The economics of human resources technology: Evidence from the UK
A new report from Strategic Gears argues that people management is structurally shifting away from isolated, point-solution HR systems toward integrated platforms that span the full employee lifecycle. However, as UK organisations make that move, they face three key challenges ahead.
Volatility has become the norm when it comes to global affairs – and businesses have spent the last six years facing one battle after another, when it comes to keeping up. Artificial intelligence has become a central part of those attempts – thanks to the agility it allows firms to leverage, without the need to recruit new skills on a large scale.
As companies around the world look to adopt AI to that end, the UK is one of the economies leading the charge. In the UK, for example, a 42% chunk of leaders recently explained they had conducted change readiness assessments to help get the most from the technology – higher than any other country polled. That includes the US on 37%, and the ANZ region on 31%.
New research from Strategic Gears shows that this has supercharged efforts among firms to adopt AI into many aspects of business at an accelerated rate – including human resources. Currently, 30% of UK employers use AI in hiring processes, up from only about 10% just three to four years ago. Recruitment agencies lead this transformation, with 48% adoption, signalling that enterprise usage will follow as agency norms and customer expectations evolve.
However, the market still faces a trio of challenges, which human resources leaders will need to reckon with, if they are to maximise returns on their investments in AI HR integration.
Obviously human recruiters are not immune to bias in the process of filling talent gaps. However, they remain clearly accountable for their actions, and – somewhere down the line – someone knows why a decision was made for or against a candidate. Even if that is only the recruiter themselves. But as in many walks of life, AI could be a “black box” – generating ‘best-fit’ approximations based on reems of data it has been fed, and potentially reproducing the bias in that data, without being able to explain why – while professionals using it may not be alerted to it.
According to Strategic Gears, narrow screening criteria, keyword matching, prestige markers, and exact job titles could therefore screen out adjacent-skills talent, for example, while generating “lookalike” hires. This cloning effect produces homogeneous teams with limited cognitive diversity, resulting in weak performance and potential discrimination claims under the Equality Act 2010 where non-job-related criteria disproportionately disadvantage protected groups.
Looking for ways around this, the experts note, “In parallel, many UK hiring stacks operate “selection-only” models, where candidates outside the shortlist hear nothing and “not progressed” is not consistently recorded as a decision, reducing the auditability and increasing candidate drop-off. Where shortlisting is automated or heavily model-driven, this can create UK GDPR and employment-law exposure, including Article 22 requirements for transparency, meaningful human review, and a right to contest decisions that materially affect individuals.”
Data sprawl
A famous idiom in data analytics is “garbage in, garbage out”. Even with the supposedly revolutionary capabilities of AI, this rule of thumb still applies: however powerful your analytics engine is, how expensive the GenAI you are expecting to help propel your company’s productivity to new heights, if you feed it bad data, it’s only going to be able to generate substandard results.
In HR, there is real risk of this, as hiring spans multiple disconnected portals, including CVs, assessments and interviews. These factors can drive candidate drop-off, and integration failures – but most troublingly, data inconsistency. Organisations managing 10-15 separate tools struggle with vendor fatigue, compliance gaps, and audit-trail visibility.
To that end, Strategic Gears recommends, “Unified platforms provide a single candidate portal or interoperable stack (SSO/SCIM and webhooks), a named system of record with mapped ownership, and clear SLAs for sync, resulting in lower friction, faster implementations, and cleaner data.”
Managing to do this can yield major benefits. In particular, the researchers argue that while single-point solutions can offer nine-times the productivity of regular HR processes, an integrated system sees that rise to 17 times.
Potential breaches
A practical truth for that last point also centres on fixing the data core – including remedying technical debt, and imposing stronger data governance policies, could unlock more productivity than just rolling out enterprise AI in the short term. In particular, clean data, clear processes, intentional decisions, and organisational readiness were all essential to making the most of these changes. This is also necessary on level of risk and compliance, however – which if not taken seriously, can see firms face an existential risk.
According to the experts, replicating candidate data across vendors increases breach risk, complicates Subject Access Requests and erasure, weakens audit trails and heightens risk of incongruency, while dispersed data creates a compliance and operational burden.
They conclude, “Best practice maintains a single golden record with live data protection Impact assessments, automated retention and deletion, unified rights portal across all processors, encryption on exports, and role-based access controls.”
