Building an interview guide means deciding before the first conversation which job-related criteria matter, what evidence will demonstrate them, and how the answer will be recorded. That makes interviews more comparable and fairer: candidates have the same opportunity to show relevant capability, while hiring decisions rely less on memory, chemistry, or the direction a conversation happened to take.

An interview guide is not a script that removes all human judgment. It is a decision framework. It connects the work of a role to evidence gathered in a conversation, so a team can explain why someone progresses and why someone does not.

What makes an interview structured?

A structured interview uses predefined criteria, core questions, and rating anchors. The interviewer can still listen, respond, and ask clarifying questions. Structure comes from holding the decision constant: every candidate is assessed against the same role requirements, rather than against a different version of the job in each conversation.

Unstructured conversations can be valuable for getting to know someone. They are less reliable as the sole basis for a hiring decision because different interviewers explore different topics, remember different moments, and give the same answer different weight. A guide reduces that variation. It does not decide who to hire; it requires the people making the decision to give role-relevant reasons.

That matters for fairness as well. Candidates vary in confidence, interview experience, and appetite for small talk. Asking for concrete situations, actions, and outcomes gives more room for job evidence to carry the conversation. For the broader process context, see Sprad’s guide to AI interviews and voice recruiting.

Start with the role, not with favourite questions

Do not begin with a list of popular interview prompts. Begin with the work a person will repeatedly have to do: the decisions they will make, the problems they will solve, and the people they will need to work with. Then ask which capability or knowledge makes success in each of those situations more likely.

For example, “works well with stakeholders” is too broad to rate. A more useful criterion might be: makes conflicting requests explicit, explains the available options, and helps reach a reasoned priority decision. That describes observable behaviour rather than an interviewer’s general impression.

A simple design rule keeps the guide usable: retain a criterion only when its absence would genuinely change the hiring decision. Everything that is merely interesting belongs outside the scored assessment. There is no universal ideal count of criteria; the realistic number is the smallest set of meaningful distinctions that an interview team can collect and discuss consistently. If two criteria are only different labels for the same capability, combine them.

• Work: What must this person repeatedly accomplish in the role?
• Criterion: Which capability or knowledge is essential to that work?
• Evidence: What concrete action, decision, or result would make the criterion visible?
• Question: What prompt is most likely to elicit that evidence?
• Rating: What would an answer below, at, or beyond the expected level look like?

This chain is more useful than a long question bank. It stops teams from evaluating whatever happens to arise in the conversation. It also separates the interview from document-led assumptions: a CV may suggest an area to explore, but the guide tests it through context and evidence. That distinction is important when planning a process for high application volumes and CV screening.

How to build an interview guide with questions that work

Questions produce useful evidence when they ask for a specific example. They invite the candidate to describe the situation, their room for action, the decision they made, and what happened next. This also gives the interviewer a legitimate route to probe without leading the candidate toward a preferred answer.

• Behavioural questions ask about a real past situation: “Tell me about a time you had to defend a priority decision despite disagreement.” They reveal what the person says they actually did.
• Situational questions present a credible future scenario: “A key requirement changes just before launch. How would you proceed?” They are useful where directly comparable experience is limited.
• Technically specific questions test job-relevant thinking: “What information would you seek before making that decision, and why?” They show depth, method, and the limits of someone’s knowledge.

Some common prompts generate little decision-quality information. Self-ratings such as “Are you resilient?” or “How collaborative are you?” invite desirable claims, not proof. Brain teasers mostly measure familiarity with brain teasers. Stress questions often measure a reaction to an artificial pressure situation rather than the capability needed in the role.

If pressure management or conflict handling matters, ask for a real example: what made the situation difficult, what did the person control, what did they do, and what was the outcome? A good follow-up rule is to deepen the evidence, not the biography. Ask about the person’s contribution, constraints, and decision process instead of collecting private or irrelevant information.

Rating anchors turn answers into comparable evidence

A rating scale without anchors creates only the appearance of consistency. Rating anchors are pre-agreed examples of what an answer at different levels looks like. They are not a template for one perfect career path. They are shared language that helps interviewers separate the evidence they heard from the impression they formed.

