The structured interview is now the best predictor of success in a role, ahead of aptitude tests, with a validity of 0.42 against 0.19 for the unstructured interview (Sackett et al., 2022). Yet most startup interviews remain a pleasant conversation followed by "I have a good feeling". Here is how to move from feeling to grid, without turning your interviews into an interrogation.
Updated 17 September 2026 · 9 min read
A scorecard is the written definition of success in the role, translated into observable criteria and a rating scale shared by every interviewer. It answers three questions before the first interview: what must this person achieve within 12 months, which skills and behaviours make that possible, and how do you recognise an insufficient, adequate or excellent level on each.
It is not the job description (which lists responsibilities) nor the question guide (which derives from it). It is the reference that makes assessments comparable from one interviewer to the next and one candidate to the next.
Key point: without a scorecard, each interviewer assesses the candidate against their own idea of the role. Three interviewers, three different roles, and a decision made on whoever is most convincing in the meeting rather than on the best candidate.
An interview is structured when all candidates for the same role answer the same questions, in the same order, and are rated with the same grid. Questions are tied to the scorecard criteria, answers are written down, and each interviewer rates alone before talking to the others. You can follow up, dig deeper, be warm: the structure is about substance and scoring, not tone.
Three ingredients set it apart from a classic interview:
Because research has been measuring it for forty years, and the latest state of the art is unambiguous:
And one argument that is harder to measure but decisive: the structured interview reduces bias. When criteria are written and tied to the role, you assess what the person can do, not where they come from or who they resemble. It is also what French labour law requires: hiring methods must be relevant to the role, and no criterion unrelated to skills may come into play.
Allow two hours with the hiring manager, once, per role. It will serve every hire on that role.
For an Account Executive role in a SaaS scale-up (fictional example, to adapt):
| Criterion | How it is observed | Anchor "1" | Anchor "4" | Assessed by |
|---|---|---|---|---|
| Customer discovery | 20-minute role play on a fictional prospect | Pitches the product without asking a question | Gets the prospect to state their problem and its cost before talking product | Manager |
| Pipeline management | Behavioural question: "walk me through your last quarter, deal by deal" | Does not know their numbers or attributes them to luck | Gives conversion rates per stage, explains what they changed and why | Manager |
| Resilience | "Tell me about the lost deal that taught you the most" | Blames the product or the market | Identifies their own mistake, what they changed afterwards, with an example of result | Peer |
| Collaboration with product / CS | "A time you had to defend a customer against an internal team" | Has no example or describes escalating to management | Describes a disagreement handled directly, with a documented compromise | Peer |
| Fit with the context | "What would make you leave after six months?" | Generic answer or avoidance | Names precise conditions, consistent with what they were told | Founder |
Five criteria, three interviewers, each with their own questions. Scores are on a 4-point scale (no midpoint, to avoid the "default 3"), and a 1 on a criterion flagged as eliminatory stops the process.
0 → 5 min
The agenda, the criteria you will explore, the fact that you take notes. The candidate knows what to expect, tension drops.
5 → 40 min
2 or 3 criteria, 2 questions per criterion, follow-ups using STAR (situation, task, actions, result). Dig into personal actions, not the "we".
40 → 55 min
The role, the first six months, the team, the salary range if not already shared. Then the candidate's questions, which are a signal too.
55 → 70 min
Before any discussion with a colleague: one score per criterion, one sentence of justification with a quoted fact, a recommendation. Fifteen minutes blocked in the calendar.
The debrief is where the structured interview wins or loses everything. Three rules:
The decision belongs to one person, the manager, with a founder veto. A committee that votes produces consensual hires, rarely the best ones. And all of it must fit the timeline: see reduce your time-to-hire.
Last point: the scorecard does not stop at signature. The criteria become the goals of the 30-60-90 plan, and at six months you compare the interview score with actual performance. It is the only way to know whether your process predicts anything, and the first lever to reduce early turnover.
FAQ
It is the written definition of success in a role: the mission in one sentence, three to five expected outcomes at 12 months, four to six observable criteria (technical skills, behavioural skills, knowledge) and a rating scale with anchors describing an insufficient, partial, solid or excellent answer. It serves as the shared reference for every interviewer on a given role and then feeds the 30-60-90 plan.
Because all candidates answer the same questions tied to the role's criteria and are rated with the same grid, individually, before any debrief. The meta-analysis by Sackett, Zhang, Berry and Lievens (2022) makes it the best predictor of job performance, with a validity of 0.42, against 0.31 for cognitive ability tests and 0.19 for the unstructured interview. It also reduces bias, since you assess observable behaviours rather than an impression of resemblance.
Four to six. Fewer and you do not cover the role; more and interviewers skim. Each criterion must be observable in an interview or a practical case, and criteria are split between interviewers so that each assesses two or three in depth. One or two criteria can be flagged as eliminatory.
On a four-level scale (1 insufficient, 2 partial, 3 solid, 4 beyond expectations), with no midpoint to avoid the default 3, and a written anchor for each level. Each interviewer scores alone, in writing, within fifteen minutes of the interview, with a quoted fact to justify each score. The debrief opens by displaying all scores and focuses on the gaps between interviewers.
Three steps are enough in most cases: a prequalification, a manager interview with a practical case, a team and founder meeting. Google showed that four interviews predict a hiring decision with 86% confidence and that further interviews add almost nothing. 90% of French managers and professionals want two or three interviews at most (APEC, 2024). Beyond that, you lose candidates without gaining accuracy.
Related guides
Sources
Contact
We build your scorecards and interview guides and train your managers. A complete interview kit in three weeks.
Book a call →