How to Diagnose Culture Before You Spend on the Wrong Fix

You run a quarterly engagement survey. It says people do not feel heard. You and your boss sit down to make sense of it and the whiteboard fills up fast: no culture of listening, not enough channels to speak up, managers who need to listen better, a loop you never closed from last time, leadership that is not visible enough, a hybrid-workplace thing, low psychological safety. You pick the two or three that feel most plausible and build the obvious fixes: a listening forum, a suggestion box, a round of skip-levels, a "you said, we heard" email.

Next quarter, the survey says the same thing. People still do not feel heard.

This is one of the most common ways culture work fails, and it has nothing to do with effort. HR tried, leadership backed it, the fixes were reasonable. They failed because every item on that whiteboard was a restatement of the symptom, not a cause of it. "No culture of listening" is just "people do not feel heard" in other words. None of them reached the layer underneath, where the real question lives: what does speaking up actually cost a person here?

That failure teaches the one thing this piece rests on. A reasonable fix was built, backed, resourced, and the problem did not move. When that happens, the cause is not in the fix. It is in changing a set of assumptions or operating behaviour that the fix never reached. This is what "culture eats strategy for breakfast" actually means: not a poster, but a claim about which layer wins when they conflict. You strategise at the top; the bottom layer decides whether the strategy survives. Culture work is not a softer alternative to solving business problems. It is what you turn to when a business problem refuses to yield to business solutions, because the binding constraint has moved somewhere your strategy cannot reach.

In my last piece, I argued that changing culture is harder than diagnosing it. True. But diagnosing it honestly is harder than the industry admits, and that is what this piece is about: finding the real cause before you spend a budget on the wrong fix.

Where culture lives

Edgar Schein described culture as three layers, and the distance between them is the whole problem. The top layer is what you can see: how meetings run, what gets celebrated. The middle is what people say they believe: the values on the wall. The bottom holds the assumptions people take so completely for granted that they have stopped noticing them. Nobody states these. Nobody argues about them. They are simply how things are done here, and most people could not put them into words even if you asked.

A survey reaches the top two layers and stops. This is not a flaw in any particular survey. A survey asks people questions, so it captures only what people can tell you, and the deepest assumptions are exactly what they cannot tell you, because they do not know they hold them. You cannot ask someone to report a rule they do not know they are following.

Here is the cost - through an example. Take three companies - all score low on the same problem - people don't feel heard. The numbers nearly identical. The cause is different in each. In the first, nothing changes after people raise things, so effort feels futile. In the second, whoever raises an issue inherits the work of fixing it, so people stay quiet. In the third, anything said before the senior view has settled is treated as naive, so people settle things in private and perform agreement in public. Same score, three companies, three opposite fixes - correct intervention depends upon what's the real cause. Diagnosis is the work of finding which cause is real - something that a survey cannot do.

First, check whether you need this

The method below is expensive, and most problems do not need it. Most business problems have causes people can report, and those you diagnose cheaply. If lateral hires leave because pay is below market, exit interviews will tell you, because the leaver knows why and has no reason left to hide it. If a process is slow because of a process bottleneck, the people in it can point to it. Ask plainly; that is usually enough to find the fix.

You need the deep method only when the cause sits where people cannot report it, and there is a signature for that: the surface explanations are plausible, they contradict each other, and the obvious fixes have already failed. That last part carries one condition - the fixes have to have been reasonable and genuinely tried. A forum nobody attended because it was scheduled badly is a failed implementation, not a cultural cause. The real signature is a well-made fix that gets quietly absorbed and reverts anyway. When you see that, the constraint has dropped below the layer your fixes can reach. When you do not see that signature, fix it the cheap way. The fastest way to lose a numbers-trusting leadership is to invoke culture for a problem that had a spreadsheet answer.

The roadmap

To keep this concrete I will follow one business problem the whole way: good lateral hires keep leaving within two years. It shows up in attrition data long before any survey, and it is not obviously about culture, which is what makes the eventual diagnosis a finding rather than a restatement.

