The Rule That Could Kill a Good Idea

In twenty-five years of changing one company, nobody ever handed me a list of the things I was not allowed to change.

I built things that had not existed before. New functions, new ways of reaching people, a whole channel where there had been a catalogue. Every one of them changed what the company was to somebody. Every one of them was approved by people who read the case and agreed with it.

Not once did anyone say: hold this, whatever else you do. I worked it out instead, the way everyone does. I got close enough to the people who already knew, and I stayed long enough to absorb what they would never have allowed. That worked, roughly, because I was there a very long time. It is a terrible way to run a company through a change.

Except in one place.

The one thing that was written down

The physical product had rules. Real ones, not a poster in a corridor. A body of encoded knowledge, taught, argued over, enforced, and older than almost everyone applying it.

How the parts connect, and the tolerance they connect at, held to three decimal places. Which shapes are allowed into the permanent vocabulary and which may only ever serve one purpose. What the design language does and does not do. What the name may sit next to, and what it may never sit next to.

I watched good ideas lose those arguments. I watched people spend weeks on a case and get turned down over a decimal.

That is a continuity line, and it is the best one I have ever worked inside. It was written well enough that someone who had never met the people who wrote it could pick it up and make the call they would have made. It held for decades. It survived a near-collapse, a rebuild, and more reinvention than most companies attempt in a century. Everything else about that company changed, repeatedly: what it sold, who it sold to, how it reached them, what business it thought it was in. The line held through all of it, and the line is a large part of why all that change stayed recognisable.

It covered the product and the name.

Plenty got written after that. Five years into my time there I sat down with the chief executive to build a brand framework, because no such thing existed and the near-collapse had made the absence impossible to ignore. It was real work and it was needed. Policies have arrived in quantity ever since.

The problem was never that nothing was written down.

It is that almost none of it could turn anything down.

Where the rules get teeth

This happens in companies that care a great deal. It is a pattern, and once you see it you see it everywhere.

Rules get teeth where the loss would be fast, visible and expensive. Get a tolerance wrong and the parts do not fit, and you find out in a week. Put the name next to the wrong thing and a lawyer calls. Those failures have an owner and a date, so the rule that prevents them is allowed to cost something. It is allowed to be expensive.

Where the loss is slow and invisible, you get documents instead. How you say no to people. What you will not do to make a quarter. The thing you always give customers that costs money and that nobody ever asked for. The tone you take when you have got something wrong and have to say so. All of that gets described, at length, in frameworks and guides that everybody in the room agrees with. None of it gets a price. Each one can be traded away quietly, in a single meeting, by a capable person with a decent case, and nothing is technically broken.

So here is the shape most large companies are actually in. One domain where the rules bite. Everywhere else, documents you can hold for free.

Which is why the question feels answered when it is not. Ask a marketing leader whether she has continuity through her transformation and she will point at real things: a brand book, a design system, trademark rules, a tone-of-voice guide. She is not bluffing. Those documents exist, they took work, and some of them are enforced. The ones with teeth govern the marks. The rest describe who the company is without ever naming what being that costs.

The decision nobody owns

The first gets an owner, a budget, a steering group, a plan and a name on a slide. The second gets asked anyway. Not once, in a room, by people accountable for the answer. Several hundred times a week, in briefs and reviews and supplier calls and pricing decisions and service scripts, by whoever happens to be building that particular piece on that particular day.

They are not being careless. They have plenty of guidance and no line. So they do what I did, which is work it out from proximity: from whatever they have picked up about what would be allowed. Proximity produces different answers in different people. That is what proximity is for. It is a wonderful way to develop somebody and a hopeless way to hold a company together while you are pulling it apart.

Drift is the sum of those answers. It is not an emotional event, and it is not a measurement failure. It is a decision nobody owns, taken several hundred times, by default.

That is also why you cannot catch it. Every instance is defensible, because every instance was defended, in a meeting, successfully. There is no bad decision to find. You go looking for the moment it turned and there wasn't one.

The reaches were all sensible

I watched the slow version of this before the company I worked for came close to going under.

