A leader who wants to know what their culture actually is faces a specific obstacle: the organization behaves differently in their presence, and every direct question returns an answer shaped by who asked it.
The way around this is to stop asking people how things feel and start looking at what the organization has already done.
Why surveys underperform
Three reasons, and none is about the instrument being poorly designed.
They measure sentiment, which is climate rather than culture, and the two can move in opposite directions.
They aggregate, which averages away the concentration that is usually the finding. An organization with one bad unit and eleven good ones produces a satisfactory overall number.
And people answer strategically, in proportion to how much they trust the anonymity, which is itself a cultural variable. The organizations where the survey is least reliable are precisely the ones where it would be most useful.
Evidence the organization already holds
Most of what a leader needs is in records kept for other purposes, and reading them culturally costs an afternoon.
Turnover by supervisor. Not by department, not organization-wide. The distribution is almost never even and the concentration is the finding.
Who left voluntarily in the last two years, and how good they were. If the strongest people are the ones leaving, the culture is selecting against capability, whatever the survey says.
Internal transfer requests. People voting with their feet inside the organization, and a signal that is usually visible only to whoever processes the paperwork.
The last five promotions, and what actually distinguished each person. This is the operative statement of what the organization values, and it can be reconstructed from the record.
The gap between when a problem occurred and when leadership learned of it. Retrievable from a handful of past incidents, and it measures the thing that matters most.
Exceptions granted in the last year. Counted, against each standard that is supposed to be firm.
Instances beat impressions
Since culture appears under pressure, the evidence is a set of specific occasions rather than a general reading.
Collect four or five and examine them properly. What happened the last time a deadline could only be met by cutting something? What happened the last time somebody senior was wrong and somebody junior knew? What happened the last time a good customer asked for something the organization does not do? What happened the last time somebody raised a concern about a colleague?
These are retrievable, because people remember them exactly. Those are the occasions on which they learned how the place works, which is precisely why they were memorable.
The material has to come from people other than the leader’s direct reports, for reasons covered in the Self-Management subchapter.
The exit interview is the wrong instrument
Conducted on the way out, by the organization, with a reference still notionally at stake, the exit interview reliably produces a neutral account.
People give the safe reason. A better opportunity, a shorter commute, family circumstances. All plausible, frequently partly true, and rarely the operative cause.
The version that works is a conversation six months later, initiated by somebody without authority over anything that affects them, framed as asking for help rather than collecting data. By then the reference is secured, the new job is settled, and there is no cost to being accurate.
Very few organizations do this. The ones that do generally find the six-month account substantially different from the exit one.
Watch what happens when you are absent
The most direct test available, and it requires only that a leader genuinely leave.
Take two weeks, unreachable, on an ordinary stretch rather than a quiet one. What holds and what does not is the answer to the question this whole subchapter is about, and it is not obtainable while you are in the building.
What holds is encoded. What lapses was being sustained by your presence, and is therefore a description of exactly how much of the culture exists independently of one person.
Expect the finding to be uncomfortable
A leader who does this honestly will discover things about their organization they did not want to know, and one of them will be about their own conduct.
That is the point of the exercise rather than a hazard of it. An assessment that returns a favourable picture of everything has almost certainly been conducted through channels that were shaped by who was asking.
The useful check on any culture assessment: did it produce at least one thing you did not already believe? If not, the instrument measured your own assumptions and reported them back.
Edited by Patrick J. Wolf, PhD