Structure Is the Candidate Experience
The Most Expensive Thing in Hiring is a Vibe
We have convinced ourselves that the interview is a conversation. It is not. It is an evaluation wearing a conversation’s clothes, and everybody in the room knows it except the person doing the evaluating.
Here is the tell. Ask a hiring manager how the interview went and you will get an adjective. Sharp. Solid. Not quite there. Ask them what the candidate actually said in response to the question everyone else also asked, and you will get a pause, then a story about culture fit, then a second pause. The adjective was not a finding. It was a feeling, retrofitted with vocabulary.
Feelings are cheap to produce and expensive to defend. That is the whole ballgame.
The Data is Not Ambiguous, Which Is Rare and Should Be Enjoyed
Schmidt and Hunter’s 1998 meta-analysis, the one everybody cites and almost nobody reads, put structured interviews at .51 predictive validity against .38 for unstructured. Sackett and colleagues re-ran the math with cleaner corrections in 2022 and the gap widened rather than closed: roughly .42 for structured, .19 for unstructured. Huffcutt and Arthur graded interviews on four levels of structure and found validity climbing from .20 at the bottom to .57 at the top.
Translation: the unstructured interview, the format most senior leaders are proudest of, is closer to noise than to signal. You are not a great judge of people. You are a great judge of people who remind you of yourself, which is a different skill and one the market has stopped paying for.
Then there is the bias literature, which is where this stops being an efficiency argument. A meta-analytic comparison of the two formats found bias effects of d = .59 in unstructured interviews versus d = .23 in structured ones. Structure does not make you a saint. It makes you harder to fool, including by yourself.
Now The Part Nobody Puts On The Slide
Every number above is a quality-of-hire argument. Fine. Here is the argument I think is more interesting, because it is the one with compounding returns: structure is the best candidate experience intervention available, and it costs less than the swag budget.
Start with the state of the market. Greenhouse found 61% of candidates have been ghosted after an interview. Criteria Corp’s 2026 read put employer ghosting at 53% of job seekers in a single year, up from 38% in 2024. Greenhouse’s benchmark data pegged median time to first offer at 68.5 days. Roughly a quarter of North American job seekers describe their experience as great. Nearly half say they would withdraw from a process over communication alone.
We have built a system where the median outcome for a person who took a half day off work, prepped for six hours, and told their spouse to hope, is silence.
Now ask why. Not the cynical why. The operational why.
Nobody ghosts on purpose. Ghosting is what a decision looks like when it cannot get made. The debrief slips because two interviewers have not submitted feedback. The feedback has not been submitted because there is no rubric, and writing free-form prose about a stranger from memory is genuinely unpleasant, so it goes to the bottom of the list. The decision waits. The candidate waits. Day four becomes day fourteen. By day twenty-one, sending the email feels worse than not sending it, so nobody sends it.
Structure attacks that chain at every link. Same questions, same order, same rubric, scored independently before the debrief. Feedback takes eight minutes instead of forty because you are filling in a form, not composing an essay. The debrief starts with evidence instead of ranking whoever spoke most confidently in the hallway. The decision arrives while the candidate still remembers the conversation.
Speed is not a nice-to-have layered on top of candidate experience. Speed is candidate experience, and structure is its supply chain.
What the applicant-reactions research actually measures
Bauer and colleagues split perceived fairness into two components: structural and social. Structural fairness is whether the process felt job-related and whether you got a real chance to show what you can do. Social fairness is whether everyone was treated the same way, respectfully, with two-way information.
Read those definitions again. They are a description of a structured interview.
The objection I hear from executives is that structure makes it robotic, that you lose warmth and the candidate feels processed. I want to be careful here, because that concern is not stupid, it is just misdiagnosed. Candidates do not resent being assessed. They resent being assessed on criteria nobody will name. What reads as cold is not the rubric. It is the interviewer who did not read the resume, riffing for forty-five minutes about the last conference they attended, then rendering a verdict the candidate will never hear an explanation for.
The unstructured interview is not warmer. It is warmer for the interviewer. For the candidate it is a slot machine with better manners. CandE and Survale’s research keeps landing in the same place: employers running structured interviews earn higher candidate experience ratings and stronger fairness perceptions. That is not a coincidence. That is people responding well to being told the rules.
On the Lavalier Page
Which brings me to what provoked this. Lavalier’s quality-of-hire page makes the standard, correct case: define the competencies that matter for a role, generate role-specific interview plans with stages and questions and interviewer assignments, arm every interviewer with the same criteria and live guidance, auto-generate feedback so nobody chases scorecards, and produce comparable candidate summaries so debriefs land on a decision instead of a rescheduled meeting.
The headline claim is 2x hiring quality in half the time. I am constitutionally allergic to any metric with a 2x in front of it and no denominator behind it, and you should be too. Ask what quality means, ask over what baseline, ask who measured it.
But notice what the feature list actually is. Strip the marketing and every single item is a candidate experience feature that has been priced and sold as a quality feature:
Role-specific question plans mean the candidate is evaluated on the job, not on the interviewer’s mood at 4:45 on a Thursday.
Same criteria for every interviewer means the eighth candidate gets the same shot as the first.
Automatic feedback capture is the thing that turns a two-week silence into a two-day answer.
Comparable summaries mean rejections can be explained instead of merely delivered.
The second-order effect is the one that compounds: candidates who felt the process was fair accept more offers, refer more people, and say better things about you after you tell them no. Roughly seven in ten candidates with a bad experience tell someone about it. Your employer brand is not your careers page. It is the aggregate of what rejected candidates say about you at dinner.
Two footnotes worth flagging while we are here. Gartner found only 26% of applicants trust AI to evaluate them fairly, so any tooling in the loop needs disclosure, not stealth. And a tool can hand you a rubric, but it cannot make your VP of Engineering fill it out. Structure is a discipline. Software is a place to put it.
The Main Thing
The unstructured interview persists because it flatters the powerful person in the room. It costs the candidate their time, their dignity, and occasionally their belief that this whole thing is on the level, and it costs the company its accuracy. That is a remarkable trade to keep making in 2026 with the data this settled.
Write the questions down. Score them before you talk to each other. Tell the candidate what happens next and when, then do that thing on that day.
You will hire better people. That is the part that gets you the budget. The part that gets you the next ten candidates is that the ones you turn down will still take your call.