Score your company on the six dimensions we assess in an engagement.
This is the scoring rubric from our AI Readiness & Workflow Audit: six dimensions, 0 to 5, anchored to what one specific workflow needs. Working through it honestly takes about twenty minutes. Nothing is sent anywhere and there is no form.
How to use the rubric
Pick one workflow
Score against what one specific workflow needs, not against an abstract maturity model. The same data layer can be a 2.0 for a workflow that needs linked history and a 3.5 for one that does not.
Score what you can show
A number you cannot support with a document, a record count, a contract clause, or something you watched happen is a guess. Note the evidence beside the score before you move on.
Half points
The anchors are written at the whole numbers. Use a half point when your evidence sits between two of them.
The realistic range
0 and 5 are theoretical endpoints. Most SMEs land between 2.0 and 4.0. A 4.5 needs exceptional evidence, and a 1.0 usually means the workflow is the wrong one to start with.
The six dimensions
Always these six, always in this order. Read the anchors, select the one your evidence supports, and add the half point if the evidence sits above it.
The scoring runs in your browser. With JavaScript switched off, all six dimensions are listed below as a plain reference instead of one at a time. The overall score is the mean of the six dimension scores; the bands are listed under Reading the result.
0 of 6 dimensions scored. Score all six to see the overall reading.
Reading the result
The overall score is the plain mean of the six dimension scores, rounded to one decimal. There is no weighting and no adjustment for judgment. If the mean looks wrong, the fix is in a dimension score you can defend, never in the aggregate.
Below the usual range
Lower than we normally see in an operating SME. Before scoring further, check whether the workflow you picked is the right place to start.
Groundwork before build
The typical position when the data or the governance layer has not been touched yet. Nothing here is a blocker, but the weakest dimensions have to close before a build starts.
Conditionally ready
Most of what a build needs already exists. The named gaps decide the sequence rather than whether to proceed.
Ahead of the usual range
Above where most SMEs land. Check that every score above 4 rests on evidence rather than on familiarity with your own systems.
The lowest dimension decides the plan
A verdict that names only strengths, or hides the constraint behind the aggregate number, is not usable. If the first thing you do does not visibly attack the weakest dimension, either the score is wrong or the plan is.
A dimension below 3.0 is only useful once it becomes a work item.
Every dimension under 3.0 has to reappear as something with a name, a date, and a test. Four things make a work item well formed.
- 01
A concrete deliverable: one indexed store, one filing convention, a signed policy. Improve data quality is an activity, not a deliverable.
- 02
A time anchor in the first phase, stated in weeks rather than quarters.
- 03
An owner or a budget, whichever the item actually needs. A precondition with neither attached does not happen.
- 04
A test that closes it, written down before the work starts: field accuracy on a back-test, a signature on a policy, one convention visibly in use.
Then write down the three things that must be true before a build starts. Three, not ten. If a fourth feels unavoidable, the scope is too wide for a first phase.
Limits of a self-assessment
A self-score records what you believe about your own operation. A score that holds up records what someone checked: a case retrieved from the archive, two people's output compared side by side, a contract read rather than summarized, a rate agreed with whoever owns the budget.
Corrections almost always run downward, and usually in data foundations and governance, where the distance between the tidy version and the observed one is widest.
The cheapest improvement is a second scorer. Have someone who does not own the process score it independently, then reconcile. Any dimension where the two of you differ by more than half a point is worth resolving before anything gets built.
How the same rubric is used in an audit
In an AI Readiness & Workflow Audit the six scores rest on observed evidence rather than self-report, and they are reviewed with management before the report is issued. Every dimension under 3.0 is carried into a first phase with dates, an owner, and a budget, and the report names the binding constraint in the client's own words.
The three sample reports on the examples page are complete and free to download. Each one scores these six dimensions end to end, so the standard is visible before anyone commissions anything.
This rubric is public on purpose. A company that scores itself honestly can act on the result with its own IT team or with any competent partner, and that is the same principle the reports are written on.
If you have scored yourself and want a second read on it, write to info@leitspur.ai.