The Half-Life of Skills Is Shrinking. Does Your Organization Have an Answer?

In 2020 a three member team took up the challenge to design a mathematical model for skillset fungibility; consequently, I published a white paper arguing that Skillset Fungibility would become a critical capability for organisations navigating accelerating change. Five years later, with AI reshaping entire skill categories almost overnight, that argument has moved from theoretical to urgent.

There is a stat that stops most senior leaders cold when they first hear it: the half-life of a technical skill — the time before half of what a professional knows becomes obsolete — has shrunk to under five years in many IT domains. In some AI-adjacent areas, it is closer to two.

Yet most organisations still manage their talent as if specific technical knowledge is their most valuable asset. They hire for today's skill stack, train for today's tools, and measure bench strength by today's demand. The assumption baked into all of this is that the skills needed tomorrow will closely resemble the skills needed today.

That assumption is no longer safe.

When I developed the Skillset Fungibility framework at BGSW in 2020, the trigger was a pattern I kept seeing in resource and capacity planning: teams with the wrong skills sitting idle alongside projects that could not find resources — not because the talent was absent, but because no one had a systematic way to understand how transferable one skillset was to another. The cost was invisible but real: bench underutilization, unnecessary external hiring, slower project ramp-up, and missed opportunities to develop people internally.

The Skillset Fungibility Index (SFI) that came out of that work was an attempt to make that invisible cost visible — and give organizations a data-driven method for managing it. What I did not fully anticipate in 2020 was how dramatically AI would accelerate the need for exactly this kind of thinking.

Why AI Has Changed the Stakes

What AI is doing — right now, not in some hypothetical future — is accelerating the obsolescence cycle of specific technical skills while simultaneously increasing the value of transferable capabilities. A developer who spent years mastering a particular testing framework now finds that AI tools can perform much of that work faster. The skill has not disappeared, but its scarcity value has collapsed. Meanwhile, the ability to understand systems, structure problems, and work across domains has become more valuable, not less.

This is precisely the dynamic that Skillset Fungibility is designed to address. Not the elimination of skills, but the shifting landscape of which skills are Sunrise, which are Mature, and which are moving toward Sunset — and what the viable transition paths look like between them.

The organizations that will navigate this transition well are not those that retrain fastest. They are those that understand the fungibility landscape of their talent pool deeply enough to make smart, data-led decisions about where to invest.

In the original framework, I categorized skillsets by market status — Sunrise, Mature, or Sunset — as one input to the Learnability Index. In 2020, that classification changed relatively slowly. Today, with AI as an accelerant across entire technology domains, the sunrise-to-sunset cycle is compressing. What this means in practice is that the fungibility of a person's current skillset — their proximity to adjacent, in-demand capabilities — has become a more important talent metric than the specific skills they currently hold.

What the Skillset Fungibility Index Measures

framework operates on a powerful insight: fungibility between two skillsets can be estimated systematically by combining two independent assessments — how adjacent the skillsets are in business context and domain, and how learnable one skillset is from the position of the other.

Adjacency Index — the outside view

This measures the structural proximity of two skillsets based on where they sit within the organisation's business clusters, domains, and skill families. Two skillsets within the same cluster and family score high on adjacency — the business context, tooling, and domain knowledge required are closely related. Two skillsets across different clusters and families score low. Each combination receives a score from 1 (lowest adjacency) to 8 (same skillset family), giving an objective measure of proximity that removes the subjectivity from what is typically a manager's intuitive judgment call.

Learnability Index — the inside view

This measures how readily one skillset can be acquired from the other, based on Competency Coverage (how much skill overlap exists), Learning Time (whether the transition requires upskilling, cross-skilling, or full reskilling), and Learning Environment (how readily training resources are available in the market and within the organisation).

The combined Fungibility Index

FI = Adjacency Index + 0.5 x Learnability Index

The weighting reflects a deliberate hierarchy: structural adjacency is a stronger predictor of successful transition than learnability alone, but learnability modifies the practical effort required. The FI score falls on a scale from 2.75 (minimum) to 13 (maximum) and maps to four decision zones: Right Fit, Near Fit, Upskill, and Reskill.

Use Case 1: Making Best-Fit Talent Decisions with Data

One of the most immediately valuable applications of the Fungibility Index is in resource allocation — answering the question: when a project needs a specific skillset and an exact match is unavailable, which existing resources are the best fit, and what investment is required to close the gap?

This sounds straightforward, but in practice most organizations answer it using a combination of manager intuition, informal knowledge, and availability — not a systematic assessment of skill proximity. The result is inconsistent decisions, missed internal candidates, and a default bias toward external hiring when internal development would have been faster and cheaper.

