Company Values: How They Get Written, and Why Most Change Nothing

TL;DR

• The core decision: whether you're writing values to describe the company or to decide things with, because the two produce completely different documents.
• When doing nothing is right: when people already make consistent decisions without consulting anything, and could each tell you the same three things this place cares about.
• What has to be true: a reasonable company could hold the opposite of each value you write. If not, you've written a quality rather than a value.
• How the options split: by who writes them and what they're derived from, which determines whether anybody recognises the result.
• Decision rule: if nobody can name a decision that would have gone differently, the value isn't doing any work.
• Outcome to expect: fewer values, less flattering ones, and at least one argument you've been avoiding.

The Poster Nobody Reads

Somebody gets handed the values project. It arrives with good intentions and a reasonable brief: the company has grown, people are making decisions differently in different teams, and a shared statement of what matters would help. So a process starts. There's a workshop, a shortlist, some wordsmithing, a launch with printed cards.

Six months later, ask anyone to name the values. You'll get two out of five, usually the two that were already true before anybody wrote them down. The cards are in a drawer. The poster is behind the plant. Nothing about how decisions get made is different, and the person who ran the project has quietly concluded that values are a soft thing that doesn't really work.

They're half right. Most values don't work, and the reason isn't the wording or the rollout. It's that the words chosen were ones nobody could disagree with. Integrity. Excellence. Customer focus. Teamwork. No company on earth has ever announced that it stands against integrity, which means the statement carries no information and can't be used to decide anything. It isn't a value, it's a description of being a reasonable organisation.

Recommended Tools

Best tools for Leadership & Culture

C Culture Amp Culture measurement L Lattice Leadership development L Leapsome People enablement platform

View all recommended tools

The real test is simple and uncomfortable. A value is only a value if a reasonable competitor could genuinely hold the opposite. Everything else is a quality, and qualities on a wall do nothing except advertise the gap between what you say and what happens on a bad Tuesday.

When You Genuinely Do Not Need to Act Yet

Your current setup is genuinely fine. People across teams make consistent decisions without consulting anything, and if you asked five of them what this place cares about you'd get roughly the same answer. Some organisations have this without ever writing it down, and running a values project on one of them adds ceremony to something that's already working.

Friction is starting to show. Two teams have made opposite calls on similar situations, or a new manager has done something that everyone senior found obviously wrong and couldn't quite explain why. That's the first real signal, and it's not a communication problem, it's that something implicit has stopped transmitting. The cheapest check is to find two recent decisions that went differently and ask what principle each one followed.

It has become a real cost. You're arbitrating the same argument repeatedly, or new joiners take months to make calls that fit. At this point the absence of an articulated standard is costing you senior time, and it's worth writing something down. What you write has to be usable in an argument, which most values aren't.

The edge case that forces it. You've merged two groups, or you've grown past the point where the founders are in most conversations. In both cases there are now two implicit sets of rules operating, and people are getting punished for following the wrong one without anybody realising that's what's happening. This is the situation where values genuinely earn their place, because the alternative isn't an unwritten shared standard, it's an unwritten contested one.

Five Questions This Reader Asks at 11pm

How many should we have? Fewer than you think, and the constraint is memory rather than completeness. If people can't recall them without looking, they can't use them in the moment a decision is being made, which is the only moment they matter. Three or four that people actually know beats seven that nobody can list.

Who should write them? Whoever can point to real decisions, which usually isn't a committee and isn't an agency. The best raw material is a set of choices the company has already made, including the uncomfortable ones, because those reveal the actual preferences. Writing from aspiration produces a document about who you'd like to be, and people can tell the difference immediately.

What if our leaders don't follow them? Then you have one problem, not two, and it isn't the values. Publishing a standard that senior people visibly breach is worse than publishing nothing, because you've given everyone a precise measure of the hypocrisy. Either the behaviour changes or the value comes down. Leaving both in place teaches people that the document is decorative, and that lesson generalises to every other document you produce.

Should we involve everyone? Consultation helps adoption and hurts clarity, and you should decide which you need more. A survey of the whole company reliably produces the words nobody can disagree with, which is exactly the failure mode described above. A defensible middle is to draft narrowly from real decisions, then test the draft widely by asking whether people recognise the place, rather than asking whether they like the words.

