Case Study

7,088 resumes.
603 human interviews.
42 hires.

The right people were already in the pile. This is what it took to surface them, and everything nobody had to do to get there.

What this is. Three months of real screening for a Customer Service Representative role, the kind of seat most organizations fill over and over again. Run repeatedly across multiple client organizations by the staffing operator that licenses BrunoIQ. The figures are a composite of those engagements, not one company's hiring push. Clients and candidates are anonymized.

The funnel

Every applicant was screened. Nobody was skipped because the pile got tall.

Resumes received7,088

Across three months, every one of them out of the clients' own job postings. Nothing was sourced or advertised on our side. The client runs the ads and points the applicant flow at the system; the system sorts it. Applying meant uploading a resume and answering the screening questions, so all 7,088 arrived with something to grade.

Passed the recorded video screen2,564

All 7,088 were screened and graded against the same questions. 2,564 of them cleared the bar, 36% of the pile.

Reached a face-to-face interview603

The first point at which a human spent time. 8.5% of the original pile. This wasn't a second filter: 2,564 people had already cleared the bar, and 603 is how many could actually be reached and booked. Scheduling closed that gap, not screening.

Recommended to the client73

The short list, drawn from the people who came through that interview well. Each one arrived with scores, transcripts, and the gaps flagged.

Offered and hired42

Hired by the client, on the client's decision.

Bars are to scale.

Two numbers worth sitting with

58%

of recommended candidates were hired

42 of the 73 people put in front of a hiring manager got the job. For comparison, a submit-to-hire rate between one in five and one in ten is the range commonly cited in staffing. When the short list said someone was worth meeting, the client agreed more than half the time.

91.5%

of the pile never reached a human interview

6,485 of 7,088 applicants were screened and graded without consuming a single interview hour. The 42 hires still happened. That is the entire proposition, measured.

"We don't source your candidates and we don't create better ones. The right people were already in your pile. We make sure they surface, and make sure the wrong ones don't take up your team's time."

What these numbers show

  • Every applicant was interviewed and graded against the same bar, at a volume no team could read through
  • The short list was dense enough that most of it got hired
  • Human time went almost entirely to people who had already cleared a standard
  • The same role, screened to the same standard, repeatedly, across different clients

What they don't show

  • That BrunoIQ predicts who will succeed in a job. It doesn't, and we don't claim it does
  • That better candidates existed because of us. They came from the clients' own job postings and would have been in the pile either way
  • That any hiring decision was made by software. Every one was made by a person
  • A validation study. This is one role over three months, not a controlled trial

Where it didn't work

A case study that ends at 42 for 42 is a brochure. Here's the part most companies leave out.

Six of the 42 were bad hires.

Roughly one in seven didn't work out. They were replaced, quickly, by the operator running the engagement, and those replacements are still in their roles today. No client relationship was lost across any of these engagements.

Worth being precise about what fixed that: a staffing operator with a service commitment replaced those people. Screening didn't. No screening process, ours included, can tell you who will still be in the seat a year from now, and any product claiming otherwise is selling you something it can't deliver.

See what one of those 73 looked like on paper.

The short list isn't a list of names. Every person on it arrives as a full dossier: scores, the transcript behind each one, the gaps flagged honestly, and the questions worth asking live. Take a real one, anonymized.