Prioritisation frameworks fail on factory floors for a boring reason: they ask too many questions. A supervisor with 30 seconds between stops is not going to fill in a weighted matrix with five criteria. They will pick "medium" for everything and the data will be worthless.
Two questions is about the limit of what survives real conditions. Fortunately, two is enough.
The two questions
Impact: when this happens, how much does it hurt? High, medium or low.
Frequency: how often does it happen? High, medium or low.
That is it. No cost estimate, no effort score, no risk multiplier. Cost estimates on the floor are guesses dressed as data, and effort belongs to whoever will do the work, not whoever spotted the problem.
Why the grid beats a ranked list
Score twelve observations this way and they land on a 3×3 grid. High impact and high frequency sits in one corner. Low and low sits in the opposite one.
A ranked list forces a false precision — is item four really more important than item five? A grid does not. It sorts your observations into a small number of buckets and lets you argue only about the ones that matter.
- High impact, high frequency — do these now. There should never be many. If there are nine, you have a systemic problem, not a list of issues.
- High impact, low frequency — these are the ones that hurt when they land. Usually a mitigation or a standard, not a fix.
- Low impact, high frequency — the quiet tax. Individually trivial, collectively expensive. This is where motion waste hides.
- Low impact, low frequency — record them, do not chase them. Their value is as pattern data later.
The quadrant most teams under-use
Low impact, high frequency is where the real money usually sits, and it is the quadrant that gets ignored, because no single instance justifies a project.
An operator walking six extra metres per pallet is a low-impact issue. Two hundred pallets a shift, two shifts, five days — it is a full working day of walking per week, per person. It never gets escalated, because nobody ever had a bad enough day because of it.
The grid makes it visible by counting. That is its main job.
Scoring drifts, and that's fine
One objection comes up every time: different people score differently. True. A supervisor's "high impact" is not a plant manager's.
It matters less than you think, for two reasons. First, you are sorting into three buckets, not thirty — coarse scales are much more robust to disagreement than fine ones. Second, what you actually use the data for is comparison over time within an area, and the drift is roughly consistent within a team.
If you want to reduce it further, the fix is not a longer definition document. It is calibrating out loud: once a month, take five recent issues into the team meeting and ask everyone to score them independently, then discuss the ones where you disagreed. Twenty minutes, and it does more than any rubric.
What to do with the grid on Monday
Pick from the top-right corner until your improvement capacity for the week is full. Then stop. The most common failure is not picking the wrong items — it is picking eleven of them and finishing none.