Do they actually change anything? Only if they're used to say no to something. A value that has never caused a decision to go differently is not operating, regardless of how often it's mentioned. That's the measure worth tracking, and it's the one nobody tracks: name the last thing this value stopped.

Three Honest Categories the Approaches Split Into

Aspirational, describing who you want to be. The values state a direction rather than a current reality, and everyone understands that the company is working towards them. It's right in a genuinely new organisation where there isn't enough history to derive anything from, and it's right when the leadership team is willing to be held to the gap publicly. It fails in the ordinary case, which is that the aspiration is stated in the present tense and the people inside can see it isn't true. That's not a small dishonesty. It's the thing that converts a neutral document into a source of cynicism, because it tells employees that leadership either doesn't know what the place is like or is comfortable saying otherwise.

Descriptive, naming what's already true. The values are derived from decisions the company has actually made, including the ones with costs attached. It's right almost always, and it's the only approach that produces a document people recognise. It also has immediate practical use: a new joiner reading an accurate description learns in an afternoon what would otherwise take six months to infer. It fails on flattery, which is why it's rare. An honest description includes traits that aren't attractive, and the version that survives the review process usually has those removed, at which point it stops being descriptive and becomes aspirational without anybody deciding that.

Directive, written as decision rules. The values are phrased as instructions for resolving specific tensions: when these two good things conflict, prefer this one. It's right when the actual problem is inconsistent decisions, which is the most common reason companies want values in the first place. It's also the most testable form, since you can check whether anybody has used one. It fails when the tensions chosen are the wrong ones, or when the organisation isn't actually willing to accept the cost of the stated preference. A rule saying you prefer speed over certainty is meaningless in a company that punishes every visible error.

Five Diagnostic Questions You Can Self-Assess Against

Take each value and write its opposite. Would a real company claim it? This is the whole test in one exercise and it takes ten minutes. The opposite of "we move fast" is "we're careful and deliberate", which plenty of serious organisations would proudly claim. The opposite of "integrity" is nothing anybody would put on a wall. Cross out everything in the second category.

Ask five people to name them without looking. Do it casually, not as an exercise, and count what comes back. If the average is under half, the set is too long or too generic to be held in mind, which means it can't be used at the moment of a decision. This is worth doing before a refresh rather than after, because it tells you whether the problem is the words or the number of them.

Name the last decision each value changed. If you can't, for a given value, it's not operating. Do this per value rather than in general, since it's common to find one that does real work and three that are decorative, and that's useful to know before you spend a quarter rewriting all four.

Find the value your company most visibly breaches. Every set has one. Ask what would have to change for it to be true, and whether anybody is willing to do that. If the answer is no, the honest move is to remove it, and removing a value the company doesn't live is one of the most credibility-building things a leadership team can do.

Ask a recent joiner which ones they'd have guessed. Somebody two months in has watched the place without having normalised it yet. Ask them what this company seems to care about based on what they've seen, and compare the answer to the list. The gap is your actual problem, and it's more accurate than any survey.

If you want a sharper version of the same exercise, ask somebody who left recently and had no bad experience. People who resigned on good terms will often describe the place with a candour that nobody currently employed can afford, and they have no reason to soften it or to sell it. What they name as the company's real character is usually the most accurate description available to you, and it costs one conversation.

Five Ways Values Get Written, Reviewed

The leadership offsite

The senior team goes away for a day or two, works through what matters, and comes back with a set. It earns its place on ownership and speed. The people who'll have to enforce the values are the ones who chose them, which matters more than it sounds, and a small group in a room can have the honest conversation that a large process can't.

Where it falls short is in what a room like that produces under mild social pressure. Nobody wants to be the person arguing against a nice-sounding word, so the list drifts towards the uncontroversial, and the sharp version that would actually decide something gets softened into a quality. The other failure is distance: a group of senior people describing the company are describing their own experience of it, which is not the experience of somebody four levels down.

It works if you add one constraint to the room: for every value proposed, somebody must name a real decision it would have changed, and a cost the company would have to accept. Values that survive that are usually real.

The second constraint worth imposing is on who speaks first. Offsites of this kind are shaped almost entirely by whoever opens, because the first suggestion sets the register for everything after it, and a comfortable opening produces a comfortable list. Ask the most junior person in the room to go first, and ask for the trait they think the company is least willing to admit to. The list that comes out of that conversation is usually shorter and considerably more useful.

