Every leader has launched a "culture initiative", a new values statement, an offsite, a poster campaign, and watched the organisation absorb it without changing at all. The reason is almost always the same: they tried to change the thing they could see (the slogans) instead of the thing that runs the place (what gets rewarded when nobody is watching). To shift culture you first have to measure it honestly, and most measurement stops at the surface.

The quick version

  • Culture is what's rewarded and tolerated, the behaviour that actually pays off here, not the values you've written down. The gap between the two is the whole problem.
  • Measure at three depths. Edgar Schein's model separates what you see (artefacts), what you say (espoused values) and what you assume (the unspoken rules). Surveys mostly catch the first two; the third is where behaviour really comes from.
  • Use real instruments, not vibes. The Competing Values Framework / OCAI maps your culture onto four types and lets you compare "now" with "preferred"; the Organizational Culture Profile (OCP) measures fit between a person's values and the organisation's.
  • Shift it through a few levers, repeated. Who gets promoted and fired, what leaders pay attention to, who you hire, these move culture. A poster does not.

The idea in depth

Start with a definition you can act on. The most durable comes from Edgar Schein, whose Organizational Culture and Leadership (5th edition, Wiley, 2017, with Peter Schein) describes culture as operating at three levels that differ in how visible they are. Artefacts are everything you can observe, the office layout, the dress code, the rituals, the stories people tell. Espoused values are what the organisation says it believes: the values statement, the strategy deck, the all-hands rhetoric. And underneath both sit the basic underlying assumptions, the things the group has come to take for granted because they once solved a problem, and which now govern behaviour invisibly. Schein's point is that the top two layers often sit on a foundation that contradicts them: a company can espouse "we welcome dissent" while its assumptions quietly punish anyone who challenges the boss.

flowchart TD
  A(["Artefacts
what you SEE, offices, rituals, stories"]) --> B(["Espoused values
what you SAY, values page, strategy deck"])
  B --> C(["Basic assumptions
what you ASSUME, the unspoken rules that
actually drive behaviour"])
  C -.->|"the gap that kills initiatives"| B
Schein's three levels. Most culture programmes change the top two and leave the bottom one untouched, so behaviour doesn't move. Leaders Loop

So stop auditing your slogans and start auditing your assumptions. The fastest way in is to look at outcomes, not statements: who got promoted last cycle and why, what behaviour got someone a difficult conversation, which projects got resourced and which quietly died. Those decisions reveal the assumptions far more reliably than any survey question. As the practitioner shorthand puts it, popularised by engineering leader Jocelyn Goldfein and others, culture is the behaviour you reward and punish. Public signals (who gets the plum project, who gets eased out) teach the organisation what is really valued, regardless of what the values page claims.

How to actually measure it

"Audit your assumptions" sounds soft, so anchor it in instruments that have survived peer review. Two are worth knowing. The first is the Competing Values Framework and its survey, the Organizational Culture Assessment Instrument (OCAI), developed by Kim Cameron and Robert Quinn and set out in Diagnosing and Changing Organizational Culture (3rd edition, Jossey-Bass/Wiley, 2011). It maps culture on two tensions, flexibility versus control, and internal versus external focus, yielding four types: Clan (collaborative, family-like), Adhocracy (entrepreneurial, innovative), Market (competitive, results-driven) and Hierarchy (structured, controlled). The OCAI's clever bit is that respondents rate the culture now and the culture they'd prefer, so the output isn't a label, it's a gap. The distance between the two profiles is your change agenda, made visible.

quadrantChart
  title Competing Values Framework (Cameron & Quinn)
  x-axis "Internal focus" --> "External focus"
  y-axis "Stability / control" --> "Flexibility / discretion"
  quadrant-1 "Adhocracy: innovate"
  quadrant-2 "Clan: collaborate"
  quadrant-3 "Hierarchy: control"
  quadrant-4 "Market: compete"
The four culture types of the Competing Values Framework. The OCAI plots where you are now versus where people want to be. Leaders Loop

The practical advice is narrow: run the "now versus preferred" version, not just a snapshot, and run it by team. A single blended profile hides the truth that engineering may be living in a Hierarchy while sales is pure Market, and that the friction between them is a culture problem masquerading as a process one.

The second instrument is the Organizational Culture Profile (OCP), from Charles O'Reilly, Jennifer Chatman and David Caldwell's paper "People and Organizational Culture" (Academy of Management Journal, 1991, vol. 34, pp. 487–516). It works by getting people to sort value statements by how characteristic they are of the organisation, then comparing that profile to an individual's own preferences. Its real contribution is person–organisation fit: in their data, the match between a person's values and the organisation's predicted job satisfaction and commitment a year later, and turnover two years after that. The implication is worth sitting with, culture is something you hire and promote for, not just describe. Fit, measured this way, is a leading indicator of who stays.

A culture survey that only asks "how are we doing?" measures mood. One that asks "where are we, and where do we want to be?" measures the change you actually have to make.

