I vividly remember one of the first interviews of my recruiter career. The candidate was due in ten minutes, and I was reading through her resume with my boss, the HR Manager. My boss noted how pretty she was, based on her LinkedIn profile picture - so we ended up scrolling through her social media pages.
Those ten minutes embarrass me to this day. Comments were made that should have banned both me and my boss from any recruitment work for good.
But here we are. Me building recruitment software, and my old manager somewhere out there, an even bigger boss for a bigger team.
The average mistake
The horrible reality is this: during the years I spent as a recruiter - working with tens of hiring managers across multiple companies and customers, building a network of hundreds of recruiters - I realized that my mistake was quite an average one.
It's bad out there. And most of the problems I've seen relate to how people handle the good ol' job interview: the preparation, the interviewing, and the evaluation.
Not everyone stalks their candidates on social media and lets it cloud their judgment. But people have their trusty interview "toolboxes": They evaluate the firmness of a handshake.
They ask about spirit animals.
And about strengths and weaknesses, without any idea what the answers should even be.
They reject people over gender, race, age, and other factors that have nothing to do with the job.
And they end up making decisions based on - what else - gut feeling. Because that's all they have.
Counting out the obvious mistakes, the way people prepare interviews sounds perfectly innocent. You dig up that Google Doc where you keep the questions that worked well before. You read your candidate's resume and notice something you want to ask about. You Google, or ask ChatGPT, for some effective questions. Or your HR team hands you a list of culture-fit gotchas you're supposed to use.
Or, if you're extroverted enough and socially gifted, maybe you just wing it, because you handled it well last time. You're curious and emotionally intelligent. Surely you can handle the task of asking questions.
We've all winged one. Like instant noodles: fine once, but a problem as a diet.
Why we disregard a century of research
The problem is what research says about that confidence. In one study, interviewers stayed just as confident in their judgment even when the interviewees had been instructed by the researchers to answer questions randomly. The interviewers didn't notice - they rated the nonsense interviews as no less informative than the honest ones. Worse, forming impressions from a free-form interview made predictions less accurate than using background information alone - the conversation diluted the signal that was already there. The paper's title says it all: belief in the unstructured interview is "the persistence of an illusion" (Dana, Dawes & Peterson, 2013).
Daniel Kahneman tells the constructive version of the same story in Thinking, Fast and Slow. As a young officer, he replaced the Israeli army's intuitive interviews with structured, trait-by-trait scoring, and predictions improved measurably. His conclusion wasn't that intuition is worthless - it's that intuition adds value only after disciplined data collection, not instead of it.
So why does winging it survive decades of evidence like this? Because interviewing feels like a skill you already have, and the feedback loop that would tell you otherwise is invisible. You never see how the people you rejected would have done. Every hire that works out confirms your eye for talent. "The interview felt great" is a fact about the conversation, not the candidate.
Unstructured interviews are like driving on a familiar road without a seatbelt. It feels comfortable. Until a truck shows up from behind the blind turn.
What actually predicts a good hire
Which selection methods actually predict job performance is one of the most-studied questions in applied psychology - and in 2022, Sackett, Zhang, Berry and Lievens published the modern revision of 85 years of that research, correcting a statistical inflation that had crept into decades of earlier estimates. When the dust settled, one method sat on top of the entire ranking: the structured interview.
Validity - what does that even mean? You don't need to care about validity coefficients to act on them. The plain translation: from the same hour of everyone's time, a structured interview extracts about twice as much signal as an improvised one. If you could keep only one touchpoint with a candidate - one hour to decide anything at all - the evidence says to make it a structured interview.
This conclusion isn't new, only newly precise. Schmidt and Hunter's landmark 1998 synthesis already had structured interviews at the top of the table (.51 vs .38 for unstructured). The 2022 revision sharpened the numbers and dethroned cognitive ability tests; it didn't change the direction.
Two honest caveats, because numbers deserve them. The exact decimals are contested - some researchers argue the 2022 corrections went too far - but the direction is robust across every serious analysis: structure roughly doubles validity. And that .42 is an average that hides a wide spread.
How well you structure matters enormously. Which raises the obvious question: what exactly is structure?
