The trap: most improvement work is launched as an event — a workshop, a new target, a fresh dashboard — and then fades once attention moves on. Real gains come from a repeating cycle that becomes how the team simply works.
The engine underneath this whole planner is Plan-Do-Check-Act (PDCA) — the Deming cycle. Every BIG, every rhythm, every Gemba walk and check-in is one turn of it. Learn the four moves once and you'll see them everywhere in the tabs that follow.
Pick a small number of Big Important Goals. Name the measure that will move if you succeed. Vague aims ("improve service") can't be improved — specific measures ("comeback rate under 3%") can.
Improvement happens in the daily cadences — huddles, boards, Gemba walks — not in the quarterly review. Do the small things relentlessly, and observe the work where it actually happens.
Bring the number back to the board and the check-in. Did it move? If not, that's not failure — it's the most useful signal you have. Root-cause it rather than working harder at the same thing.
When something works, embed it — into the job card, the checklist, the way you hire and review — so the gain holds. Then start the cycle again on the next constraint. This is what makes it continuous.
A BIG is not a task or a wish — it's an outcome that matters enough to organise a rhythm around. The test: could you put a single number against it, and a single name? If a goal has no measure, you can't tell if you're improving. If it has no owner, everyone assumes someone else has it.
Why three? Focus is the whole point. A team can genuinely drive two or three improvements at once; beyond that, attention thins and nothing moves. Retiring a BIG when it's embedded is a sign of success, not failure.
| Rhythm | Frequency | Who | Purpose / what gets decided |
|---|
It's a Japanese term from lean manufacturing — genba, the actual place where work happens. The principle is simple and hard: you cannot improve a process from a meeting room. You have to stand where it runs and watch it as it really is, not as the report describes it.
The three questions to carry on every walk: What actually happens here? (versus what's supposed to) · What makes this hard? (the friction people have stopped noticing) · What would help? (asked of the person doing the work, not assumed). Go with curiosity, not a clipboard — the moment it feels like an inspection, people perform instead of showing you the truth.
| Date | Where / what observed | What you saw | Action it triggers |
|---|
| Raised by | Idea / problem | Status |
|---|
A checklist lets you fool yourself — half-true feels like done. Scoring forces an honest read, and the total tells you something a list can't: whether the system would survive without you. The lowest-scoring dimensions are exactly where your improvement effort will stall, so they're where to put your next fortnight.
These prompts turn a capable AI into a thinking partner for continuous improvement — not to replace your judgement, but to sharpen it. Each one is built on the 4Ds from SMART AI: give the AI a clear role (Delegation), the real context (Description), a constraint so it doesn't waffle, and ask it to show its reasoning so you can check it (Discernment & Diligence).
Copy one, replace the [bracketed] parts with your own detail, and paste it into whichever AI you use. The italic note under each explains why it's shaped that way — so you learn to write your own.