A structured interview against a conversation, and which one predicts the second month

Most small employers interview by conversation. It feels natural, it is easy to run, and it reliably identifies the candidate who is most comfortable in a conversation.

That is a real skill in some jobs and irrelevant in most. The alternative takes the same forty-five minutes and produces a comparison rather than an impression.

What the unstructured version actually measures

Three well-known distortions, and each one is easy to spot in your own notes afterward.

• Similarity. Interviewers rate candidates higher when they share a background, a hobby, a hometown. This is not favoritism in any conscious sense; it comes from the conversation being easier.
• First impressions setting the rest. A judgment made in the opening minutes turns the remaining forty into a search for support.
• Different questions for different people. Once candidates have been asked different things, there is nothing to compare. The decision then rests on recall, and recall favors whoever was most memorable.

The result is not that bad hires get made every time. It is that the process cannot tell you why a good one worked, so it cannot be repeated.

The structured version, in four parts

1. Define what competent looks like, first

Before writing any question, write four to six things a person in this role must be able to do by the end of month two. Concrete and observable. Not qualities.

Good: close out a job ticket accurately without being asked. Handle two customers at the counter at once during the morning rush. Read a supplier invoice and catch a pricing error.

Not useful: strong communicator, team player, detail-oriented.

This list is the whole basis of everything below, and writing it usually clarifies the role for the employer as much as for the candidate.

2. Write one question per item, and ask it of everyone

Two question types do nearly all the work.

• Past behavior. Tell me about a time you had two customers waiting and a phone ringing. What did you actually do?
• Situational. A customer says they were quoted a lower price and there is nothing in the file. Walk me through your next ten minutes.

Ask them in the same order, with the same follow-ups: what was the situation, what did you do specifically, what happened. Follow-ups are where the answer becomes real, because rehearsed answers thin out under a request for specifics.

3. Score as you go

A three-point scale is enough. Below what the job needs, meets it, clearly above it. Write the score during the answer, not afterward.

Define what a strong answer contains before the interview. Not the words, the substance. This takes ten minutes and it is what makes two interviewers comparable.

4. Add a work sample

The single most predictive element, and the most underused by small employers.

A short, paid, realistic task. Twenty minutes to an hour of something the job actually contains: sorting a stack of invoices, answering a customer email, diagnosing a simple fault, laying out a small piece of material.

Rules that keep it fair and defensible: pay for anything beyond a brief exercise, do not use work you would otherwise pay someone to do, give everyone the same task with the same time, and tell candidates in advance so nobody is ambushed.

The comparison, plainly

Where the conversation still earns its place

Not as the assessment. As the closing.

Strong candidates are choosing too, and the part where they meet the owner, see the workspace and ask their own questions is what decides whether they accept. Run it after the structured portion, with the assessment already done, so warmth is not competing with judgment.

Ten minutes of it at the end costs nothing and is often what wins the hire.

Practical setup for a small shop

1. Write the month-two list. Keep it to six items.
2. Write six questions and what a strong answer contains. Save the file; it serves every future hire for this role.
3. Ask the same questions in the same order and score during the interview.
4. Run a short paid work sample for the final two candidates.
5. Have a second person sit in on the finalists and score independently before you compare notes.
6. Compare scores against the list, not against each other's likability.

The part that shows up in month two

Candidates who score well against a concrete list tend to arrive able to do the things on the list, which sounds obvious and is exactly what an unstructured process fails to deliver.

The second effect is on the employer's side. Having written down what competent looks like, you now have the thirty and sixty day standards already drafted, the training plan implied, and the honest job description for the next posting. One hour of preparation produces four documents, and the interview is only the first of them.