The business kept reaching into new things. Each reach had a case. There was growth to be had, the categories were adjacent, the logic held in the room. And with each reach the company became a little less the thing people had loved it for. The brand loosened its grip on them long before the accounts showed any strain.

When the crisis finally broke, it broke loudly, and the comeback is the part everyone remembers. The quiet part had been running for years by then. Nobody made a bad call. Every one of those decisions was made by capable people with a good argument and the authority to act. What was missing was anybody with the standing to say: not that one, because of this.

The numbers were fine for most of it. They usually are. The feeling goes first and the spending follows late, out of habit and contract and never quite getting round to switching, so the dashboard stays green.

Write the line for the decisions

The move here is not more brand guidelines. You have those, and they are not the problem. The move is to give the decisions the same treatment you already give the marks. Write down what does not change, before you publish what does.

Call it the continuity line. It is short. Five things, not fifty. And there are three tests for whether what you have written is a line or a poster.

• It has to be losable. If holding it never costs you anything, it is not a line. "We put the customer first" costs nothing, and nobody has ever had to turn down revenue to hold it. Name the thing you would say no to. If you cannot point at a deal, a saving or a launch you would give up rather than break it, you have written a description. Descriptions do not survive a bad quarter. This is the test almost everything in the drawer fails, including work I am proud of.
• It has to be encounterable. Something a customer actually meets. A rule about how you say no. A response you always give, even when it is expensive. A cost you eat instead of passing on. A constraint you never engineer around. The element rules worked because they were physical and could not be fudged. The rest of the business has to reach the same hardness through behaviour.
• It has to survive your absence. Take what you have written to somebody who was not in the room and has no history with you. Hand them three live decisions from this week. If they cannot make the call you would make, the line is still in your head rather than on the page, and it will not reach the people who are making those calls without you.

That third test is the one the element rules passed and almost nothing else did. It is why they held for over fifty years while everything around them moved. Written well enough to be applied by strangers, which is the only kind of rule that outlives the people who wrote it. There is a second reason to write it down, and it matters more the further you sit from the top.

If you need the argument in a finance director's language, it is this. Drift is not paid for in affection. It is paid for in re-acquisition. You spend next year's budget re-explaining yourself to people who already knew exactly who you were, and you book it as awareness. It shows up in her plan long before it shows up in a tracker.

What proximity used to cover

There is a reason companies got away with this for so long.

The whole unwritten part ran on proximity. I absorbed it by being near people who already had it, for years, and then other people absorbed it from me. That worked because change was slow, and because most of what mattered still passed through a small number of hands. The line did not have to be written down, because the people carrying it were in the room.

That enforcement is gone. Decisions are made now at a volume and a speed no small set of hands can sit across, by people who were never in that room, and increasingly by systems that never can be. An unwritten line was survivable when drift took five years. It is not survivable when it takes five months.

Which raises a harder version of the same question, one floor down, and it is the one I want to take up next month. The unwritten part was never only about the brand. It was the judgement itself.

I got twenty-five years of proximity. Most of the people making these calls in your company this week will get eighteen months and a handover document. Proximity is not coming back, and nothing has been built to replace it.

So write the line. Something shorter and harder than a values document. The short list of things that do not change, specific enough that a stranger could hold them for you.

The only line that holds is the one that can kill a good idea.

Next month: AI is taking over the work that used to build people's judgement. Your team is getting faster by the week. What's quietly not being built while they do.

Sources

Building on

• Donella H. Meadows, Dancing with Systems (2001) — wisdom #8, "pay attention to what is important, not just what is quantifiable." The green-dashboard beat rests on it: the loss that matters is the one no instrument is pointed at.
• Tony Fadell on manufactured constraints (2026) — when the constraints are not given to you, invent them; the argument that a hard limit is what makes range possible rather than what blocks it.
• Unowned Work Is the Real Startup Killer (2026) by Katie Barnes — the general case of the failure this piece diagnoses in one domain: work that matters, that nobody is accountable for, gets done badly by everyone at once.