Ten questions before deciding, and one before all the others.

An irritant rarely arrives well phrased. It arrives in one sentence, often blunt, dropped between two doors. This grid starts from that sentence exactly as it was said and takes it, in half an hour, through to a costed figure, a decision properly framed and a first dated move.

Fill it in on your own, or with one or more colleagues, before the meeting, and always before opening an artificial intelligence tool. If three questions stay blank, the matter is not ready to be decided. That is already a result.

Download the grid (PDF, in French) See a case handled end to end

How to use it.

Nobody likes being framed on a subject that irritates them, starting with oneself. You first want to put it as you live it. That is why the grid opens on a raw sentence and not on a tick box, and why the questions are taken one at a time, in order. Under pressure, you hold four things in mind, rarely more.

Give the first seven minutes to question zero and the two that follow, without looking for structure. The framing holds once the problem is stated in your own words. Never before.

If the first answer is a solution
Go back to question 1 and ask what that solution would settle. What it would settle is the problem, and it is almost never the one you thought.
If you tell yourself you have no time
Go straight to question 2. The monthly cost of the status quo answers better than any argument about method, including to oneself.
If a question makes you uncomfortable
It is often the fifth, the one about people. You have probably just touched the real subject. Leave it aside for a while, but come back to it before deciding.
If a colleague puts it better than you
Write down their wording and adopt it. A problem stated by the person who lives it is a problem they will defend in front of the team.

Question zero, and the ten that follow.

The examples follow a single case: a fourteen-employee company whose 2004 press brake breaks down too often. Replace them with your own. Print the grid, write by hand, keep it on the table during the discussion.

00

What is the exact sentence that started the matter, and have you rephrased it until everyone agreed?

Write it down without correcting it, with the words and the mood of whoever said it, including if that is you. Rephrase it by naming what weighs, then wait for the “that is exactly it”. Putting a word on a tension lowers its charge, and the discussion can begin.

Example
“That press is ruining my life, I want a new machine before the end of the year.” Rephrased: “What wears me down is no longer being able to promise a customer a date without crossing my fingers.” The right wording came on the second attempt.
01

What is happening, where, since when, and where does it not happen?

Describe the problem without slipping a solution into it. The column of what does not happen is the one most often forgotten. Yet it is the one that points to the cause.

Example
Three press-brake stoppages in September, all on the hydraulic circuit, eleven production hours lost according to the shop-floor log. No stoppage on the shear or the punch. The failures occur on the morning shift, never in the afternoon, a lead nobody had noticed.
02

What does the status quo cost, each month?

Hours lost, euros, late deliveries, regulatory risk, wear on the teams. A costed loss mobilises more than a promised gain, and it sets the real urgency of the subject.

Example
Eleven hours at €380 per hour of downtime, that is €4,180 for September alone. Two deliveries pushed back a week, with no penalty so far.
03

What is the nature of the subject?

Either the answer is known and it only has to be applied. Or it exists and needs an expert or figures to be found. Or nobody can predict it, and only a trial of a few weeks will teach it. That last case is not settled in a meeting.

Example
The answer exists. It requires a hydraulic technician's diagnosis and three figures. No trial needed.
04

What do you expect at the end, a closed question or a bounded trial?

A subject that ends neither in yes or no, nor in a dated trial with its stopping criterion, will be back on the table next quarter.

Example
Should the press be replaced before the year-end shutdown? “Modernising the shop” would have committed nobody.
05

Who decides, who applies, who can block, and by what date?

Name people and a day on the calendar. In an SME, failure rarely comes from the decision-maker. It comes from the one nobody thought to consult: an operator, a banker, an inspector or a customer.

Example
The owner-manager decides alone, on 14 November. The shop supervisor and the two setters apply it. The bank can block at the meeting on 20 November, and so can the supplier, with six weeks' lead time on the spare part.
06

Which options are open, including doing nothing, and which are irreversible?

A single option serves to justify a preference already formed. Set the deciding criterion before looking at the options, otherwise it will bend to them. An irreversible option deserves twice the examination of any other.

Example
Buy new, take on a refurbished second-hand machine, subcontract the bending to a jobbing shop, repair as you go. The two purchases commit five years of lease finance, the other two stay reversible. Where they conflict, cash flow wins over lead time.
07

Which figures do you really have, and who will go and find the others?

Separate the measured, the estimated and the missing. Write missing data rather than an invented order of magnitude, then note a name and a date beside it.

Example
Hourly downtime cost measured at €380. 2027 volume estimated from the order book up to March, missing data beyond. Corrective maintenance for the last twelve months to be pulled from the accounts by the owner-manager, for 7 November.
08

What is off limits in this case?

Data that does not leave the company, commitments nobody can make alone, regulatory obligations, decisions that belong to a third party. To be written down before submitting anything to an AI tool.

