Nine Hundred Applicants, Fifteen People, and No Way to Interview Them All

The challenge

One posting drew 981 applications. A fifteen-person company cannot interview 981 people, and it cannot carefully read 981 resumes either — the arithmetic simply does not work. What actually happens at that volume is that someone reads the first fifty properly, skims the next two hundred, and the remainder are decided by whoever happened to apply early.

Where we came in

The honest comparison is not between a structured evaluation and a careful human reading of every application, because the second one was never going to happen. It is between a written rubric applied consistently and a tired person at application four hundred. We built the system to do the first pass only: it produces a grade and a ranking with its reasoning attached, and every interview and every hiring decision is made by a person.

What we did

• Take applicant resumes and supporting documentation into a single review platform
• Draw personality and technical-aptitude signals from our own learning and assessment platform rather than inferring them from a resume
• Evaluate each applicant against the written job description, alongside HR practice and established assessment standards
• Produce a grade and a ranking with the reasoning recorded, so a human reviewer can disagree with it
• Size the interview stage against how urgently the role needs filling, rather than a fixed number
• Keep every interview and every hiring decision with people — the system shortlists, it does not select

Where it landed

• A field of several hundred applicants becomes a shortlist a small team can genuinely interview
• Every applicant is assessed against the same written criteria rather than against reviewer fatigue
• Assessment evidence and resume evidence are considered together instead of separately
• Humans conduct all interviews and make all hiring decisions