Importance Levels for Criteria + New Criteria Model
Set Must have, Preferred, or Nice to have. Scoring and the builder respect these levels. The model now writes clearer, resume‑checkable criteria.
Quick summary
You can now set each criterion to Must have, Preferred, or Nice to have. Scoring and the builder respect this nuance: must‑haves drive clear explanations, while preferences nudge rather than block. We also retrained the model to write shorter, resume‑checkable criteria with fewer duplicates.
Why this matters
Before, criteria were just Required or Optional. That forced a choice between being too strict or too loose, and it didn’t reflect the nuance that comes out of kickoff and calibration. With Must have, Preferred, and Nice to have, you can encode non‑negotiables and still surface signal from preferences without rejecting good candidates.
How scoring behaves
- Missing must‑haves are highlighted in the decision with evidence.
- Preferred and nice‑to‑have add confidence; they don’t block.
- When profiles look similar, preferences act as tie‑breakers so the right 10–20 rise first.
Practical examples
- Location preference (e.g., LA) as Nice to have bubbles locals without penalizing others.
- Work authorization or visa as Must have gives clear, explainable rejections when missing.
- Product manager experience with hands on engineering as Nice to have, and experience in research-heavy product development as Preferred.
Model improvements for screening
The builder now proposes criteria that are easy to verify on a resume—short, specific titles (customer segment, deal size, ownership) with fewer repeats—so reviews are faster and decisions are easier to explain.