The table below illustrates how the Fungibility Index creates a transparent, data-driven basis for these decisions:

S.No.	Supply Skillset	Demand Skillset	Cluster	Family	FI	% Score/ Fit
A	IT Product Life Cycle Management	IT Application Life Cycle Management	Same	Same	10.5	76% — Right Fit
B	Healthcare	Embedded Automotive Application Software	Different	Different	5.5	27% — Upskill
C	IT Business Intelligence	IT Administration	Same	Different	5.25	24% — Reskill

The contrast between these three scenarios is instructive.

In Situation A, an IT Product Life Cycle Management professional transitioning to IT Application Life Cycle Management scores an FI of 10.5 — a 76% fit score firmly in the Right Fit zone. Same cluster, same family. The decision is clear: deploy internally, with focused upskilling.

Situation B tells a very different story. A Healthcare professional moving to Embedded Automotive Application Software scores 5.5 — 27%, Upskill zone. Different cluster, different family. Not an impossible transition, but the domain knowledge gap is significant and the investment timeline needs to factor into the staffing decision.

Situation C — IT Business Intelligence to IT Administration — lands at 5.25 despite sharing the same cluster. The different family creates enough divergence to fall into the Reskill zone at 24%. Two resources a manager might intuitively consider similar because they are both 'IT' are actually quite distant from a fungibility perspective.

This is the value of measurement: it replaces a manager's gut feel — which may be broadly right but is rarely precise — with a structured, transparent score that can be explained, challenged, and improved over time.

In our implementation at BGSW, applying this framework to the deployable bench allowed us to cross-train 5% of the resource pool in a targeted, prioritised way — focusing development investment on transitions with high FI scores where effort was manageable and business need was clear. The result was measurable improvement in deployment efficiency and a significant reduction in lead time between a project need arising and the right resource being available.

Use Case 2: The Make vs. Buy Decision — Train Internally or Hire Externally?

The second use case is one that every operations and delivery leader faces regularly: when a new skill need emerges, should the organization develop it internally or acquire it through hiring? This decision has significant cost and time implications, and it is almost universally made without the data needed to make it well.

The Fungibility Index provides a principled basis for this decision. If the FI between the organization's existing talent and the required skillset is high, the cost and time of internal development is likely to be lower than external hiring. If the FI is low, the development investment is larger, the time to productivity is longer, and external sourcing may be the more efficient path.

In practice, the decision maps to three zones:

High FI (above 10, Right Fit zone): Strong case for internal development. The skill gap is small, training is likely upskilling, and time to productivity is short. Internal development also preserves institutional knowledge and supports career growth.

Mid-range FI (7 to 10, Near Fit zone): Hybrid approach. Internal development is viable but requires structured investment. Consider combining internal reskilling with targeted external hiring for senior or specialist roles.

Low FI (below 7, Upskill/Reskill zone): External sourcing is likely faster and more cost-effective for immediate needs. Simultaneously, invest in building internal adjacency in the Near Fit range — developing skills today that move more capabilities into the high-FI zone over time.

What makes this particularly relevant in the AI era is the speed at which new skill requirements are emerging. When AI-related capabilities first appeared on project demand lists at Bosch, the question of whether to build or buy these skills internally was genuinely difficult. The FI framework provided a structured way to assess which existing skillsets were closest to new requirements — and therefore which resources could be developed most efficiently — versus which gaps required external acquisition.

In one capacity planning cycle, FI-based analysis redirected training investment away from low-adjacency transitions — where reskilling would have been slow and costly — toward high-adjacency transitions where the same investment produced faster, more reliable results.

How AI Is Reshaping the Fungibility Landscape

The Fungibility Index was built with an assumption that the skill landscape changes gradually enough that periodic recalibration is sufficient. AI has changed that. Three dynamics are worth calling out specifically:

1. Skill sunset is accelerating : Several skillsets classified as Mature in 2020 are moving toward Sunset faster than anticipated — particularly where AI tooling has automated significant portions of the work. The urgency of transitioning resources away from these skillsets has increased substantially.

2. New Sunrise skillsets have low adjacency to most existing talent : AI operations, prompt engineering, AI quality assurance, and human-AI workflow design are emerging as high-demand skillsets. The challenge: most existing IT skillsets have moderate to low adjacency to these areas. The FI for internal transitions into AI-related skills tends to fall in the Upskill to Reskill range, meaning organisations that wait until demand is acute will face longer transition timelines than they expect.

3. Meta-skills are becoming the highest-fungibility assets : Systems thinking, problem structuