Four questions, four commitments, and numbers nobody made up.
SPEC is opinionated in a small number of places and silent everywhere else. These are the places, stated plainly enough to argue with.
The Board Charter
Everything else in SPEC is negotiated — which roles exist, which measures sit under each pillar, what every target is. Four things are not. They are the same in every business, they are carried by pass-or-fail gates reported beside the score and never averaged into it, and a business that does not hold all four is not running SPEC.
Zero harm, physical or mental.
A safety target derived from what a business has been managing is a statement about how many people it expects to hurt. There is no version of that a board can sign, so this one is not derived from anything.
Never lose great talent.
Somebody good leaving is the most expensive thing that happens to a business and the one thing that never appears on a P&L. It is counted here so it appears somewhere.
Gross profit at the percentage this business needs to be profitable.
This is the one number in the charter SPEC cannot supply. It is a fact about this business’s own costs, and a figure taken from anywhere else is either comfortable enough to be useless or high enough to be ignored.
No breach of the commitments the business has already made.
These are promises the business has already given to somebody else. A target of most of them is not a target, it is a plan to break some.
Three of the four are absolute: the figure is nought and it is never derived from what the business has been managing. Averaging a year of injuries into an achievable target is a statement about how many people you expect to hurt, and there is no version of that a board can sign. Exactly one is earned, and it is the money one — gross profit is the only one of the four that is a fact about your business rather than a line every business holds, so it is the only one SPEC cannot hand you.
A target has to be earned, not agreed
A target gets set in a room, everybody nods, and then it is never checked against anything again. A year later it has either been met every single month — so it was never a target — or missed every single month, so nobody is trying to meet it any more. Both look like a working scorecard from the outside.
So SPEC reads every target against your own closed months: the middle month of the last 12, not the average, so one shutdown or one enormous job cannot reset what everybody is held to.
Below what the business already does
This business already runs at 40.5% across 8 closed months, so the target is met by carrying on exactly as before. A measure that cannot be missed is not measuring anything — it is quietly adding a green light to every month.
Above it, inside what it has shown it can do
40.5% across 8 closed months, with the target at 44%. That is a stretch this business can be held to, because it is above what it does now and inside what it has shown it can do.
Further than it has ever reached
The record says 40.5% and the target says 50%. That is further than a fifth above what this business has actually managed in 8 months, so it will be missed every month — and a measure that is always red stops being read.
The business above runs at 40.5% across 8 closed months, so the sound band is 40.5% to 48.6% — up to a fifth above what it already does. The fifth (20%) and the 6-month minimum are judgements rather than laws, and they are written down so you can disagree with them. Under 6 closed months SPEC declines to judge at all and says so, because a confident answer from evidence that does not support one is worse than no answer.
The figures here are an illustration. What a business in your trade actually runs at is a fact about that business, and SPEC ships the rule that finds it rather than somebody else’s number.
The shallow J curve
Every transformation dips before it climbs. The dip is normally deep, and the reason is almost never the work — it is discovery. Weeks of interviews, spreadsheet archaeology and waiting on month end, during which nothing improves because nothing is visible yet.
Most of a quarter spent working out what is actually wrong. The business is paying for the engagement and for the disruption, and getting a picture in return.
Your systems already hold the picture. When they feed the KPIs, discovery collapses to the day they connect — there is nothing left to find out.
Execution. Getting people to do their jobs properly against numbers everybody can already see. That was always the part that moved anything.
This is a deliberate feature rather than a happy accident, so SPEC measures it instead of claiming it. Inside the product every business sees its own curve: how long its discovery actually lasted against the 70 days a hand-built picture is assumed to take, how deep the dip went, and when it reached the standard — including when that was slower than it should have been. A page that can only ever congratulate the product is not a measurement.
Discovery is the part SPEC shortens, and it is not the only thing that makes a dip deep. A target nobody can meet and a role nobody is in will both hold one open for months, so the same page measures those too, and names what would move each one.
Being exact about which half of the price list this applies to: the collapse is caused by connectors, so it happens on Advanced and not on Basic. Basic is a complete way to run the whole system and it is not a shallow J curve. Saying otherwise would make every other number here worth less.
Conversation boards
Some things about how a business is run cannot be said to the person running it. Not because they are unsayable, but because hearing them from a consultant makes them an accusation, and an accusation gets argued with rather than absorbed.
A conversation board is a purpose-built picture, drawn from what the business already recorded, engineered to raise one specific question. The data does the confronting. Nobody has to.
It shows what was recorded and stops. The reader reaches the conclusion, which is the only version of it anybody acts on.
There is no way to open somebody else’s. The moment a board can be pointed at a person it stops being a mirror and becomes surveillance.
A realisation with no next action is just bad news about yourself. Every board names the one thing that would move it.
The first three ask who decides here, who is actually in the room, and where things wait. None of them is a score, none counts towards anything, and nobody above you sees yours. They are compared against a described way of working and never against another person or another business — a board is exactly where a ranking would be most tempting and most damaging.
What none of this does
No leaderboard, no ranking, no sorting people by score, not even anonymised. Nothing is emailed with your numbers in it. No AI writes a KPI, a target or a score. Nothing recalculates a month you have already closed. If you stop paying it goes read-only and export still works, because a business that has stopped paying still owns its own record.
No card. Nothing bills until you invite somebody in.