Not sure where to start?
Start with your business.
You do not need a chosen problem, an AI project or any setup to use this. Five steps, in order, each one question at a time. Pen and paper is enough.
It is the same sequence as the printable worksheets and the assistant, so you can switch between them without starting again. Take one question at a time; no single answer finishes a step, and none of this is an assessment of your company.
1
How the work actually runs today
What it helps you decide: Where in the company an improvement could even belong. Nobody can judge an idea without seeing the work it would land in, and starting here stops the conversation jumping to a tool.
Start with: Which business outcome would you most like to be different?
Then, one at a time, once you have written that answer down:
- How does that work get done today, from start to finish?
- Where does it slow down, get redone, or go wrong?
Where it leads: A described piece of work, in ordinary words, that other possibilities can be compared against.
2
The 0–5 maturity of that work
What it helps you decide: Whether the work is already dependable, or rests on one person's memory. The same idea is a different proposition in work that runs on a documented routine and work nobody has ever observed. A process nobody has watched is unknown — something to go and look at — and never a 0.
Start with: Who keeps that work running today?
Then, one at a time, once you have written that answer down:
- What happens to it when that person is away?
- Is there anything written down that someone else could follow?
- Has anyone actually watched it run recently, or are we guessing?
Where it leads: A level for the work you can actually describe, and a list of what still needs looking at before it can be graded.
3
Two or three possibilities, side by side
What it helps you decide: Which candidates are worth comparing at all. The first idea is rarely the best first test, and a single option cannot be judged — only compared.
Start with: What else in that same area would be worth money if it worked better?
Then, one at a time, once you have written that answer down:
- Which of those would people notice most if it improved?
- Is there one you have been meaning to fix for a while?
Where it leads: A short list of candidates rather than a single favourite.
4
Score them, and say why
What it helps you decide: Which candidate deserves the first small test, and which do not yet. The five criteria make the reasoning visible: a number nobody can explain will not survive the first challenge, and the comparison usually changes the order.
Start with: For that candidate, how big is the outcome if it worked?
Then, one at a time, once you have written that answer down:
- How easily could the people doing the work absorb the change?
- How quickly and cheaply could you tell whether it was working?
- What is this resting on — something measured, or an assumption?
Where it leads: A provisional band for each candidate — a reading for discussion, never an authorisation to spend.
5
Small test, prepare and measure, or defer
What it helps you decide: What actually happens next, and what would count as having learned something. A score is not a decision. Writing down the decision, its limits and the conditions to revisit it is what keeps a small test small and a deferral honest.
Start with: What would you need to see to know whether this was worth doing?
Then, one at a time, once you have written that answer down:
- By when, and over how many cases?
- Do you have that number as it stands today, or would you need to measure it first?
- Who decides what happens when the answer comes back?
Where it leads: A bounded test with an owner, a baseline, a stop rule — or a deliberate wait with a date. The result sends you back to the comparison.
When you have a few possibilities
Use the business map to locate where each one lives, and the worksheets to record what you would need to see before spending anything. The decisions stay with you.