Structure is a recipe, not a script
Say "structured interview" and people picture a robot reading questions off a clipboard, forbidden to smile. That's the strawman. Structure is not a script, and it's not a personality test. Distilled from the classic taxonomy of interview structure (Campion, Palmer & Campion, 1997), five moves carry most of the weight:
- Decide what the job actually requires - before you meet anyone (job analysis)
- Ask every candidate the same questions, in the same order
- Ask about past behavior, and probe with planned follow-ups
- Score each answer against criteria written in advance - what a good answer contains, not how confident it sounded (rating scales)
- Rate independently, combine interview ratings mechanically - discuss after, not before
Now look at that list again. It's a recipe. First, obviously, you decide what you want to eat. You choose the ingredients before you start cooking, and the steps come in a deliberate order. A recipe doesn't make cooking mechanical - it's how good cooks stop improvising the parts that matter, so their attention is free for the part that does need judgment: the person in front of them. Structure isn't the opposite of judgment. It's what judgment stands on.
And if this sounds like theory, one very big kitchen already cooks this way. Google runs hiring on structured interviewing: vetted question banks, scoring rubrics, interviewer training, and a "Rule of Four" - after four structured interviews, additional interviewers barely improve the decision.
The objections, taken seriously
"It kills rapport - I want a conversation, not an interrogation." A good actor sounds natural because they rehearse their lines. Warmth lives in the delivery, not in improvised questions. Nothing about asking every candidate the same things forbids you from being human between them. And candidates consistently experience consistent, job-relevant processes as fair - fairer than a friendly chat whose rules they can't see (Levashina et al., 2014 reviews the reactions research; Sackett's team weighs applicant reactions explicitly in their 2023 follow-up).
"Great interviewers don't need it." That confidence is precisely the illusion the Dana study measured - it survived even random answers. And Kahneman won a Nobel for knowing when intuition fails.
"It's too much work for a team our size." This one is fair. Preparation is the honest cost of structure: a good interview, like a good meal, is mostly preparation. Hold that thought.
One more thing structure buys you, and it matters. Remember the rejections over gender, race, and age? Structure narrows exactly those openings. Same questions close the different-interview-for-different-people channel; anchored scoring replaces vibes with job-relevant criteria; independent ratings stop the loudest voice in the debrief from deciding. The research finds meaningfully smaller demographic gaps in structured than in unstructured interviews. Structure doesn't eliminate bias - nothing does - but it closes the doors bias usually walks through.
The junk-food era
Two things are happening to hiring at once. AI has made it effortless to generate interview questions - plausible-sounding ones, in seconds, vetted by no one. And AI-era screening, from one-way video interviews to chatbots, is degrading the candidate experience so visibly that candidates increasingly walk away from processes that feel automated. Faster junk, in both directions.
The fix for junk-food interviewing isn't a faster microwave. It's the opposite move: a well-prepared human in a room, asking questions chosen on purpose, scoring answers against criteria that exist. In a market automating itself into sameness, a well-run structured interview is becoming something rare - a reason for the best candidate to pick you.
Why this website exists
The evidence for structure has been settled for decades. It still loses to winging it, every single day, for one reason: preparation is the bottleneck. The old Google Doc is right there. The blank page of a proper interview plan is not.
That's why I believe this website needs to exist.
Interview Recipes compresses the preparation. Every question in the pantry is a vetted recipe card - not just the question, but why it works, the follow-up probes that get past rehearsed answers, what a good answer looks like, and the red flags of a weak one. You pick the competencies the job needs, save a handful of cards into a recipe, and you walk into the room with a clear picture of what the interview will be. The structure - same questions, planned probes, anchored scoring - comes built in.
Here's an example ingredient:
Everything is free to read - every question, every guide, this manifesto. An account just saves your recipes.
The invitation
This site is a bet that enough of us are done winging it. The beliefs behind it are simple:
- Evidence over instinct.
- Preparation over improvisation.
- Same questions over pet questions.
- Scoring the answer over liking the answering.
- The best candidate over the best talker.
If you're hiring right now, browse the ingredients and build your first recipe - it takes minutes, and you'll walk into your next interview knowing why every question is there. If you're not hiring this month, leave your email below instead: one message when new recipes come out of the oven, nothing else.
And if you know the person who runs hiring where you work - the one with the trusty Google Doc - send them this page. It might be the seatbelt they didn't know they were missing.