Skip to main content

Documentation Index

Fetch the complete documentation index at: https://nova.dweet.com/docs/llms.txt

Use this file to discover all available pages before exploring further.

Nova scores each candidate from 1-10 for one role at a time. A high score means the available evidence appears strongly aligned with the criteria for that role. The score is not a percentage, a count of passed criteria, or a hiring decision. It is a screening assessment based on the role, criteria, evidence quality, and remaining uncertainty.

Score Scale

ScoreHow to read it
10Exceptional evidence alignment. Clear evidence across the role’s core needs, plus unusually strong signals.
9Outstanding evidence alignment. Strong evidence for the core requirements and few meaningful concerns.
8Strong evidence alignment. Resume-verifiable essentials are supported by evidence. Any remaining gaps are minor or better checked in interview.
7Promising evidence alignment. Relevant and useful to review, but there is real uncertainty about at least one important requirement.
6Mixed evidence alignment. Some useful evidence, but important gaps or weaker alignment.
5Borderline evidence alignment. Possible relevance, but the evidence is thin.
3-4Weak evidence alignment. Significant gaps, contradicted requirements, or limited evidence for the role.
1-2Very limited evidence alignment. Very little relevant evidence for the role.

The 7 Vs 8 Line

Most score calibration questions come from the difference between 7 and 8. An 8 means the main resume-verifiable requirements are evidenced. The candidate may still need normal interview checks, but Nova found enough evidence to support the core screening case. A 7 means the candidate is interesting, but something important is still uncertain from the resume or application. That could be a missing skill signal, unclear seniority, thin domain evidence, or an application answer that does not fully resolve a requirement.
Interview-only unknowns should not automatically keep a strong candidate below 8. Resume-level uncertainty about an important requirement usually should.

Evidence Status

When Nova shows criteria analysis, each criterion can be marked:
StatusMeaning
PassThe evidence supports the requirement. Equivalent experience can count when it reasonably satisfies the criterion.
PartialThe evidence is related, thin, indirect, or missing enough detail to be confident.
FailThe evidence contradicts the requirement or verifies an exclusion.
Missing information is usually Partial, not Fail. A resume that does not mention a requirement creates uncertainty. A resume that clearly shows the opposite creates a concern.

How Must Have Criteria Work

Must have criteria matter more than Preferred or Nice to have criteria, but they are not automatic score caps. Nova assesses whether the available evidence supports continued review. A candidate with one absent Must have signal may still score around 7 or 8 if the missing detail is plausible to verify. A candidate who clearly contradicts a Must have requirement usually scores much lower. Use Must have only for requirements that are genuinely necessary for the role, such as:
  • Legal or regulatory eligibility.
  • The experience that makes someone viable for the role, such as production backend experience for a backend engineer or current registration for a nurse.
  • A practical requirement the job cannot work without.
  • A deal-breaker you have explicitly confirmed.
Use Preferred for strong differentiators. Use Nice to have for bonus signals.

What The Assessment Tells You

  • Verdict: The headline screening summary, not a decision.
  • Strengths: The strongest evidence Nova found.
  • Concerns: Missing, thin, or contradicted evidence.
  • Interview Focus: The questions to verify next.
Talent Search uses a shorter review: a summary for every scored candidate, plus strengths and concerns for candidates scored 7 or higher. Concern labels usually mean:
  • not evidenced: Nova did not find evidence for the requirement.
  • contradicted by: Nova found evidence pointing the other way.
  • unverified: The point should be checked with the candidate or hiring team.

When A Score Feels Wrong

What you seeLikely causeWhat to do
Most candidates score 9-10Criteria are too loose or too generic.Add role-specific requirements and make the key differentiators explicit.
Almost no candidates score above 5Criteria are too strict or too many items are marked Must have.Move secondary requirements to Preferred or Nice to have.
A candidate you expected to review scored lowTheir resume may not show the evidence your criteria ask for.Check the concerns, then decide whether to edit the criterion or verify the point in interview.
Similar candidates score differentlyCriteria may be vague or depend on details one resume includes and another omits.Rewrite the criterion around observable evidence.
Everyone clusters around 5-7The criteria may describe broad relevance instead of clear differentiators.Add the requirements that separate acceptable from strong candidates.
Tool names drive too much of the scoreThe criterion may be too vendor-specific.Reframe around the capability unless that exact tool is required.
After editing criteria, future candidates use the updated criteria. If reprocessing is enabled for your workspace, a Nova admin can use Reprocess Applications from the job actions menu to update existing scores.

Good Review Workflow

  1. Read the verdict first.
  2. Check the strengths and concerns.
  3. Open criteria analysis when available.
  4. Decide whether the concern is real, missing from the resume, or better checked in interview.
  5. Edit criteria only when the pattern repeats across candidates.
Do not use the score alone to make or change an employment decision. Review the underlying evidence before taking action. For adverse-impact checks, see Bias Testing Methodology.
There is no separate numeric weighting setup. Use Must have, Preferred, and Nice to have to show importance.
Absence is uncertainty. Nova treats a missing detail differently from evidence that proves the candidate does not meet the requirement.
The passed criteria may be lower importance, generic, or offset by a serious concern. The score reflects overall evidence alignment.
Nova can use a LinkedIn profile URL as a fallback when enabled. If no usable evidence is available, the candidate is skipped and not charged.

Configure Criteria

Write criteria that produce useful scores

AI Candidate Scoring

See how scoring and ranking work