The numbers should bother us more than they do. Corporations globally spend north of $100 billion annually on employee training. Governments pour additional billions into national workforce development initiatives. And yet research consistently shows that only around 10% of that corporate spend delivers results that hold up in the real world. Employees forget roughly 70% of what they learn within 24 hours of a training session. Nearly half of HR managers report that they simply cannot find content adequately tailored to their organisation’s specific needs.
For most people in the industry, these statistics are a call to spend smarter or adopt better platforms. For Yuvraj Singh, they confirm something more uncomfortable: the model itself is wrong.
Singh has spent years not merely studying this problem but working at the centre of it — designing and leading learning systems deployed across sovereign wealth funds, portfolio companies, government ministries, universities, and large enterprises spanning Saudi Arabia, India, and the UAE. The scope of that work — tens of thousands of employees, hundreds of public service professionals, eighteen-plus government bodies — gave him a vantage point that is difficult to replicate from the outside. What he built and ran inside those institutions surfaced failure patterns that standard industry analysis consistently misses.
“Everyone in this space is optimising the wrong thing,” he says. “They’re making content more engaging, deploying better platforms, tracking more metrics. But the underlying model — centralised creation, standardised delivery, passive consumption — that model is what’s broken.”
It is a diagnosis Singh did not arrive at theoretically. It came from the specific experience of designing governance frameworks for how institutions manage learning content, building knowledge extraction processes for organisations whose institutional knowledge existed nowhere in documented form, and trying to make content production workflows fast enough to serve organisations where the pace of regulatory and operational change meant materials were obsolete before they reached learners.
The Personalisation Illusion
One of the most persistent misconceptions in enterprise learning, Singh argues, is that personalisation is a feature that can be layered onto an existing system rather than a complete rethinking of how content is created.
Leading learning programmes across government bodies and corporations in multiple countries sharpened this for him quickly. Working directly with ministers, chief learning officers, and institutional heads, Singh found that each organisation not only had different content requirements but fundamentally different assumptions about what rigorous learning looks like — different pedagogies, different standards of credibility, different institutional cultures that shaped what content employees would engage with versus what they would dismiss.
“Institutions are particular in their approaches,” he says. “And they should be. A financial regulator has a fundamentally different relationship to knowledge than a product team at an enterprise or a faculty department at a university. When you try to build one framework and stretch it across all of that, you produce content that technically covers the topic and actually serves almost no one.”
Working inside these institutions rather than advising them from the outside gave Singh a specific insight that rarely surfaces in vendor conversations: the failure of personalisation is not a technology problem. It is a design problem — one that begins at the point of commissioning content and cannot be corrected at the point of delivery.
In response to this, Singh developed approaches to learning content design that started from institutional specificity rather than standardised templates — mapping the actual knowledge requirements of distinct roles and functions before any content was commissioned, rather than adapting generic materials after the fact. It was an approach that required sustained access to institutional leadership and the willingness of organisations to treat content design as a strategic function rather than a procurement exercise.
Knowledge Locked in People’s Heads
The second structural problem Singh encountered across his work is one he describes as almost universally underestimated until organisations are deep inside a content production project: the degree to which the knowledge that needs to become training content exists nowhere in documented form.
In the large institutions he worked with — government ministries, sovereign wealth fund portfolio companies, university departments — the understanding of what employees needed to know to perform well lived not in any system but in the experience of senior practitioners. Officials who had spent years navigating complex regulatory environments. Investment professionals who had developed analytical frameworks through practice that had never been written down. Faculty members whose approach to a discipline had evolved over decades but had never been formalised.
“When you go to build a learning programme,” Singh says, “what you are very often actually doing is trying to extract institutional knowledge that the institution has never formalised. It retires when people retire. It transfers informally, if it transfers at all. And no content library or LMS was designed to capture it.”
Confronting this problem at scale — across organisations where the knowledge base was vast, unevenly distributed, and held by people with limited time and no incentive structure to transfer it — pushed Singh to develop what he describes as structured knowledge extraction processes: methodologies for drawing out tacit expertise from subject matter experts in ways that made it usable for content development without requiring those experts to become instructional designers.
It is the kind of operational insight that only comes from having run the process repeatedly, encountered its failure points, and iterated toward something that actually works. Singh has described the development of these approaches as among the most practically significant work of his career in the field.
The Vendor Problem
Perhaps the most structurally damaging element of how most large-scale learning programmes operate, in Singh’s assessment, is the way organisations have come to depend on external vendors — and what that dependence costs them in ways that rarely make it into post-project reviews.
Large institutions routinely outsource the creation and production of their learning content to specialist vendors. On paper this makes sense: vendors offer production capacity, content libraries, and instructional design expertise that most organisations cannot justify building internally. In practice, Singh observed — and in several cases was brought in specifically to diagnose and address — a consistent set of failures in how those vendor relationships actually functioned.
“Vendors do not transfer intelligence back to the teams they serve,” he says. “They build the content, complete the engagement, and the institutional learning from that process — what worked, what learners actually needed, what subject matter experts really knew — almost none of that gets documented and handed back. The organisation ends up as dependent as when they started.”
The production process itself is manual to a degree that surprises those who have not seen it up close. Notes taken imperfectly from subject matter expert interviews. Content drafted through multiple rounds of review. Production, approval, deployment — each step handled by a different set of hands, each handoff introducing delay and quality degradation. When the underlying knowledge changes — a regulation updated, a new process introduced — the entire cycle begins again.
In his work designing learning systems for institutions in the Gulf, Singh identified this vendor dependency as a root cause of why organisations with significant budgets and genuine commitment to workforce capability consistently failed to produce training that remained current. He developed governance frameworks intended to bring more of the intelligence generated through content development back inside the institution — structures that gave internal teams greater ownership of the knowledge being codified, rather than leaving it in the hands of vendors with no long-term stake in its quality.
The Learner Who Was Never in the Room
There is a final dimension to Singh’s diagnosis that is both the most obvious and, given its persistence, the most telling.
In the overwhelming majority of learning programmes he encountered across governments, corporations, and universities, the people who would ultimately sit through the training played no meaningful role in determining what it covered or how it was structured. Content got built by teams operating one, two, sometimes three steps removed from the people it was designed to serve.
“The learner is almost never consulted,” Singh says. “What they need to know, what format would help them, what they already understand — none of that is reliably part of the design process. And when you ask why, the answer is always the same: it takes time, it requires access, it complicates a process that is already difficult to manage at scale. So it gets skipped. Repeatedly, across organisations that know better.”
Working on programmes where the consequences of this gap were visible — where employees trained on content that did not reflect their actual roles continued making the same errors, where public service professionals completed programmes that had no measurable effect on how they worked — reinforced for Singh that the learner consultation problem was not a peripheral issue but a central one.
His response was to build learner feedback mechanisms into programme design from the outset — not as post-hoc evaluation tools but as inputs to content commissioning. It required persuading institutional clients to slow down the beginning of a content development process in order to speed up its usefulness at the end. It was, as he describes it, one of the harder organisational changes to advocate for — and one of the more consistently impactful when it was adopted.
Singh’s years inside these systems produced a diagnosis that is specific enough to be actionable and grounded enough to be credible. It is a diagnosis he has since turned into the foundation of something he is building — on the basis that understanding a problem this thoroughly and this directly creates an obligation to do more than describe it.
Read More From Techbullion
