One habit doubles the accuracy of everything you do
Before we talk about types, here is the single finding that matters most. In the largest re-analysis of selection research to date, a structured interview predicted job performance more than twice as well as an unstructured one — and out-predicted cognitive ability tests, personality tests and biodata. Same room, same candidate, same hour. Different accuracy, entirely because of how the interview was built.
Values are mean operational validity coefficients (r). Higher = better prediction of on-the-job performance. Assessment centres were re-estimated at .33 in the 2023 follow-up using manager-specific reliability. Structured interviews also have wide spread (80% credibility interval .18–.66) — which is exactly the point: a well-built interview sits near the top of that range, a sloppy one near the bottom. That gap is your job.
Questions derived from the job, the same questions in the same order for every candidate, a rating scale with written anchors, notes taken during the interview, and scores recorded before you discuss with anyone else.
It is not a robot reading a script. You still build rapport, still probe, still follow up. Structure controls what you ask and how you score — not your warmth.
Same questions + written anchors = a defensible, comparable record. When a client challenges a shortlist, you show evidence instead of adjectives.
The ten interview types, easiest to toughest
Ranked by how hard they are for a recruiter to run well — not by how important they are. Click any rung to open it: what it measures, when to use it, real questions, a real example, the traps, and a short check that you understood it. Use the filters to see only the types that apply to a seniority level or a hiring situation.
What is a competency — and why it is not a skill
This is the most important distinction in structured hiring. If you blur it, you end up testing the wrong thing, hiring the wrong person, and not knowing why.
| Skill | Competency | |
|---|---|---|
| What it is | A specific technical ability that can be taught, tested and verified — usually with a right answer. | A pattern of behaviour that predicts performance across many situations — there is no single right answer, only better or worse evidence. |
| Examples | Writing SQL queries. Reading a P&L. Using Salesforce. Typing 60 wpm. Speaking B2 English. | Ownership. Stakeholder influence. Problem-solving under pressure. Learning agility. Commercial judgement. |
| How you test it | Technical interview, written test, work sample with a defined rubric. The expert checks whether the answer is correct. | Behavioural or situational interview, with a question asking for a real past example. You score the quality of the evidence, not the correctness of a fact. |
| Can you train it? | Yes — with enough practice, most people can learn most skills to an acceptable level. | Harder. You can coach someone to give better examples in interviews. Changing the underlying pattern takes much longer and is far less certain. |
| Common mistake | Listing skills that are not actually used in the job — then interviewing for them and eliminating candidates who would have learned them in week two. | Confusing competencies with personality traits: "is a team player" is not a competency. "Raises a problem to the right person instead of absorbing it quietly" is. |
How to identify the competencies a role actually needs
You do not find competencies in a job description. You find them by asking five questions about the job. The JD tells you what the person will do — these questions tell you how they will need to do it to succeed.
Ask the hiring manager: "Think of the last person who did not work out. What specifically went wrong?" The answer almost always names a competency — not a skill gap. Write it down exactly as they say it.
The decision type reveals the competency. Decides under time pressure with incomplete data? That needs judgement and composure. Decides by persuading people above them? That needs stakeholder influence. No authority = lots of influence needed.
The scope of the consequence tells you how many competencies you need and how high the bar should be. A bad week from a customer care agent affects three clients. A bad week from the operations lead affects the whole quarter.
Internally above, internally across, or externally? Each points to a different flavour of influence and communication. Lateral influence inside a matrix organisation is one of the most commonly under-tested competencies in hiring.
This is where failure hides. A strong individual contributor becoming a manager for the first time does not lack skill — they lack the leadership and delegation competencies the new role demands. Test for those, not for the ones they already have.
Competency compass
You do not choose an interview type because it is fashionable. You choose it because of the competency you need evidence on. Click a competency — the compass tells you which interview types actually measure it, which ones only pretend to, and gives you a real question you can use today.
Seniority decides the loop
The bigger the responsibility, the more expensive the mistake — so the loop gets longer, the evidence gets deeper, and the question type changes. Note one research finding that surprises most recruiters: hypothetical "what would you do" questions lose their power at senior level. Click a level to build its interview loop.
How to build and fill a scorecard
A scorecard is not paperwork you do after the interview. It is the interview, written down in advance. Build it before you write a single question, and the questions write themselves.
| Competency | 1 | 2 | 3 | 4 | 5 | Evidence (what they actually did) |
|---|
Assessing behaviour without fooling yourself
Behavioural assessment rests on one assumption: the best predictor of future behaviour is past behaviour in a similar situation. That only holds if you get a real, specific, recent example — and most candidates will not give you one unless you insist.
- S — Situation: when, where, who else was involved. Anchors it in reality.
- T — Task: what they specifically owned. Separates them from their team.
- A — Action: what they did, step by step. This is where the score comes from.
- R — Result: what changed, ideally measured, plus what they learned.
- "We usually..." → a policy, not an example. Probe: "Tell me about the last time that happened."
- "I would..." → a hypothetical in a past-behaviour question. Probe: "Has that come up? Walk me through it."
- "We decided..." → a team. Probe: "What was your part in that decision?"
Each situation describes something that happened in an actual debrief. Choose the bias. After you name it you will get the fix — the structural move that makes that bias harder to repeat.
Score two real candidates
Read the exchange, set your scores, then compare with the calibrated rating and the reasoning behind it. Being within one point on every competency is a pass. Being two points out on any competency means you are reading confidence as competence — the most common failure in this job.
Final check
Twelve questions covering everything above. Do it without scrolling back up — that is the whole point.
Run this as a 2-hour session
Suggested timing if you are delivering this live to a new joiner or a small group. Everything in bold is done by the trainee, not by you.
Sources
Everything numerical in this guide traces back to peer-reviewed selection research or a government selection authority. If a trainee asks "says who?", this is the answer.