The Last Resume Wins: How Recency Bias Is Quietly Sabotaging Your Hiring Decisions

What Recency Bias Is

Recency bias — the tendency to weight recent information more heavily than earlier information — is one of the most well-documented cognitive biases in psychology. It manifests in many domains: people remember the last item in a list more readily than items in the middle (the recency effect in memory research), juries are influenced by the last argument they heard, and performance reviewers tend to weight recent performance more heavily than a full year of data.

In hiring, recency bias operates through two distinct mechanisms. The first is simple memory: when a reviewer has assessed 50 candidates over several days, the candidates reviewed most recently are simply more available in memory than those reviewed earlier. The details of candidate number 12 have faded; candidate number 48 is vivid. This memory asymmetry affects how reviewers compare candidates when constructing a shortlist.

The second mechanism is contrast effects. Each candidate is not assessed in isolation — they are assessed relative to the candidates the reviewer has most recently seen. A strong candidate who follows a weak one will be rated more positively than the same candidate would be if assessed independently, because the contrast with the weak preceding candidate elevates their relative standing. A strong candidate who follows another strong candidate may be rated more harshly for the same reason.

The Research Evidence

The evidence for recency effects in hiring is robust and has been replicated across multiple study designs and contexts. Research published in organisational behaviour journals has examined resume review, structured interviewing, and assessment centre evaluations — and found recency effects present in all three contexts.

Studies examining resume review have found that identical resumes placed at different points in a review sequence receive systematically different ratings. Candidates presented later in a sequence tend to receive higher ratings than candidates presented earlier, even when the content of their application is identical. The effect is small for any individual candidate comparison, but across a pool of hundreds of applications, it meaningfully distorts the ranking.

Research on structured interviews — conducted face-to-face in a consistent format — has found similar effects. Interviewers who conduct multiple interviews in a single day show increasing leniency as the day progresses, followed by increased stringency as fatigue sets in. The time of day at which a candidate is interviewed has been shown to correlate with their rating in ways that should be impossible if the assessment were objective.

A notable study examining judicial bail decisions found that judges were significantly more likely to grant parole immediately after a food break than before one — a finding that generated considerable attention but is consistent with a broader body of research showing that cognitive state at the time of assessment matters as much as the content of what is being assessed.

Why High-Volume Hiring Amplifies the Effect

Recency bias is present in all human assessment — it cannot be entirely eliminated. But its magnitude varies with conditions. The two conditions that amplify recency effects most strongly are volume and fatigue — and both are defining features of high-volume hiring.

When a reviewer assesses five candidates, the early candidates are still relatively fresh in memory when the later ones are being evaluated. The recency effect is present but limited. When a reviewer assesses fifty or a hundred candidates over several days — as routinely happens in high-volume hiring — the memory and contrast effects are dramatically larger. Candidates reviewed on day one are not just less fresh in memory; they are effectively invisible by day four.

Fatigue compounds the problem. Research on decision-making quality consistently finds that cognitive performance degrades as decision-making fatigue accumulates. Reviewers who have assessed many candidates become less thorough in their evaluation, rely more heavily on heuristics, and show greater susceptibility to irrelevant contextual factors — including the order in which candidates were reviewed.

The practical implication is striking: in a process where 300 applications are reviewed over the course of a week, the candidates reviewed on Monday morning may be at a systematic disadvantage relative to those reviewed on Thursday afternoon — through no fault of their own and for no reason related to their actual qualification for the role.

The Strategic Application Problem

If recency bias were unknown, it would simply be an invisible source of noise in hiring decisions. The more troubling reality is that it is increasingly known — and some candidates and candidate coaches are beginning to exploit it strategically.

Career coaches in professional communities increasingly advise candidates to apply to roles after the initial rush — to wait a week or two after a role is posted before applying, on the theory that the reviewer will be less fatigued and that the candidate will be reviewed more recently relative to the interview process. Whether this advice is correct is difficult to verify empirically, but the fact that it is being given reflects a growing awareness that the timing of an application affects its outcomes independently of its content.

If the timing of an application influences its success, then the hiring process is partially selecting on candidates' knowledge of reviewer psychology rather than on their actual qualifications. This is a poor outcome for everyone: strong candidates who apply promptly because they are genuinely interested may be disadvantaged relative to strategic late applicants who are less interested but better coached.

Organisational Responses and Their Limits

Organisations that are aware of recency bias have developed several responses. Structured review processes — where all candidates are assessed against a fixed rubric before any comparisons are made — reduce but do not eliminate the effect. Blind review — where identifying information is removed before assessment — addresses some biases but not recency effects, which operate on content rather than identity. Panel review — where multiple reviewers assess candidates independently before comparing notes — dilutes the effect but multiplies the time cost.

None of these interventions address the fundamental driver of recency effects in high-volume hiring: the sheer number of assessments that each reviewer is asked to make, and the fatigue that accumulates across them. As long as a single reviewer is responsible for assessing hundreds of candidates, recency effects will influence outcomes.

The only structural solution is to reduce the number of assessments that fatigued humans are asked to make — which means moving more of the assessment load upstream, before human reviewers are involved, and ensuring that the upstream assessment is conducted consistently regardless of sequence.

Consistency as a Feature, Not an Afterthought

An AI interviewing system does not get tired. Candidate number 287 receives the same quality of interview as candidate number 3. The questions are the same, the probing is as deep, the scoring criteria are applied as consistently. There is no Monday morning effect. There is no post-lunch leniency. There is no end-of-day stringency. The sequence in which candidates are assessed has no bearing on the quality of the assessment they receive.

This consistency is not just a nice-to-have quality — it is a fundamental requirement for fair assessment. If the goal is to identify the strongest candidates regardless of when they applied or where they fell in the review sequence, then the assessment mechanism must be immune to sequence effects. Human assessment, under high-volume conditions, is not. AI assessment, by construction, is.

There is an additional benefit that is rarely noted: consistent assessment is also more legally defensible. In an environment where hiring discrimination claims are increasingly common, an organisation that can demonstrate that every candidate was assessed against the same criteria, by the same process, with no variation attributable to reviewer fatigue or sequence effects, is in a significantly stronger position than one relying on ad hoc human review.

The Feedback Dimension

Recency bias has one further consequence that is rarely discussed: it makes the feedback that candidates receive — when they receive any — less reliable. If a candidate was rated poorly partly because they were reviewed on day one when the reviewer was fresh and standards were higher, and partly because they were sandwiched between two unusually strong candidates, the feedback they receive about why they were unsuccessful may not accurately reflect their actual performance relative to the successful candidate.

This matters for candidates who want to improve. If the feedback they receive is partly an artefact of sequence effects rather than genuine performance assessment, then acting on that feedback may not improve their outcomes in future processes. They are trying to correct for noise rather than signal.

Consistent AI assessment changes this. When every candidate is assessed against the same criteria by the same process, the feedback each candidate receives is a genuine reflection of their performance against those criteria — not an artefact of when they happened to be reviewed. Candidates can act on this feedback with confidence that it reflects something real about their performance.

Key Takeaways

• Recency bias consistently inflates ratings for candidates reviewed later in a sequence
• High-volume hiring amplifies recency effects through increased review fatigue and longer sequences
• Candidates are beginning to exploit known recency effects strategically, selecting on reviewer psychology rather than role fit
• Structural solutions require moving assessment load away from fatigued human reviewers
• AI assessment is sequence-immune by construction, delivering consistency that human review cannot provide at volume
• Consistent assessment also produces more reliable candidate feedback — a benefit for candidates and organisations alike