Hiring Teams
How to Build a Structured Interview Process That People Actually Follow
By the Upinterview team · Published · 9 min read
Research on hiring has pointed the same way for a long time: structured interviews predict how someone will do in the job better than free-form conversations do. Most teams know this and still wing it, because a chat is easier to run than a structure, and because the chat feels like it is working. If your process is "whoever is free asks whatever comes to mind", this is how to change that without turning interviews into an interrogation.
What "structured" means
It is not a script read in a monotone. A structured interview has four features. Every candidate for a role is asked the same core questions. Those questions are tied to things the job actually needs. Answers are scored against written criteria rather than a general impression. And interviewers record their scores before they compare notes.
You can still ask follow-ups and have a proper conversation. The structure is the spine, not the whole body.
Step 1: decide what the job needs
Pick a small number of competencies, usually four to six. More than that and interviewers cannot hold them in their heads. For a backend engineer they might be problem solving, system design, code quality, communication and ownership. For a customer success manager: handling difficult conversations, prioritization, product understanding, and commercial awareness.
Write each one as a sentence describing behavior, not a vague label. "Communicates technical trade-offs so a non-engineer can decide" is usable. "Good communicator" is not.
Step 2: spread the competencies across rounds
One of the most common flaws in interview loops is that every interviewer asks about the same things. Candidates get "tell me about a time you worked with a difficult teammate" four times and nobody covers system design. Draw a grid: rounds down the side, competencies across the top, and tick who owns each. Everything should be covered by at least one round, and most things by only one.
Step 3: write the questions, and what good looks like
For each competency, write one or two main questions and a few follow-ups. Behavioral questions ask about the past ("tell me about a time you..."). Situational ones ask about a hypothetical ("you inherit a service with no tests, what do you do first?"). Technical work samples ask the candidate to do a small piece of real work. A mix is fine.
Then, and this is the part teams skip, write down what a strong answer contains and what a weak one looks like. Without that, two interviewers will hear the same answer and score it differently, and you will spend the debrief arguing about what was said rather than what it means.
Step 4: score with anchors
Use a short scale, with a description at each point. A four-point scale works well because it forces a lean one way or the other. Here is an example anchor set for one competency:
| Score | Communicating trade-offs |
|---|---|
| 1 | Could not explain the options or why they chose one. |
| 2 | Explained their choice but missed the main downside, or needed a lot of prompting. |
| 3 | Laid out options and downsides clearly and reached a reasoned decision. |
| 4 | Did all of that, adapted the explanation to the listener, and named what would change their mind. |
There is a ready-made format in our interview scorecard template.
Step 5: score first, discuss second
Ask everyone to submit written scores and notes before the debrief, and not to read each other's first. Otherwise the first strong opinion in the room, often the most senior person's, anchors everyone else. In the debrief, go competency by competency and look at where scores differ. Those gaps are where the useful conversation is.
Step 6: train and calibrate
A structure nobody has practiced with goes back to chatting within a month. New interviewers should shadow a couple of interviews before running one alone. Every so often, have two people score the same recorded interview independently and compare. If their scores are far apart, your anchors need work, or your interviewers do.
Why teams drop structure, and how to stop it
Teams usually drop structure for boring reasons. The forms are too long, so keep to a handful of competencies and a few lines of notes for each. Nobody checks, so someone, often the recruiter, has to notice when scorecards are late or empty. Hiring managers override the result without explanation, and people learn it does not matter; ask for the reasoning in writing. And sometimes it just feels stiff, which is fixed by letting interviewers ask in their own words as long as they cover the same ground.
What to use to run it
A shared document and a spreadsheet are enough to start. As volume grows, a platform helps by handling scheduling, keeping questions and scorecards together and gathering feedback in one place. Ours is described on the internal interviews page, and for the interview questions themselves the question bank keeps reusable sets by role and skill.
When there simply are not enough interviewers for the load, outsourcing part of the process can free your team to focus on the rounds only they can run.
Frequently asked questions
Do structured interviews feel robotic to candidates?
They do not have to. Interviewers can be warm and conversational and ask natural follow-ups. What stays fixed is the core questions and the way answers are scored.
How many questions should an interview have?
Fewer than you think. In a 45-minute interview, three to five main questions with follow-ups is usually plenty, since good answers need time.
Behavioral or situational questions?
Both work. Behavioral questions about the past are harder to bluff, while situational questions are useful when a candidate has limited experience. Many teams use a mix.
Can I change the questions between candidates?
Keep the core questions the same for all candidates for a role, otherwise you cannot compare them fairly. You can update the set between hiring rounds as you learn what works.