Cutting out the noise: How FDM Group redesigned their Early Careers hiring to beat AI

Challenge

In 2023, Fleur Laffont (Head of Graduate Talent, UK and EMEA) and Yana Stateva (Recruitment and Assessment Manager, EMEA) were struggling with a familiar trend in talent acquisition: their recruiters were spending a lot of time carefully reviewing applications, doing telephone interviews, and assessing candidates using video interviews and numerical and verbal reasoning tests. But 40% of candidates still failed the Assessment Centre stage later in the process.

As Yana explained: “The low pass rate told us that the process could be a lot tighter and more efficient, especially identifying which candidates had the potential to retrain and succeed in our IT and business divisions.”

But there was another issue: a huge increase in applications. Like many companies, FDM was suddenly saw hundreds of new candidates who all looked very similar on paper, making it even harder to predict who would be successful. 

After investigating, Fleur and Yana discovered the significant assessment fail rate, and the extra volume, were down to one cause: the rise of generative AI. 

Candidates were using auto-apply AI tools to apply for thousands of jobs while they slept, explaining the high volume of applications. But these AI tools were also editing candidates’ CVs and application forms. The result was high-scoring applications that all looked alike. 

Even traditional multiple choice reasoning tests (which FDM used in their recruitment process) were increasingly vulnerable to inflated scores from ChatGPT. 

The integrity of FDM’s sifting process was compromised. Candidates looked strong at the screening stage, where CVs, video interviews, and reasoning tests were being inflated by AI-generated answers, but their actual performance in the Assessment Centre didn’t align, leading to wasted recruiter time and a higher drop-off rate at the final stages. 

Fleur and Yana realised they needed a new way to identify real abilities earlier in the process.

Solution

Finding tech that works with AI-enabled candidates 

The answer was to look beyond how candidates say they behave and what they know (as with traditional reasoning tests), or what experience they have (as with CVs). 

Instead, FDM wanted to look at how candidates actually behaved to get a clearer picture of their true abilities. 

The team had already been using Arctic Shores’ Task-Based Assessment at later stages of their process. The assessment measures how candidates behave as they complete a series of tasks. This gives an accurate, unbiased, and AI-free view of their natural strengths and abilities. 

After discovering that CVs, video interviews, and reasoning tests were all vulnerable to AI manipulation, FDM made a strategic decision: move Arctic Shores to earlier in the hiring journey. This ensured that only high-potential candidates progressed to later stages, saving recruiters and hiring managers time, and improving hiring outcomes.

The pivot improved more than just accuracy: it built resilience in an AI-enabled world.

Large language models (LLMs) are capable of solving language-based problems, making traditional multiple-choice assessments, SJTs, and CVs easy to manipulate. But these models currently aren’t capable of solving task-based problems. The interactive nature of a task-based assessment is currently able to thwart even advanced AI models, ensuring that the results reflect human ability, not artificial intelligence.

This was critical for FDM. Research shows two in three candidates are now using AI in some way during job applications, leading to the massive spike in application volume that FDM experienced. Using a task-based assessment at the start of the process gave FDM’s recruiters greater ownership of their process. 

Beyond technology: Change management that delivers results

As with all new procedures, implementing new tech is just a part of the process. Success depends on people, processes, and the right tools working together.

Arctic Shores partnered with FDM for a seamless integration of the platform:

• Build global alignment: creating an assessment model that could scale across different locations, ensuring a consistent process for sifting talent worldwide. 
• Demonstrate business case to stakeholders: Fleur and Yana’s ideas for innovation, coupled with the expertise of Arctic Shores’ Business Psychologists, positioned the move as a future-proofing step for the business.
• Navigate compliance: With AI hiring regulations evolving at pace, Arctic Shores helped FDM maintain necessary legal procedures to make sure their process aligned with compliance best practices across different regions.
• Improve efficiency with simple, effective platform, and process integrations: Arctic Shores’ Task-Based Assessment worked alongside FDM’s existing hiring framework to help recruiters and hiring managers capture a holistic view of candidate potential. 
• Ongoing data-driven optimisation: regular reviews adapted FDM’s benchmarks and scoring models, making sure the assessment continued to identify the right talent.

The results: A fairer, faster, future-proof hiring process

12 months on from pivoting their traditional sifting tools to Arctic Shores Task-Based Assessment, Fleur and Yana’s innovative idea has driven results in Assessment Centre pass rates, efficiency, quality, diversity, and candidate experience measures too.

More efficient hiring

Recruiters saved 82 days of manual screening time, allowing them to focus on engaging with top candidates.

Higher-quality talent

More candidates progressed through the final hiring stages, improving the pass rate from 60% to 74%, and showing that the assessment was a more accurate predictor of success. 

A stronger, more diverse talent pool

Traditional screening often favours candidates from conventional backgrounds, whilst task-based assessment is more inclusive. By introducing the Arctic Shores assessment as an early-stage screen, FDM were able to efficiently reduce their pipeline, whilst ensuring that diversity was maintained.

Improved candidate experience

Instead of getting generic rejection emails, candidates received personalised feedback. As a result, 84% of candidates reported having a positive experience, and 83% said the assessment reflected positively on FDM’s brand.

Conclusion: A blueprint for future-proof hiring

FDM’s results demonstrate the impact of evolving the Early Careers hiring process. In an AI-enabled world, Fleur and Yana showed that hiring based on what candidates can do (not what they say they can do) is the key to future-proofing the hiring process and ensuring long-term recruitment success.

The next step? Having successfully tested this approach on UK Early Careers hiring, FDM is now working with Arctic Shores to scale this success across their entire global pipeline.