The all-company survey

Everybody is asked what they think the values are or should be, and the results are analysed for themes. It earns its place on legitimacy. People are more willing to accept a standard they were asked about, and the survey sometimes surfaces a genuine trait that leadership had no idea was seen as characteristic.

It falls short because aggregation destroys specificity. Averaging thousands of responses produces the words that appear in the most responses, which are the generic ones, and the interesting minority answers disappear into the noise. It also implies a promise: people who suggested something and see no trace of it in the result draw the reasonable conclusion that the consultation was decorative, which costs more goodwill than never asking.

The better use of the same effort is to test a draft rather than generate one. Ask whether people recognise the place in this description and where it's wrong. That produces usable information and makes no promise you can't keep.

There's one survey question in this area that's genuinely worth asking, and it isn't about values at all. Ask people what they'd warn a friend about before they joined. The answers are specific, they're about real trade-offs, and they describe the company as it's actually experienced rather than as it presents itself. Whatever comes back repeatedly is a cultural trait, whether or not you'd have chosen it, and a description that omits it will not be recognised.

The founder writing them alone

One person, usually the founder, writes what they believe and publishes it. It earns its place on coherence and conviction. The result is specific, often uncomfortable, and reads like it came from somebody rather than from a process, which people respond to. In an early-stage company it's frequently the most accurate method available, because the culture genuinely is one person's preferences at that point.

It falls short as the company grows past the founder's direct reach. What was an accurate description becomes a description of one part of the organisation, and the parts it doesn't describe are told they're the exception. It also tends to preserve traits that were useful at a smaller scale and are now costly, because the author is the person least able to see them as choices rather than as obvious truths.

If it's written this way, it needs a review by people who joined later and can say which parts no longer describe anything they recognise.

Distilling them from decisions already made

Somebody assembles a set of real decisions from the last year or two, particularly the contested ones, and asks what preference each revealed. The patterns become the values. It earns its place because it's the only method that produces a document people recognise, and because it surfaces the trade-offs the company actually makes rather than the ones it would like to make.

It falls short on comfort and on effort. It takes real work to assemble the decisions, and the honest reading of them usually includes at least one trait nobody wants to write down. The review process then removes exactly those traits, which is how a descriptive exercise turns into an aspirational document without anybody noticing the substitution.

It's still the best approach on this list. The discipline that makes it work is agreeing in advance that the unflattering findings stay in, and getting that agreement from the people who could later remove them.

Adopting a set from elsewhere

The values are taken or adapted from another organisation's published set, usually one the leadership admires. It earns its place in one narrow case: as raw material for a conversation, where reading somebody else's sharp, specific values makes it easier to articulate your own. Seeing what a real value looks like is genuinely useful when everything you've written so far is a quality.

It falls short as an endpoint for a reason that has nothing to do with originality. Another company's values encode their trade-offs, made under their constraints, in their market. Importing the words without the constraints produces a statement that the organisation has no structural reason to honour, and it'll be abandoned the first time it's expensive. People also recognise borrowed language faster than anyone expects.

Use it to calibrate ambition, then throw it away and write from your own decisions.

The borrowing worth doing is of form rather than content. Read a set that's genuinely sharp and notice what makes it so: it names a trade-off, it admits a cost, and it would read as a mistake to somebody running a different kind of company. Those three properties transfer. The specific preferences do not, because they were chosen under constraints you don't share and enforced by consequences you haven't built.

The Decision Table

The row that costs the most is the fourth, and it's the one most often left alone because addressing it requires a conversation about a specific senior person. Every week a visibly breached value stays on the wall, it teaches the organisation that the stated rules are optional at a certain level, and that lesson is expensive to reverse.

The Test a Value Has to Pass

Four tests separate something usable from a slogan. A statement that fails any of them will not survive contact with a real decision.

The second test does more work than it appears to. Every genuine value costs something, because it expresses a preference between two things that are both desirable. Choosing speed means accepting rework. Choosing thoroughness means moving slower than competitors sometimes. Choosing transparency means occasionally sharing things that are awkward. A value with no cost attached hasn't chosen anything, which is why it's so comfortable to agree with and so useless in an argument.

The third test is the one to apply annually. Values decay quietly: they get cited in all-hands presentations and recruitment materials while never appearing in a single operational decision, and from the outside that looks like health. Asking each year what each one stopped is the cheapest way to find out which ones are still alive.

