Your people trust you. That is not the same as trusting each other.

The 2026 Edelman Trust Barometer found that employees trust their employer more than any other institution: 78 percent, ahead of business in general and well ahead of government. It also found that seven in ten people are unwilling or hesitant to trust someone whose values, cultural background, information sources, or way of approaching a problem differ from their own. Set those two findings beside each other and a conclusion follows, which is roughly the one Edelman draws: if the employer is the last institution people still believe, the employer is best placed to broker trust between them. Employees say the same thing. They expect their employer to do exactly that.

It is a fair reading of the data. It also asks for something an employer brand may not be able to give, and the reason is worth sitting with before any of the work starts.

The trust in that 78 percent runs vertically. It goes from a person to an institution, and it means: this organization will treat me fairly, pay me on time, not lie to me about the things that shape my life. It is real, it is hard to earn, and if you have it you are holding something most institutions have lost.

The trust that is thinning runs sideways. It goes between people who have to get something done together, and it means: I can work with someone who reads this differently than I do.

Those two come apart more easily than the phrase “a high-trust culture” suggests. You can have a workforce that believes in the company and cannot sit in a room with each other, and that combination is stable rather than fragile. The institution stays trusted precisely because it is the neutral thing everyone can still agree on while agreeing on little else. High vertical trust can be a symptom of low lateral trust as easily as a cure for it. The company becomes the only safe subject.

So when the expectation arrives, and by employees' own account it already has, the useful question is less whether to take it on than whether what you are holding is the right material for the job.

Consider a company with a value it is genuinely proud of. Say it is something like “we speak straight.” It is on the wall, it is in the recruitment film, and — this is the part that matters — the people who work there believe in it. Ask them and they will tell you it is real.

Now watch two teams live it. One reads speaking straight as saying the difficult thing in the meeting, in front of everyone, while it still matters. The other reads it as saying the difficult thing to the person directly, first, and never ambushing a colleague in public. Both teams are living the value. Both are right about what it means. And each one experiences the other as violating it: one side sees cowardice and back channels, the other sees people using candor as cover for humiliating a colleague.

Nothing has gone wrong with the Employer Brand as it is usually measured. Trust in the company is high in both teams. The promise was made and the promise is believed. What has gone wrong is that the promise was never specific enough to be lived the same way twice, and the space it left has been filled in by whoever got there first. That failure sits below the level most employer-brand work looks at, which is why it tends to surface as a personality problem or a management problem rather than a brand one.

The standard instruments ask the vertical question. Do you trust leadership. Would you recommend us as a place to work. Do our values mean something to you. Those are worth asking, and in the company above they come back healthy, because they are healthy. Every one of those teams believes in the value.

The lateral question is not what those instruments were built to measure: can you work with a colleague who reads our values differently than you do. That is where the 42 percent of employees who told Edelman they would rather switch departments than report to a manager with different values actually lives. It does not look like a brand question. It looks like a staffing problem, or a personality clash, or the sort of thing an offsite is meant to fix. It is a brand question. A value that produces two incompatible ways of behaving is one word doing duty for two different commitments.

If your organization is going to hold trust between people who have stopped extending it to each other, it has to be able to say what it believes with enough precision that two reasonable people cannot take it in opposite directions and both feel loyal.

That is harder and less appealing than it sounds. Precision closes doors. A value specific enough to settle an argument is specific enough to exclude someone you would otherwise have hired. A values statement written so that nobody could object to it was never designed to be used at the moment people disagree; it performs well in a recruitment campaign and goes quiet exactly when it is needed.

So the work is unglamorous. It means going back to the small number of things you claim and asking of each one: when two good people read this differently, does it tell them what to do? If it does not, it is decoration, and decoration cannot broker anything.

This produces no film. It usually leaves fewer words on the wall rather than more, and the ones that survive are narrower and more uncomfortable than the ones they replace. It is also the only version of this work that would still be standing at the moment someone needs it.

Coherence is usually described as the distance between what a company says and what it does. That is the version that is easy to picture and easy to audit. There is a second distance, and this year’s findings point straight at it: the distance between what a company says and what its own people take it to mean. You can close the first and leave the second wide open. A promise that only holds where people already agree is not a promise; it is a coincidence you have been reading as strength.

The trust you have been handed is real. Whether any of it can be passed between the people who work for you depends on something smaller and far more specific than trust: on whether you ever said what you meant clearly enough that someone could disagree with it.