Interview order can affect judgment in some settings. What no credible source can give you is one percentage that applies to every hiring process.

The research is mixed because “interview order” can describe several different things. It can mean who an interviewer meets first, how recent candidates are when the debrief begins, the order in which evidence is reviewed, or who speaks first in the meeting. Those are different mechanisms, and the studies behind them do not measure the same outcome.

Metaview’s corpus does not contain a candidate-sequence analysis. It cannot show that the first candidate receives more strong-hire ratings, that the last candidate is hired more often, or that notes remove a measured share of an order effect. The corpus also stops before post-hire outcomes, so it cannot tell you whether a final decision was right.

The honest answer is therefore limited: order can shape ratings and recommendations, but the size and direction depend on the setting. A team can still make the process less vulnerable by using a job-related rubric, collecting independent evidence promptly, and keeping candidate-to-candidate comparison out of the scoring standard.

Where interview order can enter the evaluation

Imagine an interviewer meets four candidates for the same role. The first is unusually polished. The second gives a solid answer, but now sounds weaker by comparison. If the team has no clear rating anchors, the first candidate can quietly become part of the scale.

That is one possible contrast effect. It is not the same as recency, fatigue, or group influence.²

Order can enter a hiring process in at least four places:

• Candidate sequence: who the interviewer meets immediately before or after someone else.
• Information sequence: whether the interviewer sees the resume, work sample, application review, or another interviewer’s notes first.
• Timing: how much time passes before the interviewer records a rating or joins the debrief.
• Group discussion: whose conclusion the panel hears before reviewing the rest of the evidence.

The distinction matters. A study about applicants shown in sequence does not establish that the last candidate in a week gets the offer. A study about MBA admissions recommendations does not automatically transfer to a software engineering loop. A concern about the first speaker in a debrief needs meeting data. It does not need interview-transcript data.

The practical vulnerability is discretion. When the criteria are vague, interviewers have more room to use the previous candidate, the freshest memory, or another person’s opinion as the reference point. A structured process does not guarantee that order stops mattering, but it gives the team a standard against which those judgments can be challenged.

The primacy and recency effects in interview research

The research does not support a simple rule in which the first candidate benefits from primacy and the last candidate benefits from recency.

Three studies show why the answer remains conditional:

The field study is especially useful because its interviews were highly structured. Questions came from a job analysis, three-person panels used rating scales tied to benchmark answers, and the researchers still examined actual interview time and schedule position. It found no appreciable effect.³

That does not prove structure caused the null result. The studies differ in population, setting, design, and outcome. What it does show is that order effects are not inevitable and that no single multiplier belongs in a hiring article.

The admissions study also needs careful wording. It found narrow bracketing in recommendations within a day. It did not show that the last applicant was favored, and it did not measure final employment decisions.¹

How to keep the rubric ahead of the running order

The most useful safeguard is decided before interviews begin: define what the role requires and how evidence will be scored.

The quote describes Deel’s experience. It is not a measured order effect. It still captures the process problem: when criteria and evidence are weak, the debrief has more room to rebuild the assessment from memory and opinion.⁷

The U.S. Office of Personnel Management defines structured interviews around job-related competencies, predetermined questions, and a common rating scale with the same standards for acceptable answers. Its guidance also says higher levels of structure generally improve rater agreement and reliability.⁴

A practical version looks like this:

• Define the competencies and rating anchors before the first interview.
• Give candidates for the same role comparable core questions and comparable opportunities to provide evidence.
• Score each competency against the written standard, not against the strongest or weakest candidate seen that week.
• Record the specific example behind each rating.
• Complete the initial scorecard before reading other interviewers’ conclusions.
• When a rating changes after the debrief, preserve the original view and record the new evidence that changed it.

Interviewers do not need to fill every field while the candidate is speaking. They do need to document the evidence promptly, before the debrief and before other opinions can become part of their memory of the interview.

