The first interview guide I ever created was an Excel sheet, and it was a monster. This was 2016, at a virtual reality startup in Tokyo where the median age was somewhere in the twenties, which meant engineering team leads who were brilliant at their work and had never hired anyone in their lives. We had no ATS and no software to help with any of this. We had me, and we had Excel. So I built the whole method into one workbook: a question bank broken down by competency, a job analysis section, sheets feeding other sheets. It was the most impressive document I had ever prepared, it was unnecessarily complex, and the user experience was terrible.
Everybody loved it.
Not for the spreadsheet craftsmanship. The team leads, interviewing for the first time in their careers, got a script to follow in a situation where they had nothing. All of us got a shared language: when an engineer and a recruiter discussed a candidate, they were suddenly discussing the same competencies, tested by the same questions. The more experienced interviewers started improving the system with me, and the workbook grew into a methodology document that connected the engineering teams to recruitment. Former colleagues still thank me for it, a decade and much better tooling later.
That is what an interview guide does when it works, and this article is where this path's preparation phase comes together. The earlier steps produced the parts: a job analysis that names what the role requires, questions of the right type for each competency, anchored criteria for scoring the answers, and planned probes for getting past the rehearsed surface. The guide is where the parts become an interview. This article covers what an interview guide is, what goes on the page, and how to assemble one, whether you are a hiring manager preparing your own interview or the person preparing guides for interviewers who have never run one.
What is an interview guide?
An interview guide is the document the interviewer takes into the room: the questions to ask, in order, with each question's follow-up probes and scoring criteria printed beside it, space for notes, and a short script for opening and closing the conversation. The same document, with the same questions, for every candidate interviewing for the role.
That definition sounds almost too modest to matter, so it is worth being clear about what a guide is not. It is not a list of questions. Question lists are the one thing hiring has never lacked; every hiring manager has the trusty document of questions that worked before, and the internet will hand you a hundred more in a minute. What is scarce is the structure around the questions, and the guide is where that structure lives. The difference is the difference between owning ingredients and holding a recipe.
It is also not a script that turns you into a robot. The guide fixes the parts of the interview that should never be improvised, the questions, the order, the criteria, precisely so your attention is free for the one thing that does need judgment: the person in front of you. The whole argument for structure rests on that trade, and the guide is the physical form the trade takes.
Why the guide is written down
The research case behind this path is blunt: structured interviews predict job performance about twice as well as unstructured ones, .42 against .19 in the modern re-analysis of selection research (Sackett et al., 2022). But that .42 is an average hiding a wide spread, and where your interview lands on the spread depends on how much of the structure actually happens.
Now look at what "structure" is made of. In the classic taxonomy (Campion, Palmer and Campion, 1997), the content half of interview structure is a series of promises about behavior in the room: every candidate gets the same questions, in a controlled order; ad-hoc prompting is limited to planned probes; the candidate's own questions wait until the end. Every one of those is easy to intend and easy to break at minute forty of a warm conversation. The guide converts intentions into ink. You do not have to remember the next question, resist the tangent, or reconstruct the criteria from memory, because the page in front of you already did.
There is a quieter benefit, the one my Tokyo team leads taught me. A written guide is interviewer training in the cheapest form it comes. A first-time interviewer holding a good guide runs a defensible interview on day one, because the hard decisions were made for them in advance. And when several people interview for the same role, one guide means one instrument: the candidate's score stops depending on which interviewer they happened to draw.
What goes in a structured interview guide
The best public reference for guide anatomy is the US federal government's practical guide to structured interviews, public domain and used across federal agencies. Its interviewer packets carry the same parts yours should, top to bottom:
The job analysis, on top. The role, the outcomes in your own verbatim words, and the must-assess flags, exactly as the job-analysis step produced them. This is not decoration. Interviewers who know what the interview is for probe better and drift less, and when scores are compared later, the top of the page is what they are compared against.
An opening script. Three or four sentences, written out: a warm welcome, what will happen ("I'll ask everyone the same questions, and I'll be taking notes"), and that there is time reserved at the end for the candidate's questions. Writing it out is not ceremony. Holding candidate questions until the end is one of the content components of structure, because ten opening minutes of free conversation give every candidate a different interview; the script is how the interview starts the same way every time. Telling the candidate about the notes and the shared questions costs nothing and reads as fairness, which candidates consistently reward.
The question sections. One section per competency this interview covers, each holding two or three question slots. A complete slot is question, probes, and anchors printed together: the behavioral question that asks for the story, the three or four probes that fill it out, and the what-good-looks-like and red-flag anchors the answer is scored against. Keeping the three on the same page is the point of the layout: the probe you need and the anchor you are scoring should never live in a different document than the question you just asked.
Note space and the scale. Room to write under every question, because scoring happens against notes, right after each answer, on the shared four-point scale printed once on the guide. An answer scored at the end of the interview is an impression with a memory problem.
A closing script. "Is there anything else you would like us to know?", the candidate's questions, and the concrete next step with a date attached. Two minutes, written down, so no candidate leaves without it.
Timings in the margin. Minutes per section, decided in advance. The schedule is part of the standardization: a candidate who gets twenty-five minutes on a question that the previous candidate got ten on was not, in any meaningful sense, asked the same question.
