Most interview questions test one thing well: how well someone interviews. They produce confident answers from practiced candidates and stumbling answers from unpracticed ones. What they don’t reliably test is whether a candidate can do the actual job. That takes a different kind of question.

The Three-Section Interview Kit from The Structured Interview Process That Produces Better Hires in Less Time gives you the structure. This article gives you the design discipline: building questions that actually test each requirement rather than questions about the relevant subject matter.

The reframe is this: a good interview question is not a question about a topic. It is a question that can only be answered well by someone who has actually done the specific thing the role requires. Every question you write should have that property.

The Requirement-to-Question Map gives you a method for getting there.

Why Most Interview Questions Don’t Reveal Fit

The obvious problem: you ask questions that any competent candidate in this field could answer. “Tell me about your experience with customer relationships” is not a bad question. It is a question that every customer-facing candidate can answer, including the ones who will not be right for this role.

The less visible cost is that generic questions produce signals that don’t distinguish. Every candidate can describe a difficult client conversation or a project they managed or what they find rewarding. These answers give you material for the conversation but not a basis for comparing candidates.

The deepest problem is that the questions that are hardest to prepare for are the ones most worth asking. A candidate who has genuinely built a process from nothing can describe exactly what that looked like. A candidate who has only managed existing processes will struggle to give that answer with the specificity the question requires.

The question that requires specific experience to answer is the question worth asking.

The Requirement-to-Question Map

The map is built before the first interview. It takes the role’s requirements and converts each one into a question that specifically tests for it.

Step 1 lists the requirements worth testing

Start with the role’s primary output and non-negotiable traits, as defined in the success definition. Not all of them need a question. The ones that are hardest to assess from a resume, and most likely to determine success or failure, get questions. The rest can be evaluated from experience and references.

Pick two or three requirements to focus the interview on. An interview that tries to test everything tests nothing deeply. An interview that tests three things well gives you what you need to make a confident decision.

Step 2 writes the question that tests each requirement

For each requirement, write the specific question that would reveal whether a candidate has it. The test is: can this question only be answered well by someone who has actually done this?

If the requirement is “has built systems from scratch in ambiguous environments,” the question is not “tell me about a time you worked in a fast-paced environment.” That can be answered by anyone. The question is: “Describe the last process you built where nothing existed before you started. What was the context, what did you create, and how did you know it was working?”

That question fails for candidates who have only inherited existing systems. It succeeds for candidates who have built new ones. The question is testing the specific thing, not the category it belongs to.

Step 3 designs the follow-up for each question

Every primary question needs one prepared follow-up. The follow-up is what turns an initial answer into a useful signal.

The follow-up tests depth and specificity. If the candidate’s answer is detailed and specific, the follow-up confirms one more thing about what they built. If the answer is vague or general, the follow-up applies gentle pressure: “Can you be more specific about what that process actually looked like?”

A candidate who was vague on the first answer and vague on the follow-up has told you something important. A candidate who gives a detailed first answer and can go deeper on the follow-up has told you something equally important. The follow-up is where you learn which kind of answer you’re dealing with.

How the Conductor Compiles Your Interview Design

Before the first interview is scheduled, the Conductor can help convert the role’s requirements into the specific questions that would actually test for them.

The Living Library holds the role’s success definition, the 90-day criteria, and records of how past hires in this function have performed. The Conductor, Kiluma’s context-aware AI, reads from those records to compile question suggestions that are specific to this business’s version of the role.

Ask it: “Based on the role definition and what past hires in this function have required, what interview questions would specifically test for the three most important requirements?” It reads the role data and returns questions designed around what has mattered in this role, not generic interview templates.

You review and refine the suggestions. The Conductor’s starting point is grounded in what this business knows about the role, which makes the refinement faster than starting from nothing.

Test Each Question Before You Ask It

Before the first candidate arrives, test each question: can it only be answered by someone who has done this specific kind of work?

If the answer is yes, the question is too general. Sharpen it by adding the specific context, the specific output, or the specific constraint that is unique to this role.

A general question becomes specific by adding what “doing it here” actually means. “Tell me about leading a team” becomes “Tell me about leading a team where you were the first manager anyone on the team had ever reported to.” The second version can only be answered well by someone who has been a first-time manager for a team with no prior management experience.

The Test Every Interview Question Must Pass

Can each question only be answered well by someone who has done this specific work? If not, sharpen it until it can be. The interview should require the work, not just the vocabulary of it. Try Kiluma free for 14 days at kiluma.ai.