24 August 2026 · 12 min read

Culture as a Product: the iceberg, the weather, and what you can actually steer

Most culture work is aimed at the part you can see. The all-hands, the values on the wall, the behaviour in the review. That part is real, and it is roughly a tenth of the thing.

Three Mondays

Amara runs operations, and a staff survey has just told her that people do not feel able to challenge a decision. She does what a decent leader does. At the Monday meeting she says that from now on anybody can question anything, and she means it.

On the first Monday, Theo does. He asks why a supplier was chosen without talking to the team that uses it. Amara thanks him. On the way out a colleague says "brave", and smiles a little too long.

On the second Monday somebody asks a smaller question.

On the third, the supplier job slips, and a senior voice says "well, Theo did have concerns", as though he had caused it. Nobody asks anything after that. By the end of the month the Monday meeting sounds the way it always did, and the action on the survey is marked as done.

Amara changed the instruction. She did not change what the room believed a question costs.

What is the iceberg model of culture?

An iceberg is the right picture because it gets the proportions right and the mechanism right at the same time. Above the waterline sit the behaviours everybody can observe: how a decision gets made, who talks in the meeting, what happens when somebody is wrong. Below it sit the values, the assumptions and the beliefs that produce those behaviours. You can order the visible half to change. It changes back on Monday, because the submerged half never moved.

People act on what they expect to happen next, and the submerged half is where that expectation lives. An order moves behaviour for as long as somebody is watching. The belief under it moves only when the new behaviour keeps being safe, over and over, on days when nobody is.

So the surface matters after all, just not in the way the all-hands assumes. The arrow in the model runs both ways: values shape what people do, and behaviour practised often enough reshapes the values underneath. Where the picture comes from, and what we changed in it, is on the culture model page.

The weather is the part most models leave out

An iceberg is not a static diagram. It sits in weather.

On one side, a storm: a restructure, a missed quarter, a leader who says one thing in the town hall and another in the corridor. Under those conditions people's behaviour drifts away from the values they actually hold, and it drifts fast. On the other side, sun: a launch that worked, a public credit given to the person who earned it, a hard call taken well. Behaviour and values line up on their own, and nobody had to run a programme.

The practical consequence is uncomfortable: a company's culture is not a thing it has, it is a thing it is currently doing, and the weather changes what it is doing this week. Which is why an annual measurement of culture measures the weather on one particular day and calls it the climate.

How to tell weather from climate

Getting this wrong is expensive both ways: a company treats a bad month as a culture problem and rewrites its values, or treats a culture problem as a bad month and waits. Three questions sort most cases.

• Does it have a start date? Weather usually does: the morning of the restructure, the week the leader left. A habit nobody can date has been there a while.
• Did it happen in your best quarter too? If the sniping survived the launch that worked, the storm did not cause it.
• Is it everywhere, or in one team? A storm hits the whole company. One team behaving the same way in good weather and bad is telling you about that team.

If your economic slowdown has made people quiet, that is weather. If they were quiet in the good year, it is the iceberg, and the slowdown has only stopped hiding it.

What "culture as a product" actually commits you to

Treating culture as a product is not a metaphor about caring more. It is a commitment to three unglamorous things: it gets a deliberate design, it gets shipped on a cadence, and it gets a signal you read.

A deliberate design

A list of words is not a design. Across 562 companies that published their values, there was no correlation between the values they emphasised and how their employees rated the culture (Sull, Turconi and Sull, MIT Sloan Management Review, 21 July 2020). A design answers what the list leaves open: which few things matter this year, what each one looks like in a real meeting, and how an ordinary person meets it on an ordinary day. A product has somebody deciding what is in and what is out. Most value statements have a committee that agreed to keep everything.

Shipped on a cadence

The cadence is the part almost everybody gets wrong. Culture is not built by an offsite any more than fitness is built by one very good weekend. The evidence here is blunt: across 42 samples in 37 studies of workplace resilience training (Vanhove and colleagues, 2016), the effect decayed from d = 0.26 immediately after training to 0.07 at follow-up. The training worked. Then Tuesday happened.

The difference between a project and a product is the day after launch. A project has an end date, which is what makes it a project: the team disbands and the binder goes on the shelf. A product has a next release. Culture behaves like the second and gets funded like the first, and you can see the result. In a WorkNest survey of 1,017 HR professionals and business leaders, 45% experienced an initial improvement before the problematic behaviour returned (WorkNest, via HRreview, 21 August 2026). Why two-minute doses hold where a big event does not is an argument of its own.

A signal you read

No product team ships for a year and then asks its customers once how they felt about it. It reads what people do. Culture has that too: who takes part, whether that is rising or falling, whether what arrived was any use.

The usual proxy is thinner than it looks. The average internal email is opened by 76% of its recipients and clicked by 9% (ContactMonkey, Internal Email Benchmark Report 2026, 19 June 2026, updated 1 October 2026). That is a vendor's benchmark, and the gap is the point. An open is a delivery receipt. Whatever you are reading to find out whether your values reached anybody, it needs to be something a person chose to do.