Take the criterion “stakeholder prioritisation.” An answer below expectation remains general, names no personal decision, and refers only to someone else’s instructions. An answer at expectation explains the competing demands, the priority chosen, and the people involved. An answer beyond expectation also makes alternatives and risks explicit, describes how the decision was communicated, and reflects on what the person would change next time.

This is where the information gain lies. “Strong answer” adds little information for a later decision meeting. A concrete example that clearly fits an anchor reduces uncertainty about the criterion. For that reason, interviewers should capture the evidence first and assign the rating second. The rating is the conclusion; the action, result, or explanation is the basis for it.

Record uncertainty too. If an answer does not support an anchor, the right response is a standard follow-up question, not a generous guess. If the evidence remains incomplete, the assessment should remain cautious or open. That is more honest than a precise-looking score that the conversation did not earn.

One guide for human and AI-supported first conversations

The same guide can support a human first conversation and an AI-supported first conversation when the criteria, core prompts, permitted follow-ups, and rating anchors remain the same. The AI should not invent new criteria or infer suitability from tone of voice, speaking speed, or other signals outside the agreed evidence. Its role can be to ask consistently, organise answers against the guide, and flag missing evidence for review.

People still own the assessment and the decision. They can weigh context, decide whether a follow-up is appropriate, and recognise circumstances a standardised exchange cannot fully capture. A good guide therefore creates continuity across formats rather than forcing every interaction to feel identical. The voice interview use case shows how an initial conversation can be delivered through additional channels while remaining connected to a hiring workflow.

For an AI-supported process, the version of the guide is itself a governance record: who approved the criteria and anchors, which follow-ups are allowed, what is passed to human reviewers, and who is accountable for the eventual decision? The AI interview and voice tools category can help teams frame the product category before comparing implementation options.

Documentation makes the decision traceable

Traceability does not start in the final debrief. For each criterion, record the question used, the relevant answer evidence, the resulting rating level, unresolved points, and the interviewer. Also note any departure from the guide and the reason for it. This lets a team later check whether candidates were genuinely assessed against the same standard.

Documentation should remain purposeful and proportionate. A complete transcript is not automatically more useful than concise evidence mapped to a criterion. Define retention, access, and review responsibilities with the appropriate privacy, employee-representation, and legal stakeholders for your organisation and jurisdictions.

The limits of an interview guide

An interview guide can assess only what its questions and anchors are designed to assess. It cannot prove future performance or capture the whole meaning of team fit. Roles involving practical work, regulated responsibilities, or safety-critical decisions need other job-relevant assessments as well as human overall judgment.

Standardisation should not become rigidity. Candidates need room for questions, context around non-linear careers, and appropriate accommodations. If the format or communication itself becomes the obstacle, the guide is no longer measuring the intended criterion reliably. Teams need to recognise and document that limitation rather than treat every answer as directly comparable.

Where a structured guide is delivered as a voice or chat conversation, Sprad’s Voice Interview can connect the questions, responses, and further process steps. It does not replace the work of defining job-relevant criteria or human accountability for a hiring decision. If the guide must feed into a later review step, the overview of AI-assisted CV screening helps keep those process stages distinct.

Frequently asked questions about interview guides

Should every candidate receive exactly the same questions?

The core questions for scored criteria should be the same. Follow-up questions are appropriate when they clarify the same missing evidence rather than introduce a new standard for one person. Record what was explored further and why.

How should a team create rating anchors?

Build them from real work situations, expected outcomes, and the risks of the role. Then test the descriptions with people who understand the work: would this answer actually demonstrate the capability needed? Rewrite anchors that are vague, overlapping, or based on reputation rather than evidence.

Can a guide work for candidates with very different backgrounds?

Yes, if it asks for transferable evidence rather than familiar employers or identical job titles. A person may have demonstrated a relevant capability in another industry, a project, education, or community responsibility. The evidence must relate to the criterion; the career path does not need to look the same.

Are stress questions ever appropriate?

Only when responding under a particular form of pressure is genuinely job-relevant and the format can represent that pressure fairly. In most cases, a question about a real pressure situation produces better, more reviewable evidence. Artificial stress can make interviews less comparable, not more.

When should an interview guide be updated?

Review it when the role, working environment, or recurring hiring mistakes change. Look beyond the wording of questions: did the criteria and anchors actually measure what mattered in subsequent work? Version changes so that earlier decisions remain understandable.