1. Start with the business problem, not "culture."

Refuse the brief "diagnose our culture." There is no useful answer to "what is our culture," and what would the finished definition tell you to do anyway? Name the thing the business is actually losing sleep over, in business terms: lateral hires gone inside two years, alignment that never holds, people who will not stretch past their role.

You can take more than one, as long as they form a small cluster and you can say in a sentence why they share a root: "attrition, weak alignment, and low initiative might all come down to how this place handles trust and risk." A grab-bag of unrelated problems is just "diagnose our culture" as a list. Hold the shared root as a hunch - if two of them share an assumption and the third does not, drop the third.

We have already cleared the gate for this one. The business raised offers, improved onboarding, assigned buddies, and capable outsiders kept leaving. The problem survived the solutions. That is why we are here.

2. Scope a survey to the problem.

Here is where most people reach for the wrong tool. A validated, off-the-shelf culture survey is built to read a whole culture, so it scores you on a fixed library of dimensions, almost none of them aimed at your problem. You get a broad portrait and then have to hunt through it for anything that touches attrition. That is the same mistake as trying to map the entire culture at once, just in survey form.

Scope the survey to the problem instead. Build it, or adapt the relevant parts of a validated instrument, around what plausibly drives a capable outsider out: whether people feel heard and whether action follows, access to where decisions are actually made, fairness of advancement, manager openness, role clarity, belonging by tenure and entry route. Broad enough to surprise you, narrow enough that every cut bears on the problem. The validated instruments earn their place as a source of tested items, not an off-the-shelf answer.

One honesty check: the narrower you scope, the more of your hypothesis you have built in, so keep a band of items wide enough to catch causes you did not predict. And remember the survey's job is not to find the answer. It is to show you where the gap is widest so the qualitative work knows where to dig. When the results come back, you read them as a map.

Suppose the scores on two of your items - "I feel heard" and "I see action after I give feedback" - hold up fine among long-tenured staff but collapse among people with under three years here, and collapse hardest of all for lateral hires. That pattern is not the diagnosis. It is the survey doing its real job: taking a vague business pain, lateral hires leaving, and pointing you at a specific group and a specific place to go look.

3. Go deep, through people who can still see.

Most diagnoses fail here, because they reach for a focus group and call it depth. A focus group gives you more of what people are willing to say out loud, only louder - the stated values, not the assumptions beneath them. To reach the bottom you need methods that get at what people cannot say directly. Five work, and they work best together.

The survey signal sets your aim. It told you the problem concentrates among short-tenure lateral hires, around feeling heard and seeing action after feedback. So that is who you talk to, and that is the theme you chase: not "what is the culture like," but "what happens here when someone newer tries to be heard." Every question below is pointed at that theme. Without the survey first, you would be wandering. With it, you know which thread to pull.

Talk to the people who just joined. A new hire can still see what everyone else has stopped noticing. In their first weeks, before they go native, what strikes them as strange is often the assumption veterans can no longer see. Ask, one at a time: what surprised you about how people speak in meetings? When did you first sense something could not be said directly? A lateral hire might say, "At my last place people debated in the meeting. Here, they align before the meeting and perform agreement inside it." That points somewhere specific: maybe the formal room is not where influence happens. One caution - a newcomer reads the place through their last company, so they spot that something is different but are poor at saying why. Collect what they noticed; you do the interpreting.

Use a repertory grid. This is the sharpest tool here and worth learning. It comes from the psychologist George Kelly. Instead of asking what someone values, which gets a rehearsed answer, you hand them three real things and ask them to compare. Name three leaders you have worked under here - which two are alike, and how is the third different? Someone says, "These two are mature; that one is emotional," or "These two know how to influence; that one just says things directly." You have learned something they would never have offered outright: the culture may not reward openness, it may reward a politically packaged voice, while directness gets coded as immaturity. Repeat with three projects, three decisions, three people seen as successful. The distinctions that keep recurring are the yardsticks the culture runs on, and because everyone answers the same prompts you can compare across people and even count.