An honest limitation. These instruments measure the visible and the espoused far better than Schein's deep assumptions, which, by definition, people can't easily report because they don't know they hold them. A survey can tell you the organisation says it values candour; only watching what happens to the next person who challenges the CEO tells you whether that's true. Treat the surveys as a map, then verify against behaviour. And beware the fit trap: hiring purely for cultural fit can quietly become a machine for sameness, strong cultures need complementary additions, not clones, or they lose the ability to adapt.

The levers that actually shift it

Measurement tells you where you are; it doesn't move anything. What moves culture is a small set of levers, applied consistently over a long time. Schein's research keeps pointing at the same ones: what leaders systematically pay attention to and measure; how they react under stress; what they deliberately model; and the criteria they use for rewards, promotions, hiring and dismissal. Jennifer Chatman's work (with Glenn Carroll) in Making Organizational Culture Great sharpens the bar: a culture that drives results has to be strategically relevant, strong (widely shared and intensely held), and able to adapt, and leaders build that through selection, socialisation and rewards, not perks.

In practice that means picking the one or two behaviours your "now versus preferred" gap demands, and changing the signals around them. If you want more candour, the lever isn't a poster, it's publicly thanking the person who told you bad news, and making sure the next promotion goes to someone who raised a hard truth rather than the one who stayed comfortable. Culture is the accumulation of those signals.

A worked example

Take a 200-person software company, call it Meridian, whose values page says "speed and ownership," but whose engineers privately joke that nothing ships without three layers of sign-off. (Illustrative figures and details throughout; this is a teaching example, not a real company.)

The new VP of Engineering runs the OCAI across her teams. The "now" profile comes back heavily Hierarchy, control, process, approvals, while the "preferred" profile leans Adhocracy and Clan: people want to move fast and decide locally. That's the gap, and it explains the joke. The espoused value (ownership) and the lived assumption (don't act without cover) are at war.

flowchart LR
  A(["Espoused: 'speed & ownership'"]) --> B{"OCAI: now vs preferred"}
  B -->|"NOW: heavy Hierarchy
(3 sign-offs to ship)"| C(["The gap is real"])
  B -->|"PREFERRED: Adhocracy + Clan
(decide locally, move fast)"| C
  C --> D(["Pull ONE lever:
kill two approval gates,
promote the engineer who
shipped & owned the outcome"])
  D --> E(["Behaviour shifts
because the reward did"])
Meridian's culture problem, diagnosed and then moved by changing what gets rewarded, not what gets posted. Leaders Loop

The temptation is to relaunch the values with more conviction. Instead she pulls a behavioural lever: she removes two of the three approval gates, then, the part that counts, promotes the engineer who shipped a risky-but-right call and owned the outcome, and says clearly, in front of everyone, why. The next quarter, an OCAI re-run shows the "now" profile drifting toward the "preferred", not because the words changed, but because the organisation watched what got rewarded and quietly updated its assumptions. That's culture moving: slowly, behaviourally, and measurably.

Frequently asked questions

What's the difference between values and culture?

Values are what you say you believe; culture is what you actually do when those values cost something. In Schein's terms, a values statement is an espoused value, but the culture lives in the basic assumptions underneath. When the two conflict, "we value work-life balance" versus a norm of answering emails at midnight, the assumption wins every time, and people learn to ignore the statement. Closing that gap is the work of defining and embedding values.

Can you really measure something as fuzzy as culture?

You can measure it more rigorously than most leaders assume, the OCAI and the OCP are peer-reviewed instruments with decades of use, but no survey captures the whole thing. The reliable approach is triangulation: a validated survey for the espoused layer, plus behavioural evidence (promotions, exits, what gets resourced) for the assumptions the survey can't reach. Numbers plus behaviour beats either alone.

How long does it take to shift a culture?

Longer than any quarterly plan and shorter than "never," but think in years, not weeks, for a real shift. Culture changes at the speed of accumulated signals, each promotion, each tolerated behaviour, each crisis response either reinforces or erodes it. The visible turning point usually comes when people see that the new behaviour reliably pays off and the old one no longer does; until then, they're sensibly waiting to find out if you mean it.

Isn't hiring for "culture fit" just code for hiring people like us?

It can be, and that's the danger. The O'Reilly–Chatman research links value fit to retention and commitment, but strong cultures also risk becoming monocultures that can't adapt. The fix is to hire for shared values (how we treat each other, how we make decisions) while deliberately seeking diversity of background, perspective and skill, "culture add," not "culture match." This is the seam where culture work meets diversity, equity & inclusion.

Our culture varies wildly by team. Is that a failure?

Not necessarily, subcultures are normal and often healthy; a sales floor and a research lab should feel different. It becomes a problem when subcultures hold contradictory assumptions about something core (how decisions get made, whether it's safe to disagree), because that's where collaboration breaks down. Measure by team, expect variation, and intervene only where the differences sabotage shared goals, not to enforce uniformity for its own sake.

Related in the Toolkit

Measuring culture is the diagnostic step; once you can see the gap, the work branches into the formation of culture itself (how culture forms & persists) and the leadership behaviour that signals it (leadership styles & models).

Where to go next