Metaview’s corpus includes 811,298 submitted scorecards. The median was submitted 2.32 hours after the interview ended, while the mean was 36.16 hours. That gap is because of a long tail. That describes the timing of scorecards that were submitted. It does not show that faster submission reduces order effects, and it says nothing about why other scorecards were never submitted.⁵

A separate canonical analysis found that 41.9% of 296,555 advancing candidates had no submitted scorecard when they moved forward. That is a documentation finding. It is not an interview-order finding. It matters here because an independent rating cannot anchor the debrief if it was never recorded.⁵

Debrief design that starts from written evidence

A debrief should combine independent observations. It should not be the first place those observations take shape.

A simple structure helps:

1. Confirm which required scorecards are present. Treat missing feedback as missing. Do not treat it as agreement.
2. Restate the role criteria and rating anchors before discussing the candidate.
3. Review evidence by competency before asking for an overall recommendation.
4. Separate negative evidence from missing evidence. “We did not test this” is not the same as “the candidate performed poorly.”
5. Ask the interviewer with the strongest disagreement to explain the evidence behind it.
6. Make the decision against the role standard. Avoid using another candidate as the benchmark.
7. Record the final decision and what evidence changed any initial rating.

These are process safeguards. They are not findings from Metaview’s corpus. The data does not show that the first speaker controls a debrief, that seniority carries a measurable multiplier, or that this meeting structure eliminates an order effect.

The purpose is more modest. When the independent evidence exists before the meeting, the group can test it. When it does not, the debrief has to reconstruct the interview from memory, which makes the final reasoning harder to inspect later.

How Metaview captures contemporaneous evidence

Metaview records and transcribes captured interviews, then prepares editable notes from the conversation. Reviewers can jump from a note to the transcript segment it came from, correct errors, add context, and use the record to complete a scorecard.⁶

That is a practitioner describing a workflow benefit. It is not evidence that Metaview removes interview-order bias.⁷

Where an ATS workflow supports it, Metaview can autofill objective, factual scorecard sections from the captured conversation. Subjective fields, ratings, and hiring recommendations remain with the interviewer. Metaview does not score competencies when they appear, decide whether an example is strong, or make the hiring decision.⁶

Reports can filter captured conversations using calendar data, conversation content, ATS properties, and custom properties. Teams can add structured AI columns, build charts, subscribe to alerts, and export a CSV.⁶

That can help assemble the records for a local audit, but the analysis still has to define:

• what counts as one candidate sequence, such as interviewer-day, panel-day, or role-week;
• which rating was recorded independently and which decision was final;
• how role, stage, interviewer, rescheduling, and missing scorecards are handled;
• whether the outcome is an interview rating, an advancement decision, or something else;
• what statistical uncertainty is acceptable before calling a pattern real.

Metaview does not currently provide a native interview-order-bias score. Reports can help organize and export the underlying records. The hiring team or analyst still has to design the comparison and interpret it.

Interview order will always exist. The goal is to keep it from becoming an unwritten part of the rating scale.

Sources

¹ Uri Simonsohn and Francesca Gino, “Daily Horizons: Evidence of Narrow Bracketing in Judgment From 10 Years of M.B.A. Admissions Interviews,” Psychological Science 24(2), 2013. More than 9,000 MBA admissions interviews.

² Steven J. Cesare, Anthony Dalessio and Richard J. Tannenbaum, “Contrast Effects for Black, White, Male, and Female Interviewees,” Journal of Applied Social Psychology 18(15), 1988. Simulated interviews, n=120 undergraduates.

³ Michael A. Willihnganz and Lawrence S. Meyers, “Effects of Time of Day on Interview Performance,” Public Personnel Management 22(4), 1993. Records from 818 actual structured employment interviews.

⁴ U.S. Office of Personnel Management, Structured Interviews and related structured-interview guidance.

⁵ Metaview canonical 2026 corpus analysis: SCORECARDS-4, submission timing among 811,298 submitted scorecards; DECISION-8, 41.9% of 296,555 advancing candidates with no submitted scorecard. Neither analysis measures candidate order.

⁶ Metaview Help Center: Notes Overview; Ashby integration and browser extension; Reports Overview.

⁷ Metaview case studies: Beyond gut-feel: How Deel scaled their hiring with Metaview Reports; How SoSafe improved screening signals and hiring consistency with Metaview.