Ask sideways. People guard the front door and leave the side door open. Do not ask "is this place political." Ask, "What would you tell a friend joining that you would never put in an email?" Or, "Finish this: the fastest way to get into trouble here is..." If several say "never surprise a leader in a meeting" or "raise it privately first," you have not found a communication gap. You have found a consequence: public voice before senior alignment carries real risk.

Collect the stories. Every company tells stories about itself - the founder's decisive call, the person quietly pushed out, the hire who "never understood how things work here." Listen for who becomes the hero and who becomes the cautionary tale. In one place the hero challenged the plan early and saved the project; in another the hero managed the stakeholder and got the idea through quietly; in a third the cautionary tale raised the right concern in the wrong room and got labelled difficult. These stories are how people teach each other the rules, and they tell you what the place actually rewards, whatever it claims.

Follow the paper trail. This is the one source that cannot perform for you. Take a real decision and trace it: who actually signed off, when input was invited, how many levels it climbed, which objections quietly vanished. Was the meeting where the decision was shaped, or where a pre-shaped decision was announced? Set that against what people told you. Leaders may say "we had an open discussion" while the record shows the option was chosen before the team was consulted. The gap between the account and the record is the finding.

One rule on the conversations: hold them one to one, or within a single level, never with bosses and reports in the same room. People will not tell you the truth about power while sitting next to it. And if you want to know whether dissent is safe, you will learn more from watching the junior group go quiet exactly where the senior group was relaxed than from anything either says aloud. The difference between the two rooms is the data.

4. Decide what is real.

Notice what the five methods produced. The newcomer said people align before the meeting and perform agreement in it. The grid said directness is coded as immaturity. The sideways questions said raise it privately first. The stories punished the one who spoke in the open. The paper trail showed input gathered only after the direction was set. Five methods, one assumption: here, you settle disagreement in private before you speak in public, and dissent before senior alignment reads as immaturity.

There are two tests you can do before you trust you've arrived at the right assumption. First, does it predict things you did not ask about? If the rule is "align in private first," then pre-meeting one-to-ones should be where decisions move, people with back-channel access should outweigh their role, the formal meeting should rarely reverse anything, and lateral hires should feel least heard. Work out what else should be true, then check. An assumption that keeps explaining behaviour you never asked about is real. Second, does it show up across methods that fail differently? Newcomers mislead one way, stories another, the paper trail a third. When all point the same way, you have something solid, precisely because they cannot all be wrong in the same way.

5. Show the company its own reflection.

Bring the groups together and show them what you found, built from their own material - their stories, their words. The point is recognition, not verdict. "You have a hierarchy problem" puts people on the defensive. "Here is what your own stories keep saying about where it is safe to disagree" lets them see it for themselves. This is the moment a company glimpses the water it swims in, which is where any honest change begins. It also checks your work: if you read it wrong, the room will tell you fast.

6. Go back to the business problem.

Close the loop, and this is the test that gives the method its teeth. Does the assumption explain what you started with? Here it does, and not obviously. Capable outsiders arrive expecting the formal room to be where they contribute. They find it is theatre, that real influence runs through a private pre-alignment they cannot access and did not know existed, and they conclude, correctly, that they cannot shape anything. So they leave inside two years and tell the exit interview they wanted a bigger role - sincere and useless. The assumption predicts the attrition. It could also have failed: an assumption about pay would not explain why outsiders leave while insiders stay, and you would know to look elsewhere. That is the loop doing real work.

Because the cause is named, the fix can finally touch it. Not a listening forum, which only adds another public room where the same rule applies. Make the pre-alignment visible so it stops being a hidden channel; make it safe to disagree in the room; or change what gets rewarded so the person who raises the early concern becomes the hero, not the cautionary tale. If the assumption does not explain the problem, dig further. You are finished when the problem is explained, not when you have catalogued the culture.

