Score the evidence you have, and show what remains unknown. Do not convert a persuasive assertion into an observed capability.
Use a 0–4 scale, plus U
| Rating | Meaning | Evidence treatment |
|---|---|---|
| 0 | Evidence demonstrates that the criterion is not met. | Record the specific observed or documented failure. A gate breach takes precedence. |
| 1 | Material weakness. | Some relevant capability exists, but it falls well short of the agreed requirement. |
| 2 | Adequate but incomplete. | The essential approach exists, with substantive limitations or substantial assistance. |
| 3 | Strong evidence of meeting the requirement. | The relevant practice or arrangement is demonstrated and fits this learner or brief. |
| 4 | Strong, repeatable evidence. | For behavioral criteria, require more than one occasion, including a later retention or transfer check where relevant. For operational criteria, require evidence that the arrangement works, not only that it is written down. |
| U | Not established. | Missing, irrelevant, stale or assertion-only evidence. This is not a finding of poor quality. |
Weighted points = criterion weight × rating ÷ 4.
Maintain three outputs:
- Supported score: total weighted points established so far, out of 100.
- Evidence coverage: the percentage of total weight with a defensible numeric rating.
- Unresolved weight: the total weight marked U.
For arithmetic only, U contributes no supported points. Display it as U, not as a quality rating of zero. If supported points are 62 and unresolved weight is 20, report 62 supported points; 80% coverage; unresolved scoring range 62–82. The upper endpoint is a mathematical possibility, not an expected result or purchasing recommendation.
Apply evidence limits
- A supplier biography can establish what is claimed, but not that the assigned mentor teaches well.
- A document can establish a contractual commitment; it cannot establish that sessions follow it.
- One observed lesson can justify up to 3 on a behavioral criterion, not 4.
- A sample from another instructor does not establish the assigned instructor’s teaching ability.
- An anonymized report can demonstrate reporting format; it cannot establish this child’s progress.
- Retain the source, date, mentor identity and scope of every finding. Recheck evidence after a mentor, tool or project changes.
What score is sufficient to buy?
These are desk decision rules, not statistically validated cutoffs. Apply them to a completed assessment, not a brochure-only comparison.
| Result | Recommended decision |
|---|---|
| Any applicable gate unresolved | Hold the affected activity. An adult-only supplier interview may continue, but do not treat uncertainty as clearance for child contact, tool use or publication. |
| Any gate failed | Do not proceed as proposed. Consider a revised arrangement only after the issue is resolved and rechecked. |
| 85–100 | Strong candidate for the next bounded block, subject to the minimum scores below and learner agreement. |
| 75–84 | Credible candidate; approve a bounded block with explicit improvement targets. |
| 65–74 | Revise the plan or run a narrowly defined reset before a longer commitment. |
| Below 65 | Do not select for a longer engagement on the current evidence. Diagnose the mismatch before replacing or rescoping. |
| Incomplete coverage | Use the missing evidence to design the trial. Do not rank against a fully observed candidate as if the totals were equivalent. |
For a longer commitment, require all nine criteria to be established, no criterion below 2, and criteria 1, 2, 3 and 5 at least 3. This prevents excellent administration or engineering from compensating for weak learning, teaching or rapport.
A trial can be approved before the full score exists, provided the applicable gates, trial scope and commercial terms are clear. A small difference between two completed scores should prompt a closer look at fit—not a claim of precise superiority.
Can the same scorecard judge a six-week mentorship plan?
Yes, but label it proposal-stage, not delivery-verified. Score what the documents establish about scope, reporting commitments, continuity and costs; mark unobserved teaching, rapport and learner outcomes U.
A good plan should make the next verification step easy. It should not receive assumed learning points merely because it promises an impressive final application.
Sources and context
This page sets out the desk’s purchasing framework. What are the limits of this research? → Sources →