Seeing the Work, Not Just the Workforce: Notes From Our Latest HR Leaders Roundtable

We recently brought together senior HR, reward, learning, and workforce planning leaders from across financial services, energy, technology, GovTech, and professional services for an in-person roundtable. Held under Chatham House rules, the conversation was refreshingly candid - and it kept returning to a single, uncomfortable question.

There's a particular honesty that surfaces when senior leaders are in a room together, off the record and away from the slideware. This conversation had plenty of it.

The people around the table had arrived from very different starting points - reward, learning, strategic workforce planning, talent acquisition. But within minutes it was clear they were all circling the same problem. As one facilitator put it early on, these disciplines are "all ultimately converging on the same thing" - trying to understand the work, the talent, and the value inside the business well enough to plan for what's coming.

And what's coming is arriving fast. One widely-cited figure in the room: 44% of workers' core skills are expected to be disrupted within the next five years. Another: McKinsey's estimate that 30% of US work hours could be automated by 2030. The numbers aren't the story. What to do about them - when your view of your own workforce stops at headcount and job titles - is.

Job titles tell you far less than you think

The clearest theme of the morning was a loss of faith in the job title as a unit of planning.

Leader after leader described the same gap: they know their people's names, they know their titles, they have a rough description on file - but they don't actually know the work those people do. And the moment you cross a border or a business line, the title means something different again. One leader described teams in Germany and the UK carrying identical titles but doing fundamentally different jobs, shaped by the nature of the business they support. Another, in energy, contrasted a high-margin upstream business that adopts every new technology first with a downstream retail operation where "we have to sell a lot of chocolate bars to make money" and the game is pure efficiency. Same company. Same org chart. Completely different work.

This is the fidelity problem, and it creates a genuine tug of war. Generalise every role into its parts and you lose the specificity you need to hire, develop, and reward well. Split every role by region and nuance and you create so much complexity the organisation seizes up. As one leader put it: "I don't think there's any right or wrong one. It's always going to be a tug of war."

The sharpest articulation came from a leader in pharma, describing a business where 25% of the workforce is due to retire within five years:

"I know the job titles of the people who are retiring. I know their names. But do I actually know what work is leaving the business in those five years? I don't."

That's the question underneath all the others.

Jobs, tasks, and skills - and the layer most people are missing

If the job title is too blunt and the individual skill is too granular, where do you plan? Much of the conversation landed on the task - the layer in between.

Break a role into the tasks it actually contains, several leaders argued, and two things become possible. You can finally ask a precise question about AI - not "can we automate this job?" (you almost never can) but "which of these tasks can we augment?" And you can compare, reward, and hire against the work itself rather than a generic description.

Nobody pretended this was easy. The debate over sequence - top-down job architecture first, or bottom-up from tasks and skills - went back and forth, and the honest conclusion was that it has to be iterative. "You can't boil the whole ocean at once," as one leader said. You run a round of analysis, top-down and bottom-up, see where the margin and productivity gains are, and go again. One organisation had functionalised the work, built cross-functional "labs" and monthly hackathons, and accepted openly that they might not get it all right first time. Starting beats waiting.

The self-reported skills trap

Everyone wants a live view of their workforce's skills. Almost no one trusts the data they have.

Leaders mapped a familiar maturity curve - from ad hoc and documented, to self-reported, to verified, to genuinely predictive - and agreed that most organisations are stuck at self-reported, which is exactly where the data is least reliable. The examples were pointed. One bank's plan to populate its skills profiles was to pay employees £5 per skill added - a scheme almost designed to produce a thousand skills nobody actually has. As one participant admitted, if you're paying per skill, "I'm going to sit there adding as many skills as I can think of." That ruins workforce planning for everyone else downstream.

Two fixes came up repeatedly. First, proficiency, not presence. A binary "yes I have this skill" flattens everything. "Can I do Excel? Yes. Can I model in Excel like my academic colleagues? Absolutely not." Without a competency level, and without a sensible constraint on the number of skills - 10 to 20 that you actually use, not 1,000 - the data tells you nothing.

Second, and more fundamentally: stop asking employees to do the work. The insight that landed hardest was that people will only invest in accurate skills data when they get something back. One leader described the experience employees actually expect - the Netflix model, where you signal what you value and get better recommendations in return. "Right now, regardless of system, there is none of that." Pre-populate the profile, ask the employee only to correct it, and show them what accuracy unlocks - a specific role, a project that closes their gap - and the incentive problem solves itself. "If someone shows me I'm missing three skills for the role I want, and there's a six-month project that gives me two of them - now I'm glad I filled in my profile."

Nothing moves without the business case

The connective tissue through every thread was outcomes - and specifically, the ability to translate a workforce problem into a language the CFO acts on.

The advice from the room was blunt: keep asking why until you reach the number. "I want internal mobility" is not a business case. Why? Because we're losing talent. Why does that matter? Because we can't deliver projects. And? "That's 10% of revenue over the next year." That is what gets funded. Too many organisations, several leaders agreed, still treat this as "an HR side project" rather than connecting it to finance.

There's a striking asymmetry underneath this. Talent is, for most organisations, the single biggest cost - and the CFO knows precisely what it costs. But does anyone know what it generates? That gap is where the opportunity sits. The most optimistic example of the morning made the point perfectly: a well-known retailer that automated its customer service function and, rather than simply cutting the roles, retrained those people as interior design consultants - creating a new billion-dollar revenue stream. The reframe from "how many jobs do we lose" to "what capability do we now have to redeploy" changes the entire conversation.

The human tension no one has solved

None of this happens in a vacuum, and the leaders were honest about the human cost. Several described a workforce carrying a kind of organisational trauma after successive rounds of change - people unwilling to "put their head above the parapet in case it gets chopped off." Others pointed to burnout as teams shrink and expectations rise, and to a shrinking entry point for early-career talent when the repeatable work that once trained them is the first to be automated.

The recurring parallel was COVID - the last time organisations proved they could change at pace, because they fixed on the outcome and let the middle be messy. The open question is whether organisations choose to keep that muscle, or quietly drift back to the way things were.

Where Beamery fits

These are precisely the problems Beamery was built to solve.

Our Work Intelligence capability makes the work visible - not just the workforce. Starting from a standard HRIS extract, we infer the tasks performed in each role, at what volume, and with what propensity for AI augmentation - turning "we think 25% of roles could be affected" into "38% of these task-hours are automatable, and here's the redeployable capacity that unlocks." That's the task-level fidelity the room kept reaching for.

Our skills graph connects internal employees and external candidates on one governed model, with inferred, evidence-based skills - so you're not relying on employees to self-report into a spreadsheet. And our agentic AI advisor, Ray, helps leaders act on that intelligence in real time, and build the business case that ties it to outcomes finance cares about.

If the questions from this roundtable are the ones your organisation is wrestling with, we'd be glad to show you what the answers could look like.

Beamery is the AI platform for workforce transformation. To learn more or speak with one of our team, get in touch.