Defining the problem with numbers and boundaries

Two ways to define a problem badly

A problem definition can fail in either direction. Too broad — "Lakeshore has a quality problem" — names no gap, no number, no place, and so cannot be solved: every alternative addresses it and none can be shown to fix it. Too narrow — "the press's feed roller keeps failing" — pre-decides the cause and, with it, the fix; if the roller is replaced and the imaging unit fails next, the definition excluded the real answer. Between the two is a definition with a number (the size of the gap) and a boundary (where, who, when it occurs).

Worked example — three drafts for Lakeshore

1. "Lakeshore has a quality problem." Too broad. Quality of what — print, service, delivery? Measured how? A memo answering this could recommend anything.
2. "The press's feed roller fails." Too narrow. It is one cause of some breakdowns, and it commits the analysis to a roller before the breakdown log has been read.
3. "Large-format on-time delivery fell twelve points, from 96 % to 84 %, in twelve months, concentrated in the three largest accounts." A number (12 points; from 96 to 84), a period (twelve months), a place (large-format work — $310,000 of revenue), and a who (the three key accounts, $180,000 at 42 %). It is solvable: a fix works if on-time delivery in that work returns toward 96 %. And it is not the cause: the press has not been mentioned, so nothing is excluded.

The third draft is what Smith's "define the problem" step means by a definition — a percentage drop, localised — and it is what the first line of R-psp-2 (each issue stated in one sentence with a fact) is looking for.

The solution-in-disguise

A third failure hides among the broad ones: the problem stated as a solution. "We need a new press", "we need more customers", "we need better software" are not problems — they are answers, written before the question, and they are unbounded (more customers where? how many? by when?). Each one collapses stage 3 to a single alternative. Restate them as gaps: not "we need more customers" but "Saturday revenue is 18 % below the budgeted level for three months running".

Rewrite two for the clinic

• "The clinic's scheduling is a mess." Too broad. Rewrite: about 12 % of booked appointments — roughly one slot in eight — go unkept, across both dentists' books, and two double-bookings occurred last month. (Number, who, when. The 12 % is an estimate; say so.)
• "The receptionists forget to call patients the day before." Too narrow, and it blames a person for what is a process without reminders. Rewrite: no reminder reaches a patient between booking and the appointment, for any of the roughly one-in-eight who do not attend. The cause (paper, no reminders) is close by, but the definition leaves the fix open — a reminder could be a call, a text, or a booking system.

Write one cold

> A Waterloo tutoring centre's evening sessions have had empty seats since January; the manager says "we need better marketing".

A definition with a number and a boundary: evening-session attendance has run at [X] % of capacity since January, against a break-even of [Y] %, in the Monday-to-Thursday slots. You do not have X and Y — which is the point: a good definition tells you what to ask for. "Better marketing" told you nothing except the manager's favourite answer.