When application volumes passed what her team could handle, Maxine Roberts trialled Hubert alongside her recruiters’ own decisions before relying on it. Applicants now rate the interview 9 out of 10, more than half complete it on a mobile and 40% do so outside office hours.
Applications outgrew the resourcing team
PGL Travel is a brand many people in the UK recall with fond memories. Its outdoor centres are where many of them first visited as children. The resourcing team is led by Maxine Roberts, Head of Resourcing and EDI. At the start of the year, applications reached a level her team could not cope with.
PGL does not recruit on experience or qualifications, so there are no clear knock-out questions to screen applications out. A large volume of applications was left with no easy way to sift it.
A trial to address the problem
Maxine approached Hubert with the problem already defined. “I think it’s really important that you start with the problem rather than getting sucked into the really cool tech,” she says.
Plugging technology straight in would have been the easy option. Instead, PGL ran a trial in which Hubert’s decisions sat beside the ones its own recruiters would have made. The team used A/B testing, and Maxine had her recruiters assess the decisions Hubert was making. In her words, “it was really important for us to trial it and to sit it alongside the decisions that our recruiters would have made if they were doing the sifting themselves.”
Why PGL trusted the approach
Maxine trusted Hubert because of its consistency and fairness. The transparency of how it sifts mattered too. She also found it reassuring that the system is locked down and does not hand decisions to an open large language model, which gave her confidence that results would stay consistent. The deterministic scoring of Hubert’s model was a key differentiator for trusted results.
Scoring is anchored to PGL’s existing job descriptions and competency framework. “You’ve got to trust in your job descriptions and you’ve got to trust in your competency framework,” she says, “and that has to be the basis of continuing to sift out applications.” Every applicant is measured against the same criteria and the same decision is reached each time, which is what makes Hubert’s approach to AI in hiring legally defensible by design.
Applicants keep the choice
PGL attracts a high number of neurodivergent applicants, so accessibility shaped the rollout from the start. Some neurodivergent applicants prefer the Hubert process, and anyone can opt out of the automated process and go through PGL’s standard human process instead.
“I think it’s really important that the applicant still drives the ultimate process and we don’t forget about candidate experience through it all,” says Maxine.
Applicants rate the interview 9 out of 10 and often complete it on a phone out of hours
Hubert’s data for PGL Travel shows how applicants take the interview. Of completed interviews, 56% were finished on a mobile and 40% outside office hours, so applicants completed it when and where it suited them. It took 13 minutes on average, and applicants rated their experience 9 out of 10.
Recruiters back in the interview room
The change Maxine values most is where her team’s time now goes, spending it having the right conversations.
“What it’s allowed my team to do is actually focus on interviewing applicants rather than sifting out CVs. They’re spending a lot more time now actually having the right conversations with the right applicants and still making the hiring decision, which for me is hugely important.”
