- Description
- Règle à calcul avec son curseur transparent, gros plan sur les échelles graduées.
- Source
- Slashme · original page
- Licence
- CC0 Public domain
Calculation rules
Simple rules, verifiable by hand, and tested.
In brief
- Level of a dimension: the highest level for which all levels up to it are achieved.
- Score: level + share of the next level.
- Overall: guaranteed (minimum), average (weighted), gap.
The rules are deliberately simple and verifiable by hand. They are also tested automatically.
1. The statuses#
| Status | Value in the score | Counts toward achieving a level? |
|---|---|---|
| In place | 1 | Yes: the only status that allows a level to be achieved |
| Improvable | 0.5 | No: the criterion is partially met |
| To do | 0 | No |
| Not applicable | ignored | Ignored: the criterion counts neither for nor against |
| Not filled in | 0 | Counts as “to do”, and is flagged as not filled in |
2. The level of a dimension#
- A level is achieved when all its applicable criteria are in place. A level whose criteria are all “not applicable” is achieved.
- The level of the dimension is the highest level n such that levels 1 to n are all achieved (0 if level 1 is not). A higher level reached does not make up for a missing lower level.
- The score is this level, plus the completed share of the next level: the average of the values of its applicable criteria. An achieved level 5 is worth 5.
Example: the “Data sources and flows” dimension#
| Level | Criteria | Statuses | Result |
|---|---|---|---|
| 1 | 2 | in place, in place | achieved |
| 2 | 3 | in place, in place, in place | achieved |
| 3 | 3 | in place, improvable, to do | not achieved: (1 + 0.5 + 0) / 3 = 0.5 |
| 4 and 5 | 6 | not filled in | not achieved |
Levels 1 and 2 are achieved: the level of the dimension is 2. The score is 2 + 0.5 = 2.5. The “improvable” and “to do” criteria of level 3 are the next steps.
3. The three overall figures#
- Guaranteed level = the lowest dimension level.
- Average level = Σ (weight × score) / Σ weights.
- Gap = the highest dimension level − the lowest.
Example: four dimensions (excerpt)#
| Dimension | Level | Score | Weight | Weight × score |
|---|---|---|---|---|
| Data sources and flows | 2 | 2.5 | 1 | 2.50 |
| Access and permissions | 3 | 3.67 | 1 | 3.67 |
| Security and compliance | 1 | 1.33 | 2 | 2.67 |
| Testing and fixes | 2 | 2.0 | 1 | 2.00 |
| Total | 5 | 10.83 |
Guaranteed level: 1 (security). Average level: 10.83 / 5 = 2.2. Gap: 3 − 1 = 2. The product is therefore “worth” 2.2 on average, but you can only count on level 1 in every area.
4. Why three figures rather than one#
- The average alone hides weakness: a product at 5 everywhere except 1 in security has a flattering average, and level-1 security.
- The guaranteed level alone is harsh: it does not show the progress of the other dimensions.
- The gap tells whether the product is balanced or has advanced on one side only.
It is the same reasoning as a chain that is only as strong as its weakest link (see the set-based approach).
5. Frequently asked questions#
- Why is “improvable” not enough to achieve a level?
- Because an achieved level must be solid: “improvable” flags a partially met criterion. It counts for half in the score, which reflects progress.
- What about a criterion that is not filled in?
- It counts as “to do” in the calculation, and the tool shows how many criteria remain to be filled in. A fact is not presumed true until it is observed.
- What are the weights for?
- To reflect the priorities of your context in the average level (for example, counting security double). They change neither the dimension levels nor the guaranteed level.
- Can a level be skipped?
- A criterion of a higher level can be met without having achieved the previous one, but the level of the dimension remains that of the last level achieved continuously. Criteria met higher up remain visible and count toward overall progress.
- Do two teams get the same result?
- Only if they apply the criteria the same way. Hence the value of documenting the evidence for each criterion in the comment field.
6. Verification#
The rules above are implemented in a module with no user interface and covered by automated tests: completely empty set (level 0), everything in place (level 5), the detailed example above, improvable level, not applicable, higher level without the lower level, weighting, zero weights, comparison of two assessments. The worked examples on this page are those of the tests.