Example
Supplier purchase prices do not leave the company. A lease commits the owner-manager alone. No customer name and no price in an online tool.
09

It is six months from now and it has failed. What happened?

Taking failure for granted releases the objections that enthusiasm holds back. The most awkward answer becomes the objection to examine before deciding.

Example
In May, the new press runs at half capacity for want of orders beyond two years, and the instalment weighs on cash. The refurbished second-hand machine cost 40 % less.
10

How will you see that it was the right choice, and what is the first move?

An indicator, a target value, a measurement date set to the rhythm of the subject: three months for a breakdown, a year for a compliance project. Then a sentence in “if… then”, with a name and a date. An intention written in that form turns into action far more often.

Example
Under two hours of downtime a month, read on 31 March, with no overtime added. On 1 October the shop supervisor brings in a hydraulic technician. If the fault is repairable, then the repair is costed before any other option.

An investment, an irritant, a formal notice.

The same questions, applied to three entirely unrelated cases. Here is what question zero, the cost of the status quo, the expected decision and the proof come out as.

A

Machine investment

  • 00“This press is ruining my life.”
  • 02Eleven hours of downtime at €380: €4,180 in September alone.
  • 04Should the press be replaced before the year-end shutdown?
  • 10Under two hours of downtime a month by 31 March. Technician on 1 October.
B

Organisational irritant

  • 00“I spend my mornings sorting emails.”
  • 02Three people, forty-five minutes a day, twenty days: forty-five hours a month.
  • 04Four-week trial: one person does the sorting.
  • 10Under fifteen minutes of sorting a day by 30 November. Rules set on Monday.
C

ICPE compliance

  • 00“The inspectorate is going to come down on us again.”
  • 02Possible penalty payments, and risk of activity being suspended.
  • 04Should the works be commissioned before the regulatory deadline?
  • 10Formal notice lifted. Engineering consultancy within eight days.

What you do with it.

If question zero got no agreement
Stop there and go back to the sentence. A framing built on a misunderstood sentence produces an excellent answer to a question nobody asked.
If question 3 concludes that nobody can predict
The deliverable becomes a trial of a few weeks, with a capped cost and a stopping criterion written in advance. A definitive decision would be premature.
If answer 7 has more gaps than figures
The next action is to go and get the figures. The meeting will wait for them.
If answer 8 is blank
And you are considering using an AI tool on this case, fill it in first. What goes into an online model does not come back.

One page, written within forty-eight hours.

It all fits on a single page. At the top, your rephrased sentence, the one everyone agreed on. Below it, the monthly cost of the status quo, often the first figure ever put on this subject. Then come the nature of the problem, the decision or the trial chosen, the missing figures with their owner and their date, and lastly the first move.

The day you read it again, the sentence dropped in passing has become a case you can take to your banker, your team or your inspector.

Where these questions come from.

None was invented for the occasion. The analysis of what happens and what does not comes from Charles Kepner and Benjamin Tregoe, who formalised it in the 1960s for industrial diagnosis. Stating the problem without a solution opens the Six Sigma DMAIC approach. Sorting by the nature of the subject follows Dave Snowden's Cynefin framework, and the parties to consult follow the logic of ISO 9001 as well as ISO 31000 for risk.

The pre-mortem is due to the psychologist Gary Klein, and the “if… then” formulation to the work of Peter Gollwitzer. Reformulating until agreement comes from motivational interviewing, practised in healthcare for forty years. The findings retained on attention and emotions are those research has replicated. The spectacular promises of neuromanagement have been left aside.

A case taken from the raw sentence to the decision note.

A fourteen-employee company with a cutting and bending shop has to decide the fate of a 2004 press brake that breaks down too often. The case is fictional, built from situations common in engineering and metalwork.

01

The grid, filled in

Thirty-five minutes, ten of them spent on question zero alone, before any structured question. Four costed options, including the status quo, two of them irreversible. Three items of data identified as missing.

02

What was given to the tool

The answers copied out as they stood, plus a short instruction. Question 8 filtered what had to stay inside: no prices, no customer names, no employee names.

03

The decision note

A weighted comparison table, the list of what is missing with who fetches it and by when, a recommendation, the most likely failure scenario, and the condition that would invalidate the recommendation.

04

The result

The note does not decide. It establishes that the decision cannot be made as things stand, names the three missing figures and sets the technician's visit for 1 October. That is the expected outcome of honest scoping.

The three rules given to the tool

Produce no figure that was not supplied, and write missing data instead. Recommend no option without naming its cost. Flag a badly framed decision before dealing with it.

Without them, the tool fills the gaps with plausible orders of magnitude, and a case goes into a meeting carrying figures nobody has measured. A usable document rests on the scoping written by hand before opening the tool, far more than on the tool's power.

When the grid is not enough.

This grid is what I have people do at the start of every assignment. When the subject outgrows one sheet, the scoping audit takes over: half a day on site to set the priorities, the action plan and its schedule.

Let's talk See the scoping audit