How to create an interview guide, step by step
If you have followed this path, creating the guide is an assembly job, and a fast one.
- Put the job analysis on top. Half a page: role, outcomes verbatim, must-assess flags. If you skipped that step, do the three minutes now; a guide built without it is a question list with margins.
- Fill the slots with questions. Two or three per competency, behavioral as the spine, situational variants where juniors need a door. Take them from the pantry or write your own.
- Print the anchors beside each question. What a good answer contains, what the red flags sound like, in observable behavior. Ten minutes per question, and the ten minutes that most decide what your scores will be worth.
- Attach the probes. Three or four per question, neutral and fact-finding, written against the anchors.
- Frame it. Opening script, closing script, note space, timings.
Step five is where the arithmetic gets honest. An hour-long interview holds about sixty minutes: five for the opening, ten or so protected for the candidate's questions and the close, which leaves roughly forty-five for measurement. A behavioral question, probed properly and scored on the spot, costs six to eight minutes. That is five to seven questions, total. Not per competency. Total.
Run that arithmetic against your slots and something usually has to give. Give up questions, never the candidate's ten minutes or the scoring time; a guide that ignores its own arithmetic is a wish list, not a plan. And when the cuts start to hurt, that pain is information. It means the role needs more interview than one hour holds, which is the next section's problem to solve properly.
An interview guide example: the support lead
The worked example this path has carried since the job-analysis article: a support lead for a forty-person company, four competencies, customer orientation flagged must-assess. Here is that role's single-interview guide, as the hour's budget allows:
| Minutes | Section | On the page |
|---|---|---|
| 0–5 | Opening | The script: welcome, same-questions-and-notes explanation, questions at the end |
| 5–20 | Customer orientation (must-assess) | Two question slots, probes and anchors beside them |
| 20–28 | Integrity | One question slot |
| 28–36 | Planning and delivery | One question slot |
| 36–44 | Leadership | One question slot |
| 44–55 | Candidate's questions | Held to the end, on purpose, and protected |
| 55–60 | Closing | Anything else to add, next step with a date |
Five questions, every one scored against its anchors before the next begins, and the must-assess competency the only one that gets depth. Here is its first slot, filled, exactly as it comes out of the pantry:
Question, probes, anchors, all on one card: the slot needs nothing else, which is why assembling a guide from vetted cards takes minutes while writing one from scratch takes an afternoon.
One job, one plan, one guide per interview
Look at that schedule again and the squeeze is visible: three of the four competencies got a single question each. One question is a reading, not a measurement. The temptation is to cram, to trim the probes and squeeze in a sixth and seventh question, and the arithmetic has already told you what that costs.
The better move is to split. You analyzed the job once, and the analysis names the requirements; how many interviews it takes to test them honestly is a separate decision. Give the process two or three focused interviews, each owning two or three competencies, each with its own guide. The support lead's loop might be an hour on customer orientation and integrity with the hiring manager, and a second hour on leadership and planning with the person the new support hires will sit closest to. Google's structured interviewing program found that beyond four interviews, additional interviewers barely change the decision, so a short loop is not a compromise; it is the shape the evidence points at.
Write the split down as part of the plan: which interview owns which competency, and who runs it. That single written line is what stops a competency from being double-covered by two interviewers while another falls through entirely, and it keeps the promise the formats article made, that dividing the work across interviews is the guide's job to fix in writing. Every candidate then meets the same loop: same interviews, same guides, same order.
The guide that gets used
My Tokyo workbook broke every usability rule I have learned since, and its flaws are still the four ways guides fail.
Too heavy to use. Mine was a fortress of interlinked sheets, and it survived only because I was in the building to explain it. A guide has to survive a nervous first-time interviewer's solo run: a few printed pages, one question per stretch of the hour, nothing that needs a manual. If preparing it feels heavy, remember the split; if using it feels heavy, cut.
Read like a courtroom transcript. The scripts open and close the interview, the questions come in order, and everything between the lines stays human. You may rephrase, react, and be warm; you may not swap questions, invent probes, or reorder the interview for a candidate you like. Fixed skeleton, human delivery.
Questions without anchors. A guide whose slots hold only questions produces the same interview it replaced, with better formatting. The validity evidence belongs to the full discipline, questions and probes and anchored scoring together; a question list alone earns no .42, and it is better to anchor three questions than to ask seven bare ones.
Silent drift. Pet questions creep back one "just curious" at a time, and a question quietly reworded between candidates splits your process into incomparable halves. The guide can change, deliberately: if the first interviews teach you the analysis missed something real, revise the guide in writing and use the revision for every remaining candidate, the same way the job-analysis article treats the analysis itself as a hypothesis to update. What it cannot do is change silently, per candidate, per mood.
From a monster workbook to three minutes
The workbook cost me more evenings than I care to admit. You should not need them: every card in the pantry is a guide slot ready to print, question, probes, and anchors on one card, and the planner walks the whole preparation phase in one pass, the job analysis, the split into focused interviews, and a printable guide for each, free, in about three minutes, no account needed until you save.
However you build it, build it before you meet anyone. What my old colleagues still thank me for was never the spreadsheet. It was that someone finally decided what our interviews were for and put it on paper everyone could hold. That is the entire trick. The guide is where your preparation stops being a state of mind and becomes a document, and the document is what walks into the room with you.