Open an evaluation
Go to Evaluations, pick a published requirement and, if it has one, a tender reference, then choose Open evaluation. Add a supplier bid for each supplier in your directory that responded.
Score each bid
- Score each social value measure from 0 to 100.
- Enter the tendered price, which is the price axis of the award ranking.
- Enter the panel's consensus quality score from 0 to 100. It is never generated by AI and never derived from the social value score.
A bid missing a figure on a weighted criterion is held out of the ranking rather than scored zero, because a missing figure is an unread bid, not a poor one.
The AI pre-read
For each response you can generate a pre-read. It locates evidence in the supplier's response against each published criterion and quotes it word for word, lists what is not obviously addressed, and never scores, ranks or recommends a mark. A quote is only shown when it can be found in the response. If the pre-read is unavailable, read the response yourself and score as normal. Nothing is blocked.
The AI consistency check
Run consistency check comments on how consistently the panel has scored and what evidence it relied on. It is advisory and never changes a score. When AI is unavailable, or its draft is rejected, a rule based check runs instead and says so.
Limits to keep in mind
- The panel evaluates and scores every response. The AI is a reading aid.
- The weightings are checked against the lock taken at publication, and any change is reported above the ranking.
- Move the evaluation on with Send to moderation and Mark complete.