#20 Courage (Part 2 of 2) — Making It Cheap to Try

Last week I made the case that most good ideas in established companies aren't rejected. They're slowed until the energy drains out of them, because an attempt costs too much to make casually.

The obvious conclusion is to make everything small. Run pilots. Keep it reversible. Start on Monday.

That's half right, and the wrong half causes real damage. So let me start this edition by arguing against the advice I'm about to give.

(New here? This stands on its own, but the diagnosis behind it was last week's edition on why nothing starts. And the Digital Leader Canvas, the free companion to this series, is here.)

Reading time: 7 minutes. If you take one thing from this: "make everything small" is only half an answer, because some AI changes only work whole, and a hundred pilots can be a way of avoiding one decision. Ahead: four rules that lower the cost of trying, the Disney move for keeping critics in the right seat, and which instrument really builds courage in a team, which is rarely a training.

Not everything can be a pilot

Some of what AI now makes possible cannot be reached incrementally.

Redesigning a process end to end rather than bolting a tool onto the old one. Restructuring a function when most of its routine has become automatable. Committing to a platform that only pays off once everyone is on it. You cannot A/B test your way into a new operating model, and there is no two-week version of a decision that only works whole.

That MIT report I mentioned last week, the one claiming most enterprise AI pilots showed no measurable return, gets more interesting when you look at where the returns did appear: not in the scattered experiments, but in organisations that went deep enough to change how the work itself was done.

Which points at a failure mode nobody puts on a slide. A hundred small experiments can become a comfortable way of avoiding one real decision. Every individual step looks responsible. The programme looks busy. And five years later the operating model is exactly what it was, only now with a chatbot attached to it.

So there are two modes, and each demands a different nerve. Small steps take the courage to start without proof. Big moves take the courage to commit without proof, and then hold the line through the months when the new thing looks worse than the old one it replaced — which it usually does, for a while.

What you need is the judgment to tell which one is in front of you. Treating a transformation as a pilot costs just as much as treating a pilot as a transformation.

A quick aside for HR and people-development leaders. Two situations where a call makes sense right now. One: you still have development budget for this year and would rather it landed somewhere that shifts how your teams work than on an event they'll have forgotten by February. Two: you're putting next year's plan together and want the AI and human-skills part to be more than a line item. What I do takes months rather than days — coaching, workshops and longer programmes that change the conditions rather than announce a change — and I take on a limited number of organisations at a time. Twenty minutes, and you'll leave with a clear next step whether or not you work with me: book a call.

Four rules that lower the cost of trying

With that established, here's what works, in my experience.

Sort the decision first. Before anything else, decide which kind you're holding: a reversible experiment, or a change that only works whole. That one question prevents both standard mistakes — dragging a two-week trial through a full approval process, and sliding into a transformation nobody ever decided to make.

Replace the approval question. For anything reversible, try to make safe enough to try the standard. The phrase comes out of sociocracy and Holacracy, and it's the single most useful decision rule I know for this problem. It replaces are we sure?, which nobody can answer truthfully about anything new, with something you can settle in a minute: if this goes wrong, can we live with it and undo it? Save the full apparatus for what truly can't be undone. Most AI experiments are cheap, short and reversible, and get treated like a merger.

Move the worst-case sorter. Not out of the room, but to the right step. "How could this break?" is invaluable at design and review, and fatal at the gate. These people aren't blocking anything; they're in the wrong seat.

Run something small and fast. A mini-hackathon inside one team: an afternoon, a real problem, no slide deck at the end. What it builds isn't the prototype. It's the experience of having tried something and survived it, which is the only thing I've seen reliably shift an organisation's appetite for the next attempt.

The cleanest way I know to arrange that seating comes from Walt Disney, and I unpacked it in the Profiling edition. He ran ideas through three modes in sequence: the Dreamer who imagines without limit, the Realist who works out how it could be done, and the Critic who stress-tests it for flaws. Each mode got its own room. Robert Dilts formalised the approach in 1994.

