What a rule costs to keep

Fourteen rules, one year later

The work in lessons 2 to 7 ended in written rules, as this work always does. Tanvara finished the year with fourteen of them: the code discipline, the ownership table, the grant expiry, the revision ritual, the section comments, and nine smaller ones.

A year later Hisham Barakat asked the only question that matters about a rule. It is not whether people like it. What does it cost us to keep, and is it actually being kept?

The monthly governance report said compliance was 96%. Then somebody ran a sample audit: pull twenty closed orders and check each rule against the evidence, not against the ticked box. That said 75.5%. Nobody had lied. The whole of the gap sat in five of the fourteen rules.

Read the last two columns of the third row together. The five rules with the highest reported compliance were the five that were barely being followed. And they used up 84% of the whole governance budget.

Why the invisible rules cost the most and deliver the least

This looks like a paradox. It is not. It follows directly.

A rule whose breach is visible needs no measuring apparatus. The ownership table from lesson 3 is kept because somebody notices at once when a handover has not happened: the mill asks a question and nobody answers it. It costs almost nothing to keep, because the checking is a by-product of the work.

A rule whose breach is invisible has to be measured. And measuring it costs money every month, for ever. So the invisible rules attracted a monthly compliance report, a checkbox on a form, and somebody's Friday afternoon. Every one of those things measures the act of reporting, not the act itself.

A rule whose breach is invisible is not a rule. It is a hope with a number attached, and the number is the cost of hoping. Tanvara was spending 11,482.00 a year on the hoping, and the rules underneath it were being followed 41% of the time.

The first skip is a judgement; the twentieth is the method

All five invisible rules decayed the same way. It is worth naming, because it does not feel like decay from the inside.

The revision ritual was skipped for the first time on a Thursday in October. A mill cut-off was ninety minutes away, and freezing the file properly would have taken eighteen minutes. Skipping it was the right call. Nothing bad happened, and that is the important part: the outcome confirmed the decision. It was skipped again in November, under less pressure. By March it was simply how the sampling room worked, and nobody had ever decided to stop doing it.

That is the shape to watch for in your own factory. A rule does not get abandoned. It gets excepted, and the exceptions do not add up anywhere. There is no meeting where somebody proposes dropping it, so there is never a moment where anybody defends it.

Which is exactly why the grant expiry in lesson 4 works. It is the same device pointed the other way. An expiry manufactures the moment that the drift never provides.

What they did, and the number that came out of it

Each of the five invisible rules got one of three treatments. The choice was made on cost, not on principle.

The two retirements are the ones a governance committee finds hardest, and they were the easiest to justify. Both rules were being followed 41% of the time, and in a year of partial compliance neither had prevented a single incident anybody could name. A rule that is not kept, and costs nothing to abandon, is not protecting you. It is protecting the belief that you are protected. That is worse than no rule at all, because you stop watching.

They spent 10,392.00 a year less and got eighteen points more compliance. The money had been going into measuring the rules rather than into the rules.

The honest closing ratio

The temptation now is to put 29,882.57 over 3,270.00 and report that governance pays back 9.1 times. That is the number that helps you, and it is not the true one, because 3,270.00 is only the rule set. The rest of the course installed other things that cost money to keep.

Plus one-off costs of 1,385.40 — the renumbering, the ownership meeting and one required field.

5,184.30 a year against a defended annual loss of 29,882.57: 5.8 times, not 9.1. Report the 5.8. It survives the second meeting. It survives somebody adding up your own costs for you. And it is the number that lets you go back next year and ask for the next thing.

Build me a ledger of what every governance rule we have costs us to KEEP. Then sort the rules by one question: is breaking the rule visible?

Here are our rules, one per line. For each one I have written who does it, how often, and roughly how many minutes it takes: [PASTE]. Here is how we report compliance today, and where that number comes from: [PASTE].

1. Work out the yearly cost of keeping each rule: how often, times the minutes, times the number of people, times an hourly rate I will give you.
2. Put every rule in one of three groups. Group one: it cannot be skipped, because the system refuses. Group two: it can be skipped, but somebody else sees it straight away. Group three: it can be skipped and nobody sees it. For each rule in group three, say exactly who WOULD find out, and how.
3. Take the compliance figure we report. Tell me what its denominator is, meaning the number it divides by. Then tell me whether a record where the step was skipped completely can even get into that number.
4. For every rule in group three, give me three treatments with their costs: retire it, make the breach visible to a named person, or make it impossible to skip. Recommend one per rule, and say why.
5. Last, add up the TOTAL yearly cost of keeping all of them. Put that against the losses these rules prevent. Do not flatter the ratio by leaving my own costs out.