The first interview you run is usually bad in a predictable way. You talk for thirty of the forty-five minutes, you like the person who is easiest to talk to, and you come out with a strong feeling and no evidence.
That is not inexperience. It is the default outcome of an unstructured conversation, and it happens to experienced managers too. The fix is not charisma or a sixth sense for people. It is structure, decided before the candidate arrives.
The research on this is unusually clear. Schmidt and Hunter’s much-cited 1998 meta-analysis of selection methods put the predictive validity of structured interviews at 0.51 against 0.38 for unstructured ones, and more recent meta-analytic work has found an even wider gap, roughly 0.42 against 0.19 (Test Partnership’s review of the literature, retrieved November 2026). Same hour, same candidate, substantially better prediction, purely because the questions and the scoring were fixed in advance.
This page is about running the interview itself. Before you book one, be clear what the role actually owns. A vague role produces a vague interview, which is why the who-does-what question comes first.
The Bottom Line
Write the questions and the scoring before you meet anyone. Structured interviews predict performance roughly twice as well as conversational ones.
Aim to talk for about 20% of the time. If you are explaining the business for ten minutes, you are not gathering evidence.
Ask what someone actually did, not what they would do. Past behaviour has a track record; hypotheticals reward good imaginations.
Score each candidate on the same three to five criteria immediately after the interview, before you discuss them with anyone.
Some questions are illegal to ask, including age, disability, pregnancy, family plans, national origin and religion, even in friendly small talk.
Lower your delivery costs by 23%
How we reduce costs:
- No delivery vehicle expenses
- Optimized local routes
- Pay-per-delivery model
- Average 23% delivery cost reduction
Why structure beats instinct in an interview
An unstructured interview measures how much you enjoy a conversation. That correlates with how similar the person is to you, how confident they sound, and how well they interview. None of those is the job.
Structure means four things fixed in advance:
The same core questions for every candidate, so you are comparing answers rather than impressions.
The same criteria, three to five of them, drawn from what the role actually has to produce.
A scoring scale you fill in alone, before any discussion.
A set order, so every candidate gets the same amount of each section.
None of that stops you having a human conversation. It stops the conversation from being the only evidence you end up with. The meta-analytic finding above is the whole argument: structure is the single characteristic that most reliably moves interview validity, which is why it is worth an hour of preparation even when you are hiring one person.
Before the interview: the 30 minutes that decide the hour
Preparation is where a first-time interviewer gains the most ground.
1. Write down what the role must produce. Three to five outcomes, not a list of duties. “Orders leave packed correctly and on time, four mornings a week” is an outcome. “Helps in the warehouse” is not.
2. Turn each outcome into one criterion. Reliability. Accuracy under time pressure. Handling a customer who is annoyed. Working unsupervised. Keep it to five at most, because you cannot hold more than that in a single interview.
3. Write two questions per criterion. One about a past situation, one follow-up that digs for detail. Write them down and use the same ones every time.
4. Decide what a good answer looks like. Briefly, for each question: what would a 1, a 3 and a 5 sound like? This is the step that separates scoring from gut feel, and it takes about ten minutes.
5. Read the application beforehand, not in the room. Reading it in front of the candidate costs you five minutes and signals that their time matters less than yours.
6. Plan the practical part. For hands-on work, a short task tells you more than any answer. Packing a sample order, reading a route list, backing a van into a bay with you watching. Keep it brief, make it the same for everyone, and be clear it is part of the process.
How to structure a 45-minute interview
A simple shape, repeatable for every candidate.
| Minutes | What happens | Why it is there |
|---|---|---|
| 0–3 | Welcome, outline the plan, say how long it will take | Settles the candidate; nervous people interview badly and you learn nothing |
| 3–5 | Two sentences on the role. Two, not ten | Your long description teaches them the answers you want to hear |
| 5–30 | Your written questions, in order, with follow-ups | The evidence-gathering core of the hour |
| 30–38 | Practical task or walk-through of the workspace | Shows what someone does rather than says |
| 38–43 | Their questions | What they ask is itself information |
| 43–45 | Next steps and a date | Good candidates take other offers while you deliberate |
The ratio to watch is talking time. Aim to speak for roughly one minute in five. First-time interviewers routinely take half the hour, usually from nerves, and then have nothing to score.
Questions that actually predict performance
The best questions ask for a specific past event, then follow up until you have detail rather than a summary.
Behavioural questions, the backbone of the hour:
“Tell me about a morning when everything ran late. What did you actually do?”
“Describe a time you noticed a mistake that wasn’t yours. What happened next?”
“Tell me about the most difficult customer you’ve dealt with. What did you say to them?”
“Give me an example of a job where you worked without anyone checking on you. How did you organise your day?”
The follow-ups are where the value is, because the first answer is usually rehearsed: What was your part specifically? Who else was involved? What did you do first? How did it turn out? What would you do differently?
A useful listening habit: notice “we” versus “I”. Candidates describing a team achievement in “we” language and then struggling to say what their own part was are often describing someone else’s work.
