CASE STUDY — How a Melbourne event bartending agency built a repeatable way to screen for coachability, communication and other soft skills, and stopped guessing from resumes.
Company: The Fellas
Industry: Event bartending and hospitality staffing
Location: Melbourne, Australia
Size: Around 80 staff on a casual, event-based roster
Hiring model: Constant rounds — hundreds of events a year, seasonal peaks, regular turnover in a casual workforce.
The challenge: hiring for things a resume never shows
The Fellas staff bars at events. A wedding one night, a corporate function the next, a private party after that. Every event is a different room, a different layout, a different crowd. The bar gets built in the afternoon and packed down by midnight. The person behind it has to learn the setup fast, take direction from a shift lead they met an hour ago, and keep a queue of guests happy while they do it.
That is the job. None of it is on a resume.
When The Fellas looked at what actually separated a good casual from one they would not book again, the same two things came up. The good ones were coachable. You could correct them mid-shift and they adjusted without sulking. And they could communicate. They read the guest in front of them, handled a complaint without escalating it, and asked when they were not sure.
A resume asks for neither. Neither does a fifteen-minute phone screen, where a confident talker reads as a strong hire and a nervous one gets cut, regardless of who would be better on the floor. Filtering applicants by keywords told The Fellas nothing useful. Putting a chatbot between an applicant and a real assessment would have told them less.
And The Fellas hire constantly. Casual, event-based work means the roster turns over, peaks are seasonal, and there is always another round. Every round brought the same problem: a stack of applications that all looked fine on paper, a hiring lead with no time to call them all, and no reliable way to tell, before the first booking, who could take feedback and talk to a room.
Step 1: We rebuilt one of their stalls with VirtualShift
VirtualShift is a 20-minute trial shift a candidate completes from their own device. They do not read about the job. They do it. So the first step was making the trial look and feel like a real night with The Fellas.
We did not drop them into a generic bar. We built one of theirs.
Working from photos, floor plans, and a walkthrough with their team, we recreated one of The Fellas' actual event stalls in 3D. The layout of the bar. The position of the speed rail and the garnishes. The branded apron. The kind of room it sits in on a busy Friday, with a function in full swing and a queue starting to build.
Inside that environment, the candidate is on shift. They are not clicking through a quiz. They are speaking out loud to Infinisona, the speech-to-speech model that powers VirtualShift, which plays the people a bartender deals with on a real night. A guest who cannot decide. A regular who wants to chat. A shift lead who gives a correction and watches what the candidate does with it. A rush that builds while someone is still mid-order.
That back-and-forth happens in real time. The candidate speaks and InfinSona answers, the way a real shift moves, rather than typing into a box.
The point of the recreation was honesty. A candidate who has a good trial at a bar that looks nothing like the job tells you very little. A candidate who handles a Friday function at a stall built to match The Fellas' own setup tells you a lot. When they walk into their first real booking, they have, in a real sense, already done it once. Day 1 ready.
Step 2: We tuned the scoring to measure the relevant soft skills
A realistic trial is only useful if you can read it the same way every time. That is what VibEye, our scoring model, does. Hiring managers choose the values that matter for the role, and those values are what the trial tests and scores.
The Fellas chose two main ones: coachability and communication.
Neither is a single number. Each one is a set of observable behaviours, and the trial is built to give the candidate a chance to show them.
Coachability shows up in what a candidate does after a correction. The shift lead in the trial gives one, on purpose, partway through. Vibeye reads what happens next. Did they adjust, or repeat the mistake? Did they ask a clarifying question when the setup was unfamiliar, or guess and hope? Did they recover after getting something wrong, or did it rattle them for the rest of the shift?
Communication is read across the whole trial, from the speech itself. Because the candidate is talking rather than typing, Vibeye works from how they actually speak. Clarity under a bit of pressure. Warmth with a guest. Reading the difference between someone who wants a recommendation and someone who wants to be left alone. Handling a complaint without making it worse.
Those two were the priority, not the whole test. Every candidate was still scored across the full set: customer focus, collaboration, problem solving, composure, and adaptability. The hiring lead saw the complete picture, with coachability and communication weighted to the front.
A confident talker who cannot take a correction does not score well here. Neither does someone who nods along and then does the same thing again. That is the point. The trial is built so the easy-to-fake signals do not carry the score.
Vibeye reads those behaviours the same way for every candidate, built to be consistent and fair rather than a black box. The score is a lens on the candidate, not a gate that locks them out.
Step 3: We built it into every hiring round
The Fellas do not hire once. They hire on repeat. So VirtualShift had to fit a process that runs again and again, not a one-off pilot.
It sits at the top of the funnel now. When applications come in for a round, every applicant gets a link to the trial shift instead of a first-round phone screen. They complete the 20 minutes whenever it suits them, from their phone. There is no venue to book, no roster gap to cover, no calendar to coordinate. Trial shifts as a service.
The hiring lead opens the VirtualShift dashboard and sees every candidate rated against the two values The Fellas chose. Coachability and communication, side by side, for the whole round. Instead of guessing from a resume or a ten-minute call, they start from evidence of how each person behaved on a shift.
From there, a person decides. The Fellas use the scores to shortlist, then spend their limited interview time only on the candidates worth it, having a real conversation about availability and fit. The model does the screening. The people still do the hiring.
Because the setup is saved, the next round costs almost nothing to run. Same stall, same values, same scoring. When a seasonal peak hits and a hundred applications land in a week, the screen scales with them without adding hours to anyone's week.
The results
The change was less about a single headline number and more about what each round now costs and produces.
| What changed | Before VirtualShift | With VirtualShift |
|---|---|---|
| First-round screening | A phone call with as many applicants as time allowed | A 20-minute trial shift completed by every applicant |
| What the hiring lead sees | A resume and a gut feel | Coachability and communication, scored, for everyone |
| Time to a shortlist | Days of calls, squeezed around events | The same day the round closes |
| Cost to run the next round | The whole process again, from scratch | The saved setup, re-run |
| First-shift readiness | Found out on the night | Already demonstrated in the trial |
The hiring lead stopped spending evenings on phone screens. Candidates who would have been cut for a shaky resume or a nervous call got through on what they could actually do. And the people The Fellas booked were, more often, ready on the first night rather than learning the basics in front of paying guests. Know before you hire.
It was not instant. The first version of the stall was close but not right, and their team flagged details we had missed: the way a particular rush builds, the kind of guest who turns up at a corporate function. We adjusted. That feedback is part of why the trial reads true now.
In their words
> Now I know with confidence that the people I'm interviewing are from the top percentile of those applying for the job. This took tons of hours and dollars beforehand. > — Director, The Fellas
Why it works
The Fellas did not need a smarter resume filter. They needed to see who could take a correction and talk to a room, before the first booking, and to see it the same way every round. Soft skills are the only honest signal for this kind of work, and a trial shift built on their own bar is how you actually read them.