One rule makes it work, and it's simple enough to introduce in your next meeting: nobody criticises while the room is still dreaming. Most meetings run all three modes at once, which is why an idea can die before anyone has finished describing it.

If you're the one meant to build it

For HR and people leaders, the instinct is to put on a training. Courage responds poorly to that, and well to a handful of other things. The trick is matching the instrument to what is really blocking people:

• The system may be the blocker. People have watched someone get punished for trying, and no workshop will touch that. Change what gets recognised: put a well-run failed experiment on the agenda of the meeting where wins are celebrated, and have a senior person say out loud that they backed it.
• The risk feels personal. A leader can remove that in one sentence: if this doesn't work, it's on me. Cover is cheaper than courage and works immediately.
• The individual is stuck. Fear of looking incompetent, or one attempt years ago that went badly. That's coaching rather than training, it's specific to the person, and it usually resolves faster than anyone expects.
• The team has no shared rules. That's a workshop: agree what's safe to try here, what needs a real decision, and what happens when something fails. The Culture Design Canvas gets it onto one page.
• Nobody has practised. Build the reps: pre-mortems, mini-hackathons, rehearsed difficult conversations. In the words of the 5Cs authors, this develops through challenge, not through modules.

Start where the resistance sits. Most organisations run the workshop when their problem is the reward system, then conclude that development doesn't work.

One more variable if your teams sit in different countries. Erin Meyer's Culture Map, a regular companion in this newsletter, has scales for how openly people disagree, how hierarchical leadership is, and whether decisions get made by consensus or from the top. All of them change what speaking up costs. What the map doesn't have is a dimension for tolerating uncertainty, and for that you want Hofstede, where Germany scores 65 on uncertainty avoidance against Sweden's 29 and Singapore's 8.

Both are national averages, and as the Profiling edition insisted, the variation between two colleagues usually exceeds the variation between two countries. Still, the combination produces a pattern worth recognising: a culture can be direct about disagreement and slow to try things at the same time. People will tell you plainly what might go wrong, and then wait. Which describes the sentence this pair of editions started with rather well.

The part that stays uncomfortable

None of this removes the moment where somebody has to decide with incomplete information and put their name against it. Lowering the cost of an attempt makes that moment smaller and more frequent. It doesn't abolish it, and any advice promising otherwise is selling something.

I say that plainly, because the four rules above can be read as a way of engineering the fear out of decisions. They aren't. They're a way of making sure the fear attaches to the decisions that deserve it, rather than to a feedback pilot in one department.

Your turn

Your 1% this week. Take one thing you've been calling a project and sort it:

• Really a two-week experiment, just labelled as a project? Run it. Tell one person you're doing it, so you can't drop it unnoticed.
• A real commitment you've been slicing into pilots to avoid making? Then the next step isn't another trial. It's a decision, and it needs a date in the calendar.

If you lead a team, one more. Add a single sentence to the next attempt somebody proposes: if this doesn't work, it's on me. Then watch what it does to the room. It costs you nothing until something fails, and by then you'll have learned more than the pilot would have taught you.

That closes courage. Next up are the two I haven't touched at all: compassion and curiosity. Creativity has been circling this newsletter since the Delegation edition and will get its own turn later. Which would you rather read first? Tell me in the comments — and if you've seen a big move that only worked because someone committed to it whole, I'd like to hear that one especially. I read every reply.

One favour before you close the tab. If you know a leader or an HR colleague sitting on a hundred pilots and one decision they haven't made, forward this to them. Word of mouth is how most people find this newsletter, and a recommendation from you carries further than anything I can write.

The Digital Leader Canvas is free for individual use. And if you're in HR or L&D and want to bring these themes into your organisation, here's that link again — book a 20-minute call. You can also explore the Leadership Essentials Workshop Series for teams.