What is the difference between culture fit and culture add?

Because in practice it became a vote on familiarity. Interviewers asked themselves whether they would enjoy a drink with the candidate, and that question rewards shared background, shared humour and shared communication style. The result is a team that gets progressively more similar and progressively worse at spotting its own blind spots, while the assessment itself is almost impossible to defend when challenged. Fit is not a meaningless idea. A person who needs deep autonomy will struggle in a heavily supervised environment, and that is a real compatibility question. The failure is using a vague label to license an intuition nobody has to justify with evidence. So keep the compatibility question and discard the label, because the label is what carried the bias.

Two things kept separate. First, whether the person can operate effectively given how your company genuinely works: its decision-making speed, its tolerance for ambiguity, how directly people give feedback, how much structure exists. That is a behavioural question with evidence behind it, and candidates should hear the honest version so they can assess it too. Second, what perspective, experience or skill they bring that the team currently lacks. Add is not diversity of demographics used as a proxy. It is a specific gap you can name before the interview: nobody here has worked at scale, nobody has run a regulated process, nobody challenges the founder. Naming it in advance is what stops add becoming as vague as fit was.

Write down three or four operating values in behavioural terms, then assess each with the same structured question for every candidate. If one value is direct feedback, ask for a time they disagreed with a manager and what they said. If it is ownership, ask about something that went wrong that nobody had assigned to them. Score against anchors held in your hiring software describing what strong and weak evidence looks like. Ban the summary judgement entirely: no scorecard should accept the words good fit without evidence attached. This is standard structured interviewing applied to behaviour rather than skill, and it produces comparable answers instead of impressions. Then have two interviewers score the same values evidence independently, which surfaces disagreement about the standard.

Leadership sets it, but you should assess against what the company genuinely does rather than what the values page claims. The gap between the two is where new hires get disappointed and leave within a year. Be honest with candidates about the real operating environment, including the difficult parts: long decision chains, heavy travel, a founder who changes direction, limited process. Some people want exactly that. Others should self-select out, and that outcome is a success. If interviewers cannot agree on what the culture is, you have a definition problem to solve before you can assess anyone against it, and no interview technique compensates for that. Write the operating reality into the kickoff notes so every interviewer describes the same company.

Organisational culture is the set of behaviours a group actually rewards, tolerates and punishes, which is not the same as the values written on a wall. It shows up in four observable places. Decision rights: who can commit the organisation to something and who has to ask. Information flow: what is shared by default and what has to be requested. Recognition: which behaviours get promoted, given the visible work, or quietly overlooked. And conflict: what happens when two people disagree in a meeting. Stated values matter only where they predict those four. Where they do not, candidates learn the real culture in their first fortnight, and the gap between what they were told and what they found is a familiar reason early exits happen.

The familiar typologies — the four-quadrant models that sort organisations into labels such as clan, market, hierarchy or adhocracy — are descriptive frameworks built for organisational analysis, not selection instruments. They are useful for a leadership conversation about what the organisation is becoming. They are close to useless inside an interview, because a label does not tell an interviewer what to ask or how to score the answer, and two companies given the same label will reward visibly different behaviour. What works in a loop is the level below the label: three or four written behavioural definitions specific enough that two interviewers who have never met would rate the same answer the same way. A definition that cannot survive that test is a value statement rather than a criterion.