How to Interview When You've Never Hired Before

If you have never hired, the instinct is to have a conversation, see how it goes, and trust your read. That instinct is well-documented as unreliable — unstructured interviews mostly measure how comfortable the conversation was, and comfort correlates with similarity rather than capability.

The fix is not complicated. Same questions, in the same order, scored independently, with a piece of real work in the middle.

Structure the conversation

Write the questions before you meet anyone. Six to eight, tied directly to the outcomes in the job description. Ask them in the same order every time.

This feels stiff and it's the entire point. When two candidates have answered the same question, you can compare answers. When they've had two different conversations, you're comparing how each one made you feel.

Ask about the past, not the hypothetical.

| Weak | Strong | |---|---| | How would you handle a difficult client? | Tell me about the last difficult client you dealt with. What happened? | | Are you organised? | Walk me through how you tracked your work in your last role. | | How do you handle pressure? | Tell me about the last time you missed a deadline. What did you do? |

Hypotheticals measure how well someone talks about work. Past behaviour is the better predictor, and it's much harder to invent under follow-up.

Follow up on specifics. When an answer is vague — "we improved the process" — ask what it was before, what changed, and what the result was. Two follow-ups separate people who did the thing from people who were nearby when it happened.

Use a work sample

The highest-signal part of any hiring process, and the part small businesses skip.

Give a real, small task that resembles the actual job:

• Bookkeeping role → reconcile a messy sample month and note what you'd query
• Support role → draft replies to three real (anonymised) customer emails
• Marketing role → outline a piece for our audience and explain the angle
• Ops role → read this process description and tell us what would break first

Pay for it at your normal contractor rate for anything over about an hour. Unpaid multi-hour exercises filter out the candidates who have options and no spare evenings, which is the opposite of what you want.

Keep it to two or three hours. And give everyone the same task, so you can compare.

Score before you discuss

Anchoring is real and it's fast: the first opinion spoken aloud shapes everyone else's.

So each interviewer fills in a scorecard independently, before anyone talks:

CANDIDATE: ______________          ROLE: ______________

OUTCOME 1 — [from the job description]      1  2  3  4     Evidence:
OUTCOME 2 — [from the job description]      1  2  3  4     Evidence:
OUTCOME 3 — [from the job description]      1  2  3  4     Evidence:

WORK SAMPLE                                 1  2  3  4     Evidence:

WORKING NORMS — [the ones you named]        1  2  3  4     Evidence:

Would I want this person doing this work?   Yes / No
Biggest reservation:

The evidence column does the work. A score with no evidence is an impression wearing a number.

And use an even scale. A five-point scale collects threes.

What "no" looks like

Two failure modes, and the second is worse.

Hiring on rapport. You liked them. That is not nothing — you'll work together — but it is not evidence they can do the job, and it correlates with similarity in a way that quietly narrows a small team.

Hiring because you're tired of looking. The most expensive hire is the one made in week seven of a search out of exhaustion. A vacancy is a known cost; a wrong hire in a five-person business is a much larger and slower one.

If nobody clears the bar, the honest answer is to re-open the search and consider whether the role, the range, or the description is the problem.

Tell people the outcome

Everyone who had a conversation gets a reply. Everyone who did a work sample gets a reply and gets paid, whatever the decision.

This is not just decency. In a small market your reputation as an employer travels through exactly the people you rejected, and the ones you treated well refer others.

The mistakes

1. Different questions for each candidate. You've collected impressions, not comparisons.
2. Hypothetical questions. They measure articulacy.
3. Unpaid multi-hour exercises. They filter out the strongest candidates.
4. Discussing before scoring. The first opinion anchors the room.
5. "Culture fit" as a gut call. Name the actual working norms and assess against those.
6. Five interview rounds. After the third you are collecting reassurance, not information.

What to do next

Write your six questions and your scorecard before you speak to anyone. If you have already started interviewing, write them now and re-interview the shortlist — it's a short conversation and it converts a set of impressions into something you can actually compare.

Then plan the first ninety days before you make the offer, not after.