The assumption no method can catch

This method finds assumptions by finding gaps - where leaders and staff disagree, where the account and the record diverge, where one room goes quiet and another does not. But the most dangerous assumption is sometimes the one everyone shares completely, top to bottom, so it leaves no gap at all. No one argues about it. It quietly decides what the company can even see as a problem. You will not catch it by looking inward, because inside there is nothing to compare it against. You catch it only by looking outward, asking where the company's deepest shared belief is heading for a collision with what its market or its customers will soon demand. There is no reliable method for this, and usually only a crisis makes it visible - the same crisis that alone can move the deepest layer is often the only thing that lets an organisation see it. Sensing the collision before the crisis arrives is the hardest thing a diagnostician does.

That aside, the method stands. The objection to a survey plus five methods is that it is too much, and I feel it too. But you only reach for it on the rare problem that has already survived the cheap fixes, and the arithmetic is gentler than it looks: two weeks of survey, one week for the methods if planned in advance, one week of analysis. Four weeks buys far more than two weeks of survey alone. A survey hands you a picture of the surface and a comparison against strangers. This hands you the cause of a problem you already named, in language your own people recognise. One tells you something is wrong. The other tells you what to do starting next week.

Photo by Kindel Media: https://www.pexels.com/photo/a-scientist-doing-an-experiment-8325966/

References

1. Schein, E.H. & Schein, P. (2017). Organizational Culture and Leadership (5th ed.). Hoboken, NJ: Wiley. Key insight: Culture operates at three levels, and the deepest, basic underlying assumptions can only be reached through inquiry tied to a specific problem, not measured by survey. Relevance: The source of the three-level model and of the rule to anchor any diagnosis to a concrete problem. Read further: https://www.wiley.com/en-us/Organizational+Culture+and+Leadership%2C+5th+Edition-p-9781119212041

   
   

2. Jung, T., Scott, T., Davies, H.T.O., Bower, P., Whalley, D., McNally, R. & Mannion, R. (2009). "Instruments for Exploring Organizational Culture: A Review of the Literature." Public Administration Review, 69(6), 1087–1096. Key insight: Of seventy culture instruments reviewed, fewer than half showed adequate internal consistency and only about one in five could justify aggregating individual scores to an organisational total; the authors conclude there is no ideal instrument, only fitness for purpose. Relevance: The evidence that culture surveys measure the surface reliably and overreach when they claim to measure the culture, and the basis for scoping a survey to a problem rather than running a generic one. Read further: https://onlinelibrary.wiley.com/doi/10.1111/j.1540-6210.2009.02066.x

   
   

3. Cooke, R.A. & Szumal, J.L. (1993). "Measuring Normative Beliefs and Shared Behavioral Expectations in Organizations: The Reliability and Validity of the Organizational Culture Inventory." Psychological Reports, 72(3), 1299–1330. Key insight: The foundational validation of the OCI, built on behavioural norms rather than opinions. Relevance: Why the OCI is the strongest source of validated items when rigour matters most. Read further: https://journals.sagepub.com/doi/10.2466/pr0.1993.72.3c.1299

   
   

4. Cameron, K.S. & Quinn, R.E. (2011). Diagnosing and Changing Organizational Culture: Based on the Competing Values Framework (3rd ed.). San Francisco: Jossey-Bass. Key insight: The OCAI's four-quadrant model gives a fast, intuitive read of espoused culture that people grasp immediately. Relevance: The instrument to reach for when the brief genuinely is a whole-culture read, and a source of fast, intuitive items otherwise. Read further: https://www.wiley.com/en-us/Diagnosing+and+Changing+Organizational+Culture-p-9780470650264

   
   

