Scoring your committee can defend
Invite evaluators, assign by expertise, collect conflict declarations and run as many rounds as the call needs. Rubrics carry weights and descriptors, calibration evens out strict and generous scorers, and the leaderboard shows human, AI and blended views side by side.
- 42
- 5
- 3
- 3
A ranking is only as good as the process behind it. Evaluation is the part of IdeaScore that a committee will be asked about, so every score carries who gave it, against which criterion, in which round, and what the descriptor said at the time.
How evaluation works
Get the right reader on the right application
Invite faculty, mentors and investors by email, collect their areas of expertise and let IdeaScore auto-assign submissions to match, or assign by hand when you know better. Workload is visible per evaluator, so nobody is quietly given forty applications. Conflict-of-interest declarations are collected before any assignment is confirmed, and a declared conflict withdraws that evaluator from the record.
A rubric with descriptors, not a number out of ten
Build criteria with weights and write a descriptor for each band, so a seven means the same thing to a professor and to an investor. Evaluators score in a stepper with the descriptor in front of them, save a draft and come back, and submit when the batch is done. Blind review hides applicant identities where the programme requires it, and the running total is always visible so nobody has to add up.
Correct for the strict reader and the generous one
Two evaluators can read the same application and differ by four points because of how they use a scale, not because of what they thought. Calibration normalises each evaluator's scores against their own distribution, so a ranking reflects judgement rather than scoring habits. It is optional, shown alongside the raw numbers rather than replacing them, and the committee can see both.
Promote forward, round by round
A call is rarely one pass. Screening narrows the pile, a jury round scores what is left against a heavier rubric, and a final round runs on pitch day. Each round carries its own rubric, its own evaluators and its own deadline, and promotion moves a defined number of teams forward with the decision recorded. Earlier rounds stay readable, so a committee can see how a team got to the final.
Human, AI and blended, in the same table
The leaderboard is what the committee reads. Switch between the human score, the IdeaScore AI score and a blended view, and watch the ranking reorder so you can see which teams the two disagree about. Rank one carries the gold marker, movement against the human ranking is shown as a delta, and the whole table exports to the committee report.
Leaderboard
- 1#014246.5 / 50 (93%)—
- 2#012743 / 50 (86%)—
- 3#013941.5 / 50 (83%)—
- 4#011839 / 50 (78%)—
- 5#008337 / 50 (74%)—
Everything in evaluation
Evaluator invites
Workload view
Auto-assign by expertise
Conflict of interest
Deadlines and reminders
Blind review
Weighted rubrics
Draft and submit
Calibration
Multi-round pipelines
Leaderboard views
Score export
Questions committees ask
How does calibration work?
Can evaluators see each other's scores?
What does blind review hide?
How is the blended score calculated?
Can an evaluator change a submitted score?
How many rounds can a programme have?
See IdeaScore run a call with your own rubric
A 30-minute walkthrough with a founder, using your programme's form and criteria.