The Interview That Tells You Nothing

The interview went well. They were personable, said the right things, and started three weeks later.

It did not go well or badly. It told you almost nothing, and the impression it left was mostly about how comfortable the conversation was.

What an interview cannot establish

Whether they will turn up. ⚠️ The single most important thing about a shift worker, and nothing in a conversation predicts it.

How they behave under pressure. A calm room tells you about a calm room.

How they treat people they are not trying to impress. ⚠️ The thing you most need to know, and structurally unobservable in a setting where they are trying to impress you.

Whether the hours actually work. They will say yes. Everyone says yes in an interview, including people who will discover in week three that it does not work.

What it can establish

Genuinely useful, if you ask for it:

Whether they understood the job. Ask them to describe it back. Misunderstandings surface immediately and are much cheaper here than in week two.

Whether the practical facts fit. Transport at 03:00, other commitments, the actual pattern. ⚠️ Ask about the specific night, not "are you flexible" — the second question gets a yes from everyone.

What they are looking for. Someone wanting something you cannot offer will leave, however good they are.

Whether they ask anything. ⚠️ Not as a test of enthusiasm — a person who asks nothing has not thought about the job concretely, and that predicts something.

The trial is the interview

For this work, watching someone for a few hours outperforms any conversation.

Watch three things. Whether they ask when unsure. Whether they move when there is nothing to do. Whether they listen to people all the way through.

⚠️ Not how polished they are — that is training. These three are much harder to teach and they are visible in an hour.

Pay for the trial. It is work, it should be paid, and it changes the relationship in a way that shows.

Give them something real. ⚠️ A trial spent shadowing tells you nothing — you need to see them do something, however small.

The reference nobody takes

Frequently skipped, and one specific question makes it worth doing:

"Would you take them back?"

A hesitation is the answer. ⚠️ Everything else in a reference is guarded — this question is not, because it is about the referee's own judgement rather than about the candidate.

The decision

Decide on the trial, not the conversation. ⚠️ The interview should confirm the practical facts and be short.

And decide quickly. A good candidate is being considered elsewhere. Speed is a competitive advantage that costs nothing.

What predicts leaving

Worth weighing more heavily than most interviewers do:

A pattern of very short stays. ⚠️ Not one — a pattern. Ask about it directly, and the answer is frequently reasonable.

Travel that only just works. Marginal at hiring means unworkable in winter.

Hours agreed reluctantly. They said yes; watch whether they wanted to.

An interview measures interviewing

Who performs well  people who are good at interviews
What you need   people who are good at 22:00 on a Friday

⚠️ In this trade those two overlap very little, because the floor skill is sequencing under pressure, not speaking under attention.

Which is why getting better at interviewing does not help

⇒ ⚠️ The improvement worth making is deciding less in the interview, not conducting a better one.

---

Questions that cannot work

"What would you do if it got busy?"

⚠️ This measures who can construct a good answer on the spot. 　 ⇒ Constructing a good answer and moving well at 22:00 are different 　 　 abilities, and the interview only ever sees the first.

Ask in the past tense instead

"What happened on the busiest night at your last place?"

⇒ ⚠️ Past events can be checked; future intentions cannot be. Changing the tense of the question changes the quality of the answer more than any other adjustment.

---

The three things an interview does establish

Whether they arrive when they said
Whether you understand each other
Whether they accept the terms as stated

⚠️ These three are genuinely settled in an interview. Everything else is not, and treating the rest as settled is where the mistake happens.

---

The trial night is the real interview

Without deciding what to watch  "seemed nice", and you have run a second interview
With three things decided   an actual observation

⚠️ Decide in advance: do they ask when unsure; do they notice what they were not told; does their pace hold when it gets busy.

---

The reference almost nobody takes

One question is enough  "Would you work with them again?"

⚠️ The hesitation before the answer carries most of the information, and the call takes two minutes.

---

What actually predicts leaving

Not anything in the interview
But whether they were left alone in the first two weeks

⚠️ How you receive someone decides retention more than how you selected them. 　 ⇒ Which means the effort is better spent on the first fortnight than 　 　 on the hour before it.

---

---

Three to hold

Six-month retention, by whether they did a trial. ⚠️ The number that settles whether the trial is worth it — and it usually settles it decisively.

Days from application to decision. Long means losing the good ones to faster venues.

Leavers within eight weeks, and the reason. Frequently a fact that was available at interview and not asked about.

Where the record has to sit

Comparing who worked out against how they were hired needs both facts recorded at the time. Without it, hiring stays a matter of impressions and the same mistakes repeat.

tasteck keeps staff records alongside shift history, so how someone was hired sits with how long they stayed.

They were good in the interview. So is nearly everyone, and that is exactly why it told you so little.

---

Read next

• The Staff Who Bring Their Friends
• The Advert That Attracts The Wrong People
• The Advert You Never Took Down