Questions worth less than they seem. “Where do you see yourself in five years?” rewards fluency. “What’s your greatest weakness?” has a standard script by now. Hypotheticals like “what would you do if a delivery was refused?” test imagination, not behaviour. If you want to know how someone handles a refused delivery, ask about a time something was refused, rejected or returned on them.
One question most people skip. “What would make you leave a job within six months?” It is the most honest conversation you will have in the hour, and it surfaces the mismatch that would otherwise cost you a rehire. If the answer names something your operation does every week, such as early starts, weekend routes or physical loads, you have learned the most useful thing available.
What you cannot ask in a job interview
Some questions are unlawful regardless of intent, and they most often slip out as friendly small talk while you walk someone to the door.
Do not ask about, or around, any of these: age or date of birth, race, color, national origin or accent, religion, sex, pregnancy or plans to have children, marital or family status, childcare arrangements, disability or health history, or genetic information. Federal law prohibits basing employment decisions on these grounds (EEOC), and many states and cities add more. Several restrict asking about salary history or criminal records at this stage.
The workable rule: ask about the job, not the person.
Instead of “do you have kids?”, ask: “This role starts at 5 a.m. four days a week. Does that work for you?”
Instead of “any health problems?”, ask: “The job involves lifting crates up to 50 pounds repeatedly. Can you do that, with or without an accommodation?”
Instead of “where are you from originally?”, ask: “Are you authorised to work in the United States?”
Instead of “how old are you?”, ask: “Do you hold a current licence for this class of vehicle?”
Each version asks about the actual requirement, which is both legal and more useful. Write your questions down in advance and this problem largely disappears.
Scoring: how to decide without relying on a feeling
Score immediately, alone, before you talk to anyone else about the candidate. Memory decays within the hour and group discussion overwrites individual judgement with whoever speaks first.
A one-page scorecard is enough:
| Criterion | Score 1–5 | Evidence from the interview |
|---|---|---|
| Reliability | ||
| Accuracy under time pressure | ||
| Handling an unhappy customer | ||
| Working unsupervised | ||
| Practical task |
The evidence column is the part that matters. A score with no quote or example next to it is a feeling with a number attached. If you cannot write anything in that box, you did not ask enough follow-ups, which is itself useful information about the interview, not the candidate.
Then compare candidates criterion by criterion rather than as whole people. “Who did I like most?” and “who scored highest on accuracy?” produce different shortlists surprisingly often, and the second question is the one connected to the job.
One more discipline: do not lower a criterion because you like someone, and do not keep interviewing to find a reason to hire the person you already chose. If every candidate scores poorly on something the role really needs, the honest conclusion is to keep looking. Hiring the least-bad option costs far more later, when the alternative is retraining or working through a termination a few months in.
Common first-timer mistakes, and the fix
Talking too much. Fix: write the questions down and keep your eyes on the list. If you have explained the business for more than two minutes, stop.
Letting the conversation drift because it is going well. A good rapport is pleasant and proves little. Fix: finish the written questions, then chat.
Deciding in the first two minutes. First impressions are confident and often wrong. Fix: scoring at the end, with evidence, which makes an early judgement visible to you.
Different questions for different candidates. This makes comparison impossible. Fix: a fixed core, with follow-ups varying as needed.
Skipping the practical. For hands-on roles, this is the most predictive part of the hour. Fix: keep it to eight minutes and run it identically every time.
Taking two weeks to decide. Good candidates in hands-on roles get hired within days. Fix: schedule interviews close together and commit to a decision date before you start.
Hiring from desperation during your busiest week. Fix: notice that the pressure is real and separate from the decision. If your week is too full to interview properly, that is a scheduling problem to solve first, not a reason to lower the bar.
Frequently asked questions
How do you conduct an interview as an employer for the first time?
Write three to five criteria from what the role must produce, write two questions per criterion, use the same questions for every candidate, run a short practical task, and score each person on a one-page sheet immediately afterwards. The preparation takes about half an hour and does most of the work.
How long should a job interview be?
Forty-five minutes is enough for most hands-on roles, and about 25 of those minutes should be your prepared questions. Longer interviews rarely add evidence; they add rapport, which is already over-weighted.
What are behavioural interview questions?
Questions that ask what someone actually did in a specific past situation, such as “tell me about a morning when everything ran late”. They predict performance better than hypotheticals because they describe behaviour that has already happened rather than behaviour a candidate can imagine.
How many people should interview a candidate?
One prepared interviewer beats three unprepared ones. If a second person joins, have both score independently before comparing notes, or the first opinion spoken aloud becomes everyone’s opinion.
What questions are illegal to ask in an interview?
Anything about age, race, national origin, religion, sex, pregnancy or family plans, marital status, disability or health history, or genetic information. Ask about the requirements of the job instead: the start time, the lifting, the licence. That version is both lawful and more informative.
Should I tell candidates about the pay in the interview?
Yes, give the range early. It saves both sides a second conversation, and a growing number of states require pay ranges to be disclosed in any case.
Write the questions down
If you do one thing from this page, do that. The questions on paper, the criteria on paper, the scores on paper at the end.
That is the entire difference between an interview that predicts how someone will work and a pleasant conversation you will second-guess for a month.
