INSTRUMENTS

Do not send a survey. A survey is a request for a declaration, and declarations are exactly what this population has learned not to make. You will get zeros, you will believe them, and you will be wrong.

Ask yourself these five questions instead, and answer them honestly — on paper, which is why there is no form here.

1. What do I do in a week that I have quietly made easier, without telling anyone how?
2. Who on my team is faster than their method explains?
3. If someone on my team were paying out of their own pocket to do their job better, how would I ever find out?
4. What is the biggest programme in my area, and what has it shipped this year?
5. What would it take for a six-hour tool to get five minutes in the same room?

None of these asks anyone to confess to anything, and that is deliberate. By the end of Chapter 11 you will understand exactly why asking directly is the one method guaranteed not to work.

The next time somebody in your organisation describes a problem as new — the AI shadow-tool problem, the remote-work-visibility problem, the citizen-developer problem — write it on the left of a page. On the right, write the oldest version of the same problem you can find.

Not the same technology. The same shape: someone routing around the official way to get the work done, and hiding it.

You will almost always find one, and it will almost always be older than you expected. That gap — between how new the problem feels and how old its shape actually is — is the finding. A five-year-old problem might yield to a new tool. A forty-year-old one will not.

Take one improvement in your area, a real one, something a single person could conceivably build, and estimate what it would have cost to create and to conceal in three eras.

1995, the shadow IT era. Weeks of evenings, real technical depth, and something visible — a database, an application, a machine under a desk. Someone would notice.

2015, the SaaS era. A cloud tool, a corporate card, maybe a departmental subscription nobody scrutinised. Cheaper to build, moderately concealable, until procurement or security caught up.

Today. A chatbot, a personal account, an afternoon.

Answer these about the last good small idea that reached you, or that you brought to someone else.

Did it die at a meeting, or before one? If before, you are looking at anticipation rather than evaluation, and no amount of fixing the meeting will help. Most deaths are before.

Was it rejected for what it was, or for where it came from? "It needs to align with the roadmap" is a where-it-came-from rejection wearing a what-it-was costume.

Take the last ten ideas, requests or proposals that crossed your desk. For each, ask a single question before any other: is this a one-way door or a two-way door?

A one-way door is expensive to reverse, or impossible. A hire. A platform migration. A press bolted to a floor. Get these wrong and you live with it.

A two-way door is cheap to try and cheap to undo. Most software. Most process tweaks. Most tools one person could build in an afternoon. Get these wrong and you close the tab.

Now the diagnostic question, and answer it honestly: how many of the two-way doors did your process treat like one-way doors?

Less a calculator than a comparison you have almost certainly never been shown, in four parts.

One. Find the forecast. Take the largest completed programme in your area — completed, not running, so the answer exists. Find its original business case and read the expected benefits. Not the slide. The document, with the numbers that got it funded.

Two. Find out what happened. Establish what it actually delivered, in the same units. This is the hard part, and the fact that it is hard is itself the finding.

Three and four. Take one small improvement that already works, and put the two side by side.

You are going to find out what your organisation has already tried, which in most places nobody knows, because the people who remember have moved and the record was never kept.

Count the graves. List every programme launched in your area in the last ten years whose purpose was to get more small improvements out of the people doing the work — suggestion schemes, idea portals, protected time, innovation days, hackathons, improvement circles, kaizen waves, digitalisation initiatives, whatever the local name was. Include the ones nobody mentions any more.

Aimed squarely at anybody who sits on a panel. Unlike the earlier instruments it asks you nothing about culture and nothing about how anyone feels. Every question below has a factual answer.

The last five promotions. List the last five people promoted into decision-making roles in your area, and for each write down the single thing they are most associated with — the thing that comes up when their name comes up. Then sort those five things into three piles: a launch, a turnaround, or a thing that has quietly run well for years.

The last five. Write down the last five ideas anyone brought you. Not proposals that arrived through a formal process — ideas, including the ones raised in a corridor or at the end of a meeting when the agenda ran out.

For each, answer three questions. What happened to it? — "taken forward" is not an answer, but "went to the March council and was funded" is, and so is "I meant to look at it." Did the person find out what happened, and how? Not whether an outcome exists — whether they were told. And how long did it take?

Silence from your side reads as futility from theirs, whatever was actually happening.

The first instrument in this book aimed at the person holding something rather than the person receiving it.

Name it and date it. Write down what you have built, in one sentence, as if to somebody who does not do your job. Notice how long that takes. Most people find they have never said it out loud, and the first sentence you produce under pressure is the one that will be in the room for the rest of the thing's life.

Then two dates: when you started using it, and when it started being worth something. They will not be the same date, and the gap is the thing to look at.

Every question below is factual and answerable from records or from a five-minute conversation. None asks you to rate your own culture, because you cannot and neither can anybody else.

What decision changes? Write down the decision that will be made differently because of this data. Not the goal — the decision. Who makes it, when, and what are the options in front of them.

If you cannot name one, you have not bought an instrument. You have bought a camera.

For the person with a budget and an org chart, because everything in it requires one.

Sort the last ten. Write down the last ten changes of any size that happened in your area. For each, mark which physics it belongs to: irreducibly large — capital, coordination, all-or-nothing — or local, reversible and compounding. Then mark which route it actually travelled.

If everything went through the same route regardless of which column it sits in, you have one system. That is the finding.

The three moves that follow from this — the protected window, the answer guarantee and the reversible router — live on the mechanisms page, with their parameters exposed and versioned.

The last instrument, and unlike every other one in this book it does not ask you to audit anything. One action each, for the three people who read this, and each small enough to do without asking anybody. That constraint is not modesty — a move that requires a programme has already failed the argument it comes from.

If you have built something. Do not surface it. Get a second user. One person, by name, this week — somebody who will run the thing themselves rather than watch you run it.

If people bring you ideas. Go back to the last one you did not answer, and answer it. Late is survivable; silence is not.

If you set the rules. Grant one window. Six weeks, a small cap, and nothing owed at the end but a conversation.