5. Denison, D., Nieminen, L. & Kotrba, L. (2014). "Diagnosing Organizational Cultures: A Conceptual and Empirical Review of Culture Effectiveness Surveys." European Journal of Work and Organizational Psychology, 23(1), 145–161. Key insight: Reviews the major culture-effectiveness surveys and situates the Denison model's performance linkage and benchmarking within them. Relevance: Why Denison is the most persuasive board-facing option, with the caveat that it is proprietary. Read further: https://www.tandfonline.com/doi/abs/10.1080/1359432X.2012.713173

   
   

6. Van Maanen, J. & Schein, E.H. (1979). "Toward a Theory of Organizational Socialization." In B.M. Staw (Ed.), Research in Organizational Behavior, Vol. 1 (pp. 209–264). Greenwich, CT: JAI Press. Key insight: Newcomers learn an organisation's unwritten rules during early socialization, and those rules are most visible to them before they acclimatise. Relevance: The basis for using recent joiners as the diagnosis's clearest sensor of the taken-for-granted. Read further: https://www.semanticscholar.org/paper/eb8b4aa450019035f8c4d72c3c19126615121d65

   
   

7. Kelly, G.A. (1955). The Psychology of Personal Constructs. New York: Norton. Key insight: People organise experience through tacit, bipolar constructs that can be drawn out by comparison rather than direct questioning, using the repertory grid. Relevance: The source of the repertory grid, the method that turns invisible cultural yardsticks into structured, comparable data. Read further: https://www.routledge.com/9780415037976

   
   

8. Morgan, G. (1986). Images of Organization. Beverly Hills, CA: Sage. Key insight: The metaphors people hold for their organisation are tacit frames that shape how they interpret it, and can be surfaced projectively. Relevance: Grounds the sideways and metaphor questions used to reach assumptions people will not state directly. Read further: https://us.sagepub.com/en-us/nam/images-of-organization/book225545

   
   

9. Martin, J., Feldman, M.S., Hatch, M.J. & Sitkin, S.B. (1983). "The Uniqueness Paradox in Organizational Stories." Administrative Science Quarterly, 28(3), 438–453. Key insight: Organisations tell strikingly similar stories about themselves, and what a recurring story rewards or punishes encodes a shared assumption. Relevance: Establishes the stories a company tells as a readable, naturally occurring source of its assumptions. Read further: https://www.jstor.org/stable/2392252

   
   

10. Edmondson, A. (1999). "Psychological Safety and Learning Behavior in Work Teams." Administrative Science Quarterly, 44(2), 350–383. Key insight: People surface uncomfortable truths only where they feel safe to take interpersonal risk; absent that, they withhold. Relevance: Why the deep conversations are run one to one or within a single level, never in mixed rooms across power lines. Read further: https://journals.sagepub.com/doi/10.2307/2666999

    
    

11. Campbell, D.T. & Fiske, D.W. (1959). "Convergent and Discriminant Validation by the Multitrait-Multimethod Matrix." Psychological Bulletin, 56(2), 81–105. Key insight: Agreement between measures confirms a finding only when the methods are genuinely different; agreement among similar methods inflates false confidence. Relevance: Why a finding counts as real when newcomers, stories, and the paper trail converge, since each can fail in a different direction. Read further: https://psycnet.apa.org/record/1959-15239-001

    
    

12. Weick, K.E. (1995). Sensemaking in Organizations. Thousand Oaks, CA: Sage. Key insight: Organisations make sense of themselves collectively and after the fact, by putting experience into words they can recognise. Relevance: What the closing reflect-back session does, and why showing a company its own words is what lets it see its own water. Read further: https://us.sagepub.com/en-us/nam/sensemaking-in-organizations/book4988

    
    

13. Lincoln, Y.S. & Guba, E.G. (1985). Naturalistic Inquiry. Beverly Hills, CA: Sage. Key insight: Reflecting a finding back to the people studied, member checking, is among the strongest ways to confirm it is credible. Relevance: Why the reflect-back session doubles as a check on whether the surfaced assumption is real or projected. Read further: https://us.sagepub.com/en-us/nam/naturalistic-inquiry/book842