Criteria Scoring

Criteria scoring is how Optible turns an application into a ranked score. Your rubric is organised into two layers: criteria groups and criteria.

The two layers

Criteria groups — broad categories that reflect the dimensions of your grant, such as Financial Viability, Community Impact, or Project Readiness. Each group gets its own score.
Criteria — specific things being assessed within each group, such as "Has audited accounts" or "Positive cashflow." Each criterion gets its own score, which feeds into the group.

How scores roll up

Scores flow from criteria up to groups and then to an overall rank:

Criterion score — Optible assesses each criterion and produces a score from 0 to 100
Group score — weighted average of all criteria scores within that group
Overall rank — weighted average of all group scores; this is the number shown in the ranked list

Warnings

You can configure warnings tied to scores. For example: "If Financial Viability drops below 60, flag the application as Weak financials." Warnings are reconciled after every scoring run — new ones are added and old ones that no longer apply are removed.

There are three kinds of warnings:

Missing document — a required document was not uploaded. Critical severity.
Failed document check — a document was uploaded but failed one or more requirements.
Low score flag — a criterion or group scored below the configured threshold.

Reviewer overrides

Every score is a starting point, not the final word. Reviewers can open any criterion and override the score. When they do:

Their score replaces the original in the ranked list
On a re-run, their override stays in place — only the original assessment gets rebuilt
Their comment appears alongside the original reason so the audit trail is complete

Note on scores

Criterion scores must be greater than zero. A criterion cannot be saved with a score of 0.

Testing a rubric before going live

You can run a rubric against real applications in test mode before activating it. Test scores appear separately and do not touch live data. If the results look good, promote the rubric. If not, tweak and run again. See Criteria Versions for more detail.