Illustrative, not a factoryMEAN 2.4 1.4 0.4 0.0 / FLOOR 0
GFT instruments / Performance maturity
Assess capture, cadence, improvement and data-driven practice. The illustrative example has evidence floors of 2, 1, 0 and 0: the overall level is not established.
01 / An instrument for the decisionIllustrative model
System evidence floor
/ 4
Not established
/ 80 answered remaining mean of twenty
Weakest Improvement, question 12: Root-cause work tests a cause before choosing a remedy.
02 / Floor against mean
The mean shows the spread of evidence across a dimension. The floor records its weakest practice. The hatched band is what an average hides. Save A in the instrument to draw a comparison; a shared link carries both answer sets.
03 / Method and judgement boundary
Σ totals + remaining = 80
Point at a bar to read its question and score.
Let q be a whole-number evidence score from zero to four. Each dimension contains five questions. Its total is the sum of its answers. Zero includes unverified claims, so an absent record is drawn as an ochre dot on the baseline, never as a gap in the chart.
A dimension’s mean is its total divided by five. It shapes the diamond and locates unevenness. A larger area is not a measured productivity gain, and a step between two anchors does not imply an equal change in capability, cost or output.
A dimension’s evidence floor is its smallest answer. The hatched band between floor and mean is the part an average would hide. This deliberately conservative convention stops a high score for digital tooling from compensating for a missing capture or action-closure practice.
The system level is the smallest of the four floors. Here capture’s weakest answer is two and cadence’s is one; improvement and data-driven practice each hold a zero. The total is twenty-one, the remaining score fifty-nine, the system floor zero. The evidence set is incomplete. That does not say production is absent, unsafe or commercially unviable.
04 / The floors, read across
1 · Area
mean of twenty 1.05
An average would read 1.
2 · Floors
2 · 1 · 0 · 0
Each dimension keeps its weakest practice.
3 · Point
0 · Not established
A high number elsewhere cannot lift it.
05 / The rollout, as a sequence
The programme design proposes this sequence from project start. The windows are not elapsed time inferred from your score, and not a forecast for an operating factory. A low score in one dimension can coexist with useful work elsewhere. Move the month: the corridor brings that stage’s gate forward.
Source: GFT Cameroon textile programme, performance-management design, section 7, page 16, prepared 31 January 2024. Windows and stage names are from the design; acceptance evidence is this instrument’s illustrative interpretation. The source document is not offered for download.
06 / Defaults and limits
Level names and rollout windows Programme design
Programme design, section 7, page 16. A proposed progression, not measured implementation results. Where it shows: the ladder beside the first diamond and the four gates of the rollout. Changing it would rename a level or move a window. It would not rescore any answer.
Twenty ordinal evidence scores, five per dimension. A dimension's level is its lowest answer. The overall level is the lowest dimension level. Means shape the radar; they do not award maturity. Total score plus remaining score equals 80. This is an illustrative GFT workshop rubric, not a validated certification.
GFT Cameroon textile programme, performance-management design, section 7, page 16 (prepared 31 January 2024). The questionnaire and scoring rules are an illustrative adaptation.
The related paper programme includes Measure Before You Model (WP02), on performance-management systems from the first shift, and The Industrial Data Spine (WP14), on governance and lightweight MLOps. These are related-paper references from the GFT paper register, not claims that reviewed editions are available here.
This instrument cannot inspect records, determine compliance, certify a management system or confirm implementation readiness. A planning aid. Not investment, legal or tax advice.
07 / Carry the level forwardIllustrative model
/ 4
Not established
Next gate: months 1 to 3, stand up capture.
Open: questions 12, 14, 15, 16, 17, 18, 19, 20
A level needs a record behind it.
Raise an open knot to anchor 1 only when the record exists. Use the questionnaire with the people who capture the data, run the reviews and close the actions. Retain the evidence alongside the assessment. Revisit the same operating boundary after changes have had time to settle.
Sharing copies the answer set into a URL on your device; export creates a CSV on your device. Neither submits an assessment to GFT. Links reveal their encoded answers to anyone who receives them.
No scenario saved.