• Event: Learning Technologies UK 25
• Date: 23 April 2025
• Speakers
  • Sandra Loughlin, Chief Learning Scientist, EPAM Systems
  • Anandi Shankar, Global Head of Learning and Leadership Development, Unilever
• Chair: Amanda Nolen, Co-Founder, NilesNolen
• Estimated read time: 10 minutes

 

---

 

Quick read summary

This session explored what it really means to work with skills across a large organisation, beyond frameworks, taxonomies, and vendor promises. Drawing on contrasting experiences at Unilever and EPAM Systems, the discussion examined why many skills initiatives stall, where complexity creeps in, and how skills can become meaningful to business decision making.

The topic matters now because organisations are under pressure to respond to rapid technological change, particularly AI, without clear visibility of workforce capability. Leaders are being asked to make decisions on hiring, redeployment, and development without reliable skills data.

Readers will gain a grounded view of what has gone wrong in real organisations, what has worked over time, and how to reset skills efforts around shared language, ownership, and practical business outcomes.

 

Why skills initiatives struggle at scale

Large organisations rarely fail at skills because of lack of ambition. They fail because complexity builds faster than understanding.

At Unilever, early momentum created confidence that skills were being “cracked”. Talent marketplaces, gig opportunities, and role deconstruction showed measurable engagement and progress. The problem emerged as layers of terminology accumulated faster than employees could absorb them.

Leadership skills, functional skills, domain skills, priority skills, focus skills, and future fit skills all had rational definitions. In practice, the pace of change meant there was no stable, shared understanding of what a skill actually was.

When business leaders were asked to define skills, many struggled to give a clear answer. The language existed, but it was not embedded or consistently understood.

The lesson was not that language does not matter, but that it must stabilise long enough to become shared.

 

When skills data fragments decision making

A second challenge emerged around technology and ownership.

Skills data sat in multiple systems, learning platforms, recruitment tools, and performance processes, each with its own taxonomy and owner. Together they created a fragmented signal that could not easily answer basic business questions.

Leaders could not get a single view of which skills mattered most in a function, whether those skills were being developed, or whether they already existed in the organisation.

This fragmentation slowed decision making and eroded confidence. Skills insights took days to assemble, while commercial data was available instantly.

The practical takeaway was simple. If skills data is owned by multiple functions and systems, it will not support real time business decisions, regardless of how advanced the technology appears.

 

Starting with work, not skills

EPAM Systems approached skills from a different starting point.

Rather than treating skills as an HR construct, the organisation began with the work itself. Tasks were defined first, then the skills required to complete those tasks, and finally the people who possessed them.

This task led approach allowed skills to function as operational data rather than abstract descriptors. Skills became a way to match people to projects, inform staffing decisions, and identify development needs in real time.

Because the organisation was designed this way from the outset, skills data flowed across hiring, learning, performance, and project systems without needing retrofitted integrations.

The insight for other organisations was not to copy the technology, but to recognise that skills only create value when anchored to real work.

 

Ownership, accountability, and trust

A recurring theme in the discussion was ownership.

At EPAM, the business owns skills definitions because the business understands the work. HR and learning functions align around those definitions to support hiring, development, and performance.

At Unilever, the emerging model places learning as the end to end owner of skills data, while the business defines what skills matter and why.

Both perspectives converged on a shared principle. Skills initiatives fail when ownership is unclear or fragmented. Someone must be accountable for coherence, governance, and usability.

Without that accountability, skills remain an internal HR conversation rather than a tool leaders trust.

 

Skills, AI, and the future of work

The session also connected skills strategy directly to AI.

AI changes work by automating tasks. As tasks shift, so do the skills humans need. This makes task intelligence and skills data inseparable from AI strategy.

Organisations will increasingly require a unified enterprise data layer to support AI. Skills data will form part of that foundation, not because of learning priorities, but because AI systems require integrated data to create value.

Separating skills conversations from AI discussions risks leaving organisations blind to how human capability and automation interact.

 

Practical application: making skills usable

Questions leaders should be asking

• Can people across the business clearly explain what a skill means here
• Is there one trusted source of skills data
• Are skills being used to inform real business decisions

Signals to watch in the organisation

• Leaders asking for skills insights to support hiring and redeployment
• Employees engaging with skills when they see career impact
• Frustration with duplicated data across systems

Common pitfalls

• Expanding skill taxonomies faster than shared understanding
• Treating skills as an HR or learning problem
• Investing in disconnected point solutions

What good looks like in practice

• Skills anchored to work and tasks
• Clear ownership and governance
• Skills used alongside performance, experience, and leadership

 

Key takeaways

• Skills only add value when tied to real work and decisions
• Shared language matters, but only if it stabilises
• Multiple systems without a single source of truth undermine trust
• Business ownership of skills content is essential
• Skills, AI, and task data must be considered together

 

Quote of the session

“Skills are good, but they are not the end all or be all.”

Anandi Shankar, Global Head of Learning and Leadership Development, Unilever

 

Final thoughts

Working with skills across an organisation is less about ambition and more about discipline. Simplicity, ownership, and relevance matter more than sophistication.

As technology reshapes work, organisations that ground skills in tasks and business outcomes will be better placed to adapt. Those that continue to treat skills as abstract frameworks risk creating complexity without clarity.

 

---

 

Speakers

Anandi Shankar, Global Head of Learning and Leadership Development, Unilever. A senior leader with extensive business partnering experience, focused on making learning practical, relevant, and commercially grounded.

Sandra Loughlin, Chief Learning Scientist, EPAM Systems. A learning scientist overseeing large scale education ecosystems and advising on skills based organisational design.

 

---