Your team will always do as you do, never as you say.

When I was a boy, my dad used to tell me: do as I say, not as I do. Even at eight I remember thinking it did not quite add up. If the thing you are doing is fine, why can't I do it? And if it is not fine, why are you doing it?

I never got a satisfying answer. What I did not know then is how much of my working life I would spend watching grown, capable, well-paid leaders say exactly the same thing to the people who report to them, and be just as surprised when it does not work.

A perfectly reasonable sentence that fools nobody

“Do as I say, not as I do” is a kind and human phrase, when you look at it gently. It usually means: I know I get this wrong, I would rather you did not, so learn from my advice instead of my flaws. Parents say it out of love. Leaders say it out of a similar instinct, most of the time. Nobody is being a villain here.

The trouble is that it rests on a belief that simply is not true. It assumes people follow your words. They do not. They follow your behaviour, and they always have. A child watches what you do and quietly files the words under “things adults say.” A team does precisely the same thing, only with a spreadsheet.

So the sentence is not really an instruction. It is a confession dressed as one, and everyone in the room can tell the difference.

The engagement I keep thinking about

I came off a long piece of work recently that has stayed with me, because it made the point better than I ever could in an essay.

The executive team had asked the whole organisation to adopt a new framework and a new way of working. Fair enough, it was a good way of working, it is rather the point of what we do. And then, quietly, they exempted themselves from almost all of it.

They were the last to adopt the method they had mandated. When it came to writing the narrative under their goals, the reasoning, the beliefs, the choices, they wrote the least of anyone, a line or two where their teams wrote paragraphs. And the outcomes they set for themselves were, let us say, carefully chosen to avoid anything resembling real risk or accountability.

In some ways that reflects badly on me, and I have made my peace with owning that. But as my dad would also have said, you can lead a horse to water, and you cannot make it drink.

Here is the part that has stayed with me though. The teams reporting to those leaders did exceptionally. They wrote rich narratives, they set genuinely ambitious outcomes, they leaned in. The good work did not come from the top. It happened one level down, among people who decided to take the thing seriously whether or not the people above them did.

Which tells you two things. That you can have wonderful people doing wonderful work despite their leaders. And that no organisation ever rose above the ceiling its leaders set by example, it only ever had pockets that briefly poked through it.

Rules for thee, and rather nicer rules for me

We are all, at some level, unusually sensitive to this particular unfairness. One set of rules for you and a softer set for me is the thing that curdles trust faster than almost anything else a leader can do.

You do not need me to labour it. If you want the fully worked example, complete with the part where everyone found out, ask Boris Johnson how the parties went down while the rest of the country was under the rules he wrote. The specifics are political and I will leave them there. The lesson is not political at all. The moment people conclude that the rules apply to them and not to you, you have not lost an argument. You have lost their belief, and belief is the only thing that was ever making the rules work.

Leaders tend to think they are judged on what they decide. They are mostly judged on what they do when they think the decision does not apply to them.

The same old sentence, now wearing an AI badge

Which brings me, inevitably, to the version of this I am watching play out in nearly every company right now.

The message from the top is genuine and it is everywhere. I want you to train on AI. Use it to make yourself more effective. Automate the repetitive work. Build agents. This is the future and I need you all to embrace it.

And then the same leader, in the same week, uses AI to tidy up an email and write the odd report, and considers the matter handled.

It is do as I say, not as I do, in a fresh outfit. And it lands exactly as well as it did when my dad tried it, which is to say, not at all, because the organisation is doing what it has always done. It is watching what you actually do.

This one matters more than the others

In the previous eras of business, a leader who would not adopt the new tool lost a little efficiency and the company muddled on. The stakes were mild. That is no longer the case.

If the advantage now comes from faster learning loops, from creativity, from teams genuinely reinventing how work is done with these tools, then AI adoption is not a productivity tweak. It is the whole game. And adoption is a behaviour, not a memo. A workforce takes up AI in earnest only when the people at the top are visibly, slightly awkwardly, learning it in public alongside them. Building the clumsy first agent. Sharing the prompt that failed. Being a beginner where everyone can see.

A leadership team that mandates AI while using it to spellcheck their emails is not driving adoption. They are quietly signalling that this is a thing for other people, and the organisation, ever the faithful mirror, will believe them.

Going first is the actual job

None of this is a counsel of perfection. I am not asking leaders to be flawless, which is impossible and, honestly, a bit exhausting to be around. My dad was not a hypocrite for getting things wrong. He was only ever wrong about one thing: believing the words would do the work that the example was quietly undoing.

The fix is not to say the right things more loudly. It is to go first. Adopt the method before you ask anyone else to. Write the longest, most honest narrative in the company, because yours sets the watermark for everyone else's. Set an outcome ambitious enough that you might genuinely miss it, and let people watch how you handle that. Learn the AI properly, in the open, badly at first. Put yourself inside the frame you drew, not above it.

Do that and you will not need the sentence any more. You can retire “do as I say, not as I do” for good, because the two halves will finally match, and your team will do as you do, which for once will be exactly what you were hoping for.

The one line to leave you with

You can lead a horse to water, and you cannot make it drink. But you can do the one thing that has always worked, and drink first.

Your team is not reading your mandate. They are watching your hands.