Configuring Scoring Criteria

Create, edit, and refine the criteria Nova uses to evaluate candidates for each role.

Criteria are the rules Nova scores against. Nova evaluates every candidate against your criteria consistently across your entire pipeline.

Criteria are configured per job. You need at least 1 before Nova can score.

How It Works

From the Jobs page, click Add Criteria to open the Criteria Builder. Nova drafts criteria for you based on conversation.

1. Provide context

   Nova asks for context beyond the job description: intake call notes, hiring manager requirements, deal-breakers, target companies. Skip if the job description covers it.

2. Answer clarification questions

   Nova generates 5-7 questions with multiple-choice options. Pick options and see criteria diffs side-by-side. Skip any and refine later.

3. Review and refine with the AI assistant

   Nova generates criteria (typically 5-8) with importance levels. The screen splits: criteria panel on the right, AI assistant on the left. The assistant flags anything worth adjusting and offers calibration.

The AI Assistant

The assistant refines criteria in plain language. It flags issues proactively: criteria hard to verify from a resume, overlapping requirements, or must-haves that are too strict.

Examples of things you can say:

• “Add a criterion for 5+ years of Python experience”
• “Remove the startup experience requirement”
• “Consolidate the first three criteria into one”
• “Make the leadership criterion a must-have”
• “These criteria feel too strict, can you loosen them?”

Changes appear as interactive cards. Approve or reject. Approved changes save immediately.

Select criteria using the checkboxes before typing. The assistant knows which criteria you’re referring to.

Importance Levels

Level	What it means
Must have	Required. Missing this disqualifies.
Preferred	Important, not a blocker.
Nice to have	Adds value but optional.

Click the dropdown on any criterion to change importance. Saves immediately.

Editing Criteria

All edits save automatically:

• Edit text: Click to edit inline.
• Change importance: Click the importance dropdown.
• Delete: Trash icon (minimum 1 required).
• Add new: Add New button.
• Add from library: Pull in saved criteria.
• Refine with AI: Select criteria, describe what you want in chat.

Calibration

Test your criteria against real candidates before enabling scoring.

Running calibration

Ask the assistant to calibrate. Nova samples candidates from your ATS and scores them in the conversation.

Results

Each candidate shows an overall score (0-10), criteria passed out of total, and a source badge. Click to expand the full per-criterion breakdown with evidence.

The assistant analyzes the batch and suggests criteria changes. If a strong candidate fails due to wording, it recommends a rewrite.

Approving changes and re-scoring

Approve a criteria edit from the interactive card. Nova automatically re-scores the same candidates. Re-scored cards show deltas (e.g., “was 4 (+3.0)”). Multiple rounds supported.

Calibration is a test run. It doesn’t affect your live pipeline.

Starting Nova

When criteria are ready, start screening from the builder. The assistant offers this when criteria look good, or ask (“start Nova”).

Criteria Library

Save criteria that apply across roles to a company-wide library. Click the star icon on any criterion. All team members share the same library.

Good library candidates: tenure requirements, remote work experience, communication standards, startup experience.

To add library criteria to a job, click Add From Library, select, and click Add Selected.

Reprocessing After Changes

Existing scores aren’t automatically recalculated when you update criteria. Use Reprocess Applications from the job’s actions menu.

Reprocessing replaces all existing scores with new ones.

Scoring Input Sources

Source	Details
Resume/CV	Primary input
Job description	Context and terminology
Your criteria	The rules Nova scores against
Cover letters and portfolios	Considered when available
Application question responses	Included by default. Exclude in Company Settings.
LinkedIn profile	Fallback when no resume attached (if enabled)

No external data sources influence scores unless Talent Experience Research is explicitly enabled.

Can I weight criteria differently?

No numeric weights. Use the 3 importance levels. Must-haves weigh more heavily than nice-to-haves.

How many criteria should I set?

5-8 works well for most jobs. Fewer gives Nova less to differentiate. Too many penalizes candidates who are strong overall but miss minor points.

What if I change criteria mid-search?

Changes apply to future scores. Use Reprocess Applications to update existing ones.

How specific should criteria be?

Specific enough to verify from a resume, broad enough to avoid excluding equivalent experience. “3+ years in backend development with Python or Go” beats “5 years Python experience at a Series B+ company.”

Can I redo the initial setup questions?

No, but tell the assistant what changed. “I said 5 years but 3 is fine” updates the relevant criteria.

Understanding Scores

How to read and interpret Nova assessments

Enabling Nova

Start scoring on your job postings