The problem with classical interviews
"Unstructured interviews predict future performance at only 14%. Structured interviews: 51%." — Schmidt & Hunter, 1998 (reference study)
This is THE decisive phase of recruitment. And this is where most companies go wrong by conducting "gut feeling" interviews. The recruiter improvises their questions, asks different questions to each candidate, relies on intuition, forgets half of what was said. Result: biased decisions and costly mistakes.
Cognitive biases that distort your interviews
Main interview biases
| Bias | Description | Impact on decision |
|---|---|---|
| Halo effect | One positive quality influences overall judgment | Overall overestimation of candidate |
| Confirmation bias | We seek to confirm our first impression | Evaluation skewed from the start |
| Similarity bias | We prefer candidates who resemble us | Lack of diversity in recruitment |
| Recency effect | We remember the last candidate better | Biased comparison between candidates |
| Contrast effect | An average candidate seems excellent after a bad one | Judgment error through relative comparison |
Aurelia neutralises these biases
What Aurelia does for you
- 1
Generation of personalised interview grids
Based on the scorecard defined in step 1, Aurelia generates a custom interview grid in 10 minutes instead of 1 hour.
- –Typical structure: introduction, background, technical skills, soft skills, projection, candidate questions
- –STAR questions (Situation, Task, Action, Result) for each criterion
- –Evaluation indicators to score each answer
- –Alerts on potentially illegal or discriminatory questions
- 2
Automatic interview transcription
No more frantic note-taking during interviews. Aurelia transcribes for you with 95%+ accuracy.
- –Accepted formats: MP3, WAV, M4A, MP4
- –OpenAI Whisper technology — state-of-the-art in speech recognition
- –Speaker identification (recruiter vs candidate)
- –Timestamping of key passages for quick navigation
- 3
AI analysis of responses
Aurelia doesn't just transcribe, it analyses the quality of responses.
- –Consistency between responses and sought profile
- –Quality of examples given (STAR method respected?)
- –Points of strength and caution identified
- –Alignment with declared company values
- 4
Objective and documented scoring
Each candidate is evaluated on the same criteria, with the same scale, for fair comparison.
- –Score per criterion on the scorecard (1-5)
- –Automatic weighting according to defined weights
- –Overall score calculated automatically
- –Qualitative comments attached to each score
The STAR method explained
Aurelia generates all its questions according to the STAR method, scientifically validated as the best predictor of future performance:
- Situation: "Describe the context you were in"
- Task: "What was your role and responsibilities?"
- Action: "What did you actually do, step by step?"
- Result: "What was the measurable result of your actions?"
Example of interview scoring
Scoring grid — Candidate A (Senior B2B Sales)
| Criterion | Weight | Score /5 | Comment |
|---|---|---|---|
| B2B sales expertise | 30% | 4 | Concrete examples, long cycle mastery, solid customer references |
| Resilience to rejection | 20% | 5 | Recovered after sales failure, examples are detailed and convincing |
| Communication and clarity | 20% | 4 | Clear and structured, slight lack of synthesis on complex questions |
| Cultural alignment | 15% | 3 | Correct motivation but professional project still unclear |
| Evolution potential | 15% | 4 | Ambition compatible with position trajectory |
| OVERALL SCORE | 100% | 4.0/5 | Solid profile, explore further on long-term vision |
Evaluation best practices
Before moving to step 4
- Same questions asked to all candidates
Comparison is only possible if criteria are identical
- Notes taken immediately after interview
Memory distorts, evaluate within 15 minutes
- Facts noted, not interpretations
Note what the candidate SAID, not what you think they meant
- Interviews evaluated by 2-3 people if possible
Multiple evaluators reduce individual biases
- Recording agreement obtained from candidate
Legally mandatory before recording interview
- All shortlisted candidates have been evaluated
Scores are filled in for each scorecard criterion
How to get agreement to record an interview?
Can AI analyse non-verbal language or tone of voice?
How to structure interviews with multiple evaluators?
What to do if a candidate gives an evasive or generic answer?
Generate your personalised interview grid
Based on the position and criteria you've defined, Aurelia creates a complete interview grid with STAR questions and evaluation indicators.