Limit: predictions about tenure stay predictions, however confidently you write them.
Referral pressure from someone senior	Keeps sourcing channel and evidence in separate boxes. Limit: it costs you a slightly awkward conversation, every time.
Backfilling someone who left	Re-scopes to what the team now needs. Limit: writing the card after meeting a favourite candidate just launders a preference — write it before the pipeline opens.

Watch out for

Treating the intake list as the profile. Ten must-haves from one call is a wish list. Cap it at three, propose the cut in writing, and let the hiring manager swap rather than add. If they add a fourth, ask which of the three it replaces.
A must-have you cannot test. “Stays past two years” is a prediction, not a requirement — nobody can evidence the future in a 45-minute loop. The testable version is a pattern: did they leave before the hard part, repeatedly? Three moves each with a landed outcome is not a hopper; two moves that both ended right before the migration shipped is a pattern worth naming.
Letting a referral count as evidence. A referral is a sourcing channel. When the panel comes back “solid, nothing standout,” that is the evidence, and average evidence has to fail a must-have — otherwise the bar quietly becomes “who knows us.”
Hiring the highest individual score. The strongest analyst in the pool can still be the fifth person who is strong at the same thing. Ask what the team cannot do on Monday, and whether this hire changes that.
Scoping the role so narrowly that only one person fits. If no realistic candidate clears the card, the problem is the req, not the market — split the role, move a must-have to teachable, or admit the card was written around someone you already met. And when ambition points elsewhere (another team, management), the honest question is not “will they leave” but “does this role deliver what they want inside the time they’ll stay?”

Worked example

Debrief for the billing analyst req. Priya is the strongest interviewer anyone has seen this quarter: clean experiment design, a story about turning a finance director around, deep payments fluency. Tomás is the best SQL modeller in the pool by a distance. Marcus comes from healthcare logistics and has never seen a chargeback. Dana was referred by a staff engineer and every panel said the same thing: solid, nothing standout.

Against the card, three of those four resolve fast. Dana misses influence without authority on a “one example, no depth” answer — the referral told us where to look, not what she can do. Tomás clears every must-have except experiment design, and everything he’s strong at, the team already has four of; hiring him is buying a fifth SQL modeller for a team that needs an experimentalist. Marcus misses payments entirely, but payments sits in teachable — a quarter of reading, shadowing and one embarrassing week — and he covers both gaps outright.

Priya is the real decision. She fails only because someone promoted stays past two years to a must-have after reading three roles in six years and “I want to be managing again within eighteen months.” The defensible move is to name it: we are trading the strongest analyst in the pool for confidence about tenure. Then test the trade instead of assuming it — what does year two look like for her here, is there a lead-analyst path in eighteen months, and would twenty months of her beat three years of Marcus for these two outcomes? If the answer is yes, take “stays past two years” off the must column and say so; if it stays, you have chosen, in writing, and Marcus is your hire. Either way nobody leaves the room saying “we just liked the other one more.”