Making Values Do Work After the Launch

A launch is where most values work ends, and it's the point at which none of it has happened yet. Four places determine whether the document survives the year, and all four are ordinary processes rather than culture activities.

Hiring, specifically rejection. If a value has never been the reason a capable candidate wasn't hired, it isn't operating in hiring. This is the most testable use available, and it's also the one that most reliably reveals whether the organisation means it, since rejecting somebody good on the grounds of fit is expensive and gets challenged.

Promotion, and the reasoning given for it. People study promotions harder than any communication, so a promotion justified partly in the language of a value teaches the value to the whole organisation at once. The reverse is equally powerful: a promotion that obviously contradicts one quietly retires it.

Difficult conversations with senior people. The first time a value is applied upward is the moment it becomes real, and until then everybody is waiting to see whether it will be. This is almost always the hardest of the four and the one that decides the rest.

And the everyday decision where two good things conflict. This is where a directive value earns its keep: somebody facing a genuine trade-off can point at the written preference and decide without escalating. If the values never come up in those moments, they aren't written as decision rules and should be rewritten or retired.

One thing not to do with them, which is common and does quiet damage: don't attach them to the performance review as a scored dimension. Rating somebody out of five on a value converts a decision rule into a personality assessment, and it produces exactly the behaviour you'd expect, which is people performing the value rather than using it. Values work as constraints on decisions. They work badly as measures of a person.

What to Put in Writing

The values themselves are the least useful thing to record. What gets forgotten is the reasoning, which is what makes them usable a year later when the people who wrote them have moved on.

The last row is the one no organisation writes and the one that would do the most good. Naming your own gap internally, before somebody else does, is the difference between a leadership team that's honest about a problem and one that's discovered to be unaware of it.

Questions to Ask Before You Commit

On the test. For each value, could a reasonable competitor claim the opposite? A bad answer explains why the word is important rather than answering the question.

On cost. What do we give up by holding this? A bad answer says it makes us better at everything.

On use. Name the last decision this changed. A bad answer describes where the value appears rather than what it did.

On the breach. Which of these do we most visibly fail? A bad answer is none of them.

On derivation. What real decisions did we write these from? A bad answer is a workshop.

On recognition. Would somebody two months in have guessed these? A bad answer is that they will learn them in onboarding.

What Getting This Wrong Costs

The first cost is the one people expect and it's the smallest: wasted effort. A values project consumes a few weeks of senior attention and some design budget, and if the result goes in a drawer the direct loss is modest. Treating this as the main risk is why the subject gets so little rigour, since a low-stakes exercise doesn't warrant a difficult conversation.

The second cost is larger and lands on credibility. A published standard that's routinely contradicted doesn't sit inertly. It gives every employee a precise, company-authored measure of the distance between what leadership says and what happens, and people use it. What was previously a vague sense that things aren't quite right becomes a specific, quotable gap, and the document that was meant to align people becomes the reference point for their cynicism. That effect is strongest among the people who took the values seriously in the first place.

The third cost is the one that compounds. Once an organisation has run a values exercise that changed nothing, it becomes much harder to run anything similar again. Employees learn what this kind of initiative amounts to, and the next attempt, even a good one, is received as another round of the same. Companies in this position often conclude that their people are cynical, when what's actually happened is that their people are correctly generalising from evidence.

So before you begin, answer a narrower question: is this a consistency problem, a clarity problem, or a credibility problem? Consistency means teams decide the same question differently and a directive value would genuinely help. Clarity means nobody has articulated what's expected and a descriptive one would. Credibility means everyone knows what's expected and watches it being ignored, and no document addresses that at all. The third is the most common reason people ask for a values project and the one values can't fix.

When You Are Ready to Go Further

None of this needs a budget or an agency. It needs a set of real decisions to read, a leadership team willing to name the trade-offs those decisions reveal, and the discipline to keep the unflattering ones in.

The step after that is making the values load-bearing in the places people actually watch: what gets somebody hired, what gets somebody promoted, and what happens when a senior person contradicts one. That's less a communications question than a question about how consistently those decisions get made across the organisation.

HROpsLab publishes independent comparison work across HR tooling, applicant tracking and payroll. We sell nothing, we take no vendor money, and we publish no paid placements. If the next step is looking at the consistency of the decisions your values are supposed to govern, our comparison work is one place to start.

Frequently Asked Questions

A value nobody has ever used to say no isn't a value. It's a description of wanting to be liked.