Assessments
Assessment library, create and edit flows, categories, results, journey linking, and AI scoring of text responses.
Assessments are timed or untimed evaluations — multiple choice, free text, case studies — used standalone or inside journeys. Recruiters manage assessments from the dashboard and link them via journey Assessment nodes.
Key routes
| Action | Path |
|---|---|
| My assessments | /dashboard/assessments |
| Create assessment | /dashboard/assessments/create |
| Assessment detail | /dashboard/assessments/[id] |
| Edit assessment | /dashboard/assessments/[id]/edit |
| Results (all candidates) | /dashboard/assessments/[id]/results |
| Individual result | /dashboard/assessments/[id]/results/[candidateId] |
| Template library | /dashboard/assessments/library |
| Assessment analytics | /dashboard/analytics/assessments |
Assessment library
Dashboard → Assessments → Template Library (/dashboard/assessments/library) browses pre-built assessments by category, difficulty, and industry.
Workflow:
- Filter and preview templates
- Clone to your org — creates an editable copy under My assessments
- Customize questions, time limits, and category weights
- Publish when ready
Cloned assessments are independent — library updates do not overwrite your copy.
Create and edit
Create flow
/dashboard/assessments/create opens the assessment builder:
| Setting | Description |
|---|---|
| Title & description | Shown to candidates at start |
| Time limit | Optional global or per-section cap |
| Passing score | Threshold for score-gate branching |
| Categories | Group questions; each category can have a scoring weight |
| Questions | From question library or inline |
Edit flow
/dashboard/assessments/[id]/edit — same builder for drafts and published assessments. Structural changes to live assessments affect new attempts only; completed results remain tied to the version taken.
Publish and archive
From /dashboard/assessments/[id]:
- Publish — makes the assessment assignable in journeys and invitable standalone
- Archive — hides from new assignments; preserves historical results
Categories and weights
Questions belong to categories (e.g. Technical, Problem Solving, Communication). Each category has a scoring weight used when computing the overall assessment score:
overallScore = weightedAverage(categoryScores, categoryWeights)Configure weights on the assessment edit screen. Categories with zero weight still display in results but do not affect ranking.
Standalone invites
From the assessment detail page, invite candidates directly (outside a journey):
- Enter email, optional name, message
- Manage invitations — resend, cancel, grant extra attempts
- Results appear under
/dashboard/assessments/[id]/results
Results
Aggregate results
/dashboard/assessments/[id]/results lists all completions with scores, completion time, and pass/fail status. Filter and export from this view.
Per-candidate drill-down
/dashboard/assessments/[id]/results/[candidateId] shows item-level responses, category breakdown, and AI analysis for text items.
Journey-embedded assessments also surface on:
/dashboard/journey-report/[sessionId]/[candidateId]/dashboard/reports/[journeyId]/[candidateId]/assessment/[assessmentId]
Linking to journeys
- Open journey editor:
/dashboard/journeys/[id]/edit - Add an Assessment node
- Select a published assessment from the dropdown
- Configure completion rules (must pass threshold, required vs optional)
- Connect pass/fail edges if using score gates downstream
Candidates see the assessment inline in the journey — no separate login. Progress syncs to the journey session overall score.
AI scoring of text responses
Free-text and long-answer items can receive AI-generated scores when scoring criteria are configured:
| Input | Scoring behavior |
|---|---|
| Multiple choice | Deterministic — correct option matching |
| Numeric / structured | Rule-based validation |
| Free text / essay | AI rubric scoring against evaluation criteria |
Authenticity analysis
Text responses may run through authenticity analysis — signals for generic AI-generated content, template copying, and response consistency. Results attach to authenticity_analysis records and surface on assessment reports alongside the content score.
Authenticity is a signal, not an automatic fail. Review low-authenticity responses in context.
Improving AI accuracy
For each question used in assessments:
- Open
/dashboard/questions/[id] - Configure Scoring criteria (basic, weighted rubric, or AI-generated rubric)
- Add ideal response exemplars where helpful
See Questions & scoring for rubric authoring.
Best practices
- Pilot with library clones — faster than building from scratch
- Balance category weights — avoid 100% weight on a single category unless intentional
- Set time limits — reduces abandonment on mobile
- Use journeys for hiring — standalone invites suit screening campaigns; journeys keep evidence unified
- Review text AI scores — spot-check first cohort before trusting thresholds for auto-reject gates
Related guides
- Journey builder — assessment node placement
- Questions & scoring — shared question library
- Review queue — how assessment scores feed composite ranking