Skip to content
KatafactsBeta

Measurement

Metric definition sheet

Metric definition sheet is licensed CC BY 4.0. Attribution: Katafacts (katafacts.com).

Customise with AISkill file ↓

1 · What it is

What it is

A metric definition sheet is a real, checkable answer to 'what does this number actually mean, and who says so' — one formula, one data source, one owner per metric name. The problem it exists to solve is common and well documented: two teams both report 'on-time delivery' or 'active users,' each computing it a different way, and both numbers are defensible on their own terms — until someone tries to reconcile them. This sheet doesn't invent a resolution when none exists; a metric with no agreed formula or no real owner is flagged plainly, not smoothed over.

2 · When to use it

When to use it — and when not to

Use it when

  • Two teams have quietly started reporting different numbers for a metric with the same name, and nobody has actually reconciled why.
  • A dashboard or huddle board is about to go live and you want every metric on it traceable to a real formula and a real owner, not an assumption.
  • You want to know honestly which metrics in active use still have no agreed definition, rather than assuming they're all settled.

Not when

  • You don't actually know how a metric is computed today — go find out from whoever owns the system first; this sheet records real facts, it doesn't invent a formula that sounds plausible.
  • The metric is genuinely one-off or exploratory, not something two or more people will reference by name going forward — the overhead isn't worth it yet.

3 · How to fill it in

How to fill it in

Scope area
What part of the business do these metric definitions cover?
Metrics
For each metric: the formula, data source, owner, refresh cadence, and unit you actually use today — your own facts, never guessed. Leave any of these blank if it's genuinely not agreed yet.
Narrative
Drafted from the metrics actually defined, naming which ones still have a real gap.

4 · What good looks like

What good looks like

The example below continues the Line 2 paint-booth storyline running through this catalogue's A3, scoping canvas, and FMEA examples — two metrics are fully specified against real systems already in use, and the third, on-time delivery rate, honestly names a real, unresolved conflict between Scheduling and Quality rather than picking a winner.

Same example, as a downloadable xlsx workbook.

Download .xlsx

Metric definition sheet

Line 2 — metric definitions

Line 2 production and changeover metrics

Priya Nair, Line Supervisor · 2026-03-10

Scope area

Line 2 production and changeover metrics

Metrics

No agreed formula: On-time delivery rate

First pass yield — Line 2 final inspection

The share of units that pass Line 2's final inspection the first time, with no rework or scrap, counted at the final-inspection station specifically — not a whole-line figure.

Formula: (Units passing final inspection without rework) ÷ (Total units entering final inspection) × 100

Source: MES final-inspection station log, not the shift-end paper tally sheet · Owner: Priya Nair, Line Supervisor

Known limitations: Excludes units scrapped before reaching final inspection — a true whole-line yield figure needs the first-pass-yield tracker's full stage-by-stage view, not this single-station number.

Changeover time — Line 2 paint booth

Elapsed time from the last good unit of the previous color to the first good unit of the new color, measured at the booth itself.

Formula: Timestamp of first good unit (new color) − timestamp of last good unit (previous color)

Source: Line 2 booth changeover log — currently a paper form, transcribed weekly, not yet automated · Owner: Marcus Webb, Paint Booth Lead

Known limitations: Manually transcribed, so a missed log entry understates true changeover frequency. This is the same procedure the FMEA on the Line 2 changeover already flagged — the cleaning-step timer, not this metric itself, is the real risk behind a rushed changeover.

On-time delivery rate

No formula

No single agreed definition exists yet. Scheduling currently counts an order as on-time if it ships by the promise date. Quality counts it as on-time only if it also passes final inspection by that date — a stricter bar. The two teams are reporting different numbers for the same name.

Formula: Not yet agreed

Source: — · Owner: Dana Ruiz, Scheduling

Known limitations: Scheduling and Quality compute this differently today, and this sheet doesn't manufacture a false resolution. Dana Ruiz owns getting the two teams to agree on one formula — until that happens, neither team's number should be trusted against the other's.

Narrative

Two of three metrics are fully specified with a real formula, source, and owner — first pass yield and changeover time both trace to real systems already in use on Line 2, and changeover time's own known limitation connects directly to the FMEA's cleaning-step finding on the same procedure. On-time delivery rate is the one still genuinely contested: Scheduling and Quality compute it differently today, each defensible on its own terms, and this sheet names that conflict plainly instead of picking a winner on its own. Dana Ruiz owns resolving it — until she does, this is one metric name with two real numbers behind it, not one.

5 · Common mistakes

Common mistakes

  • Writing a formula that sounds precise but was never actually confirmed against the source system.

    A confident-looking formula with no real basis is worse than an honest gap — it lets two teams both trust their own number right up until the day someone tries to reconcile them.

  • Leaving a metric unowned because 'everyone kind of owns it.'

    A definition nobody owns is a definition everyone quietly forks — a real owner is what keeps the formula from silently drifting as systems change.

  • Resolving a genuine cross-team conflict unilaterally just to make the sheet look complete.

    Picking a winner without the other team's buy-in doesn't resolve the conflict, it just hides it — naming the conflict plainly is what actually gets it fixed.

6 · What it connects to

What it connects to

upstream

    downstream

    • KPI tree

      A driver metric in a KPI tree is only as trustworthy as this sheet's own definition behind it — the tree tracks the number, this sheet is what makes the number mean one specific thing.

    7 · Where AI helps

    Where AI helps

    Judgement — stays yours

    • Deciding whether a metric's formula is actually well-specified or just sounds precise
    • Deciding which unresolved gap to prioritize closing first

    Analysis — AI helps

    • Tightening raw definition and known-limitations notes into specific, scannable language
    • Drafting the narrative naming which metrics are fully specified and which still have a real gap

    Drudgery — automated

    • Identifying which metrics are missing a formula or an owner
    • Exporting to xlsx in the house format

    9 · Rate this kata

    Rate this kata