🎨 Understanding Stoplight Grading
The AI automatically evaluates candidate answers against your specific criteria using four tiers:
🟢 Strong (Green): The candidate provided a thought out answer that hits the core intent of the prompt.
🟡 Fair (Yellow): The candidate gave a conditional answer, or only met the requirements under specific circumstances.
🔴 Weak (Red): The response missed the core competency entirely or showed clear misalignment with the role.
⚪ N/A (Grey): The question was not answered.
🔑❗ Crucial Step: Setting up your initial job evaluation criteria correctly is absolutely key to this process! The AI scores responses strictly against the rules and benchmarks you define during job setup. Accurate criteria ensure highly accurate, reliable grades. How to configure criteria for AI interviews covered in this article (jump to Step 3)
🔍 Step 1: Navigating to AI Interviews
Navigate to AI Interviews > click into the AI Interview. This will show you a table view of candidates
2. Select a candidate (e.g., "Hannah Tester") to open their interview details.
3. Beside each question in the transcript view, you will see a color-coded indicator (Green, Yellow, or Red) detailing how closely their response matched the criteria.
4. Quick-Switch Profiles: Click the Next (➡️) arrow icon in the top panel to jump straight to the next candidate's interview without returning to the main list view.
📝 Step 2: How to change criteria scores
If you would like to modify the evaluation at the criteria level, click on the dropdown next to the Strong/Fair/Weak that you'd like to change.
In the popup, you can perform the following changes:
Type notes or feedback in the provided text field as to why you are adjusting the evaluation score.
Check the Flag AI score checkbox to note that the automated grade was wrong and should be reviewed by the Puck team.
2. Click Save.
❗Notice that the Overall score for the candidate has changed according to your modified evaluation.
💡 Why It Matters
Instant Visual Triaging: The stoplight system allows you to skim long transcripts in seconds and immediately pinpoint where a candidate excelled or struggled.
🟢🟡🔴
Granular Breakdown: Unlike a single global score, this rates answers on a per-question basis so you can see exactly which specific competencies were missed. 📊
Audit and Calibrate: Flagging scores and typing adjustment feedback keeps a transparent audit trail for your team and helps improve future AI calibration! 🧠