Almost every bad hire traces back to a decision made before the first CV was opened. This is the order the process runs in, and what to settle at each stage.
Assessing candidates means gathering comparable evidence about each applicant against a role you have already defined, then deciding. For a company without an HR department it is mostly a sequencing problem: almost every bad hire traces back to a decision made before anyone read a CV, not after.
The most common way a small company hires badly is to open a search for a good person rather than for a defined job. A good person is not a specification. It cannot be compared against, it cannot be disagreed with, and it quietly becomes whoever the last interviewer liked.
Before the first CV, write down what the person will own in their first six months, what has to be true for that to go well, and which of those things you are willing to teach. That last column is the one people skip, and it is the one that decides whether you are hiring for experience or for aptitude.
The fourth item feels pessimistic and is the most useful. Naming the failure conditions in advance is what lets you notice them in month three rather than month nine.
Nothing in that list is expensive. The order is doing the work: every step exists to stop the next one from being decided by whoever spoke most confidently.
A comparison of three candidates shown two ways. Ranked against each other they are three numbers and the highest wins. Read against what the role requires, the top-ranked candidate is weakest on three of the five things the job needs.
Ranking candidates against each other feels like progress and produces a reliable failure: the best of a weak field looks strong, and a strong candidate who is wrong for this particular job looks like the obvious hire because they were the most impressive person in the room.
A single ranked score makes this worse by hiding it. If one number says 78, you cannot see that it is high because the candidate is exceptional at something the role does not need, and acceptable at the thing it does. Keep the requirements separate for as long as you can bear to.
This is what the assessment stack is arranged around — How the different kinds of assessment compare
Write down the requirement each question was meant to settle, what the candidate actually said, and what you concluded — separately. Conflating those three is how a hiring debrief turns into two people trading impressions they can no longer trace.
Do it immediately. Not that evening. An hour after an interview you are reconstructing a feeling rather than recalling an answer, and the reconstruction is reliably kinder to candidates you liked.
If you want the questions built for you, Generate a structured question set for the role
Structure makes your judgement comparable and traceable. It does not make it correct. You can run every step here and still hire the wrong person, because the role changed, or because someone was better at being assessed than at doing the job.
What it does buy you is the ability to find out why afterwards. An unstructured process that goes wrong teaches you nothing, because there is no record of what you believed and when.
Enough that you are choosing rather than confirming, which in practice usually means more than three. But a small company hiring against a clear definition can stop as soon as someone clearly meets it. Interviewing more people to feel thorough is a cost, not a control.
Where the job has an observable output, yes — it is the most direct evidence available. Keep it to a couple of hours, make it resemble the actual work rather than a puzzle, and pay for it if it takes longer than that.
Before the interview, to inform what you ask. It tells you which areas are worth probing with this particular person. Used after the interview as a score to confirm a decision you have already made, it is doing nothing except making you more confident.
The person who will manage the role, and one person who will work alongside it. More than three interviewers rarely adds information and reliably adds calendar delay, which is how small companies lose candidates to faster ones.
Go back to the role definition and find out which requirement each person is disagreeing about. Most hiring disagreements are not about the candidate; they are two people applying different definitions of the job and discovering it late.
Operated by Alano Tech Pte. Ltd., Singapore.
Terms of Service Privacy Policy
Also from us: Alano.ai Opptymizer Omu Labs