What MeaningFuel actually does with the model

Two doses, and they are deliberately different medicines.

The multivitamin. One insight, one quote behind it, four faces, one question - who said this - by email, Slack or Teams. Thirty seconds. It goes to everybody, every working day, whether or not anything is wrong, because the point is to hold the posture rather than to fix a fault. The quote is the prompt; the insight underneath it is what a reader is actually handed, and it is written per theme, because the same line lands differently under Resilience than under Trust.

The company chooses the values and skills it runs on. For Amara's meeting those might be Courage and Trust, with Feedback among the skills, so that over an ordinary week her people meet a few short lines on what it takes to ask the awkward question, and on what a good answer to one sounds like. This is the waterline, where the submerged half and the visible half meet.

The painkiller. A module is booked onto a team and a problem when a named issue emerges: a short course of daily lessons, ten over two working weeks, with two ten-minute debriefs its lead runs. One team is sniping and another is fried; they should not be getting the same module, and in most products they do, because most products only sell one thing.

Take teams that have drifted apart under one logo, where "quality" means double-checking in one room and polish in the next. One Shared Culture is the module for that. Booked for a team, it invites the people on it, and those who join get one short lesson each working day. In one lesson the team takes a value word it uses and defines it by behaviour, because a definition that rules nothing out defines nothing. In another, each person picks the three values they actually use to decide things, and at the debrief the picks go side by side. The overlap is usually smaller than the official list, and truer.

That second dose carries a guardrail that travels with the metaphor: a painkiller is prescribed to a team and a theme, never to a person. Culture is a property of a group. Targeting an individual would not just be intrusive, it would be worse medicine.

What we don't do, on purpose: four things the model rules out

A model is only worth something if it rules things out. This one rules out four:

1. It does not run a survey on the people receiving it. They are asked one short question at most - every 14, 30 or 90 days, or never, as the employer sets it - and can skip it in one tap. What comes back is pooled, with nobody named. The quarterly pulse goes to owners and team leads only.
2. There is no individual score. No leaderboard, no per-person streak a manager can pull up. The team is the unit, always.
3. No manager reads anybody's answers. The signal is aggregate or it does not exist.
4. There is no rollout. No facilitators, no certification, no binder. Two minutes and zero admin, or the cadence dies in week three like the last one did.

What gets read instead of a survey

Engagement, not correctness. Whether somebody knew that Amy Edmondson said it is close to worthless as a signal; whether they opened it, answered it, and marked it as landing is the thing that tracks whether the ritual is alive.

So that is what Insights shows, team by team and never person by person: engagement, how many people are taking part, and whether it is landing well. Landing well is the only one of the three that asks whether the material was any good; the other two count volume. A team that opened everything and valued none of it should not look healthy, and under an older single score, for a while, it did.

The honest limit is that all of this reads the waterline. It tells you whether the habit is alive in a team and how the insights land. It does not tell you what that team believes, or why it went quiet, and a healthy-looking chart can sit on top of an assumption nobody has tested.

Nothing here would have stopped that "brave" in the corridor. That takes a leader who noticed it, and who says something on the fourth Monday. The habit keeps the question of what a value means in front of Theo's colleagues on the days nobody is asking it. It promises no change in the iceberg, and it does not do the leader's half.

Questions managers ask about the culture model

Can you change a culture by changing behaviour?

Yes, if the behaviour is repeated and it stays safe. An instruction moves behaviour for as long as somebody is watching. A practice done often enough, where nothing bad follows, changes what people expect, and what people expect is most of what a culture is. One session does not do that, however good the session was.

How often should you measure organisational culture?

Often enough to tell weather from climate, which means continuously rather than once a year. A yearly survey reads one day. What is worth reading every week is behaviour that is already happening, such as who takes part, rather than a fresh form each time. A form still has its uses, as an occasional question rather than the main instrument.

Who should own culture as a product?

One named person, senior enough to say what stops and close enough to read the signal every week. The job is a product manager's: decide the design, keep it shipping and act on what comes back. A committee can advise on it. It cannot own it, because nobody on a committee has to explain why the third Monday went quiet.

Is the iceberg model of culture still useful?

Yes, as a way of deciding where to act rather than as a diagram for a slide. Its one big idea is that what you can see is not what is steering. It is a simplification, and the plain version has two gaps: it leaves out the weather, and it draws the arrow only one way. Add those two and it starts telling you what to do on a Tuesday.

One thing this library will tell you about itself

Most of the lines in it are attributed to a man.

That is not the shape of where good ideas came from. It is the shape of whose byline history kept. We say so here rather than waiting to be asked, we withhold speakers whose record does not survive scrutiny, and when an attribution turns out to be wrong - the "survival of the fittest" line Darwin never wrote, imported more than once - it gets corrected rather than defended. Credit in proportion to what earned it is a culture value we sell. It would be an odd one to break in the content.

The iceberg is the model. The weather is why a once-a-year read of it is useless. Two minutes a day is what we think the design costs.

• Culture as a Product
• Organisational Culture
• Workplace Culture
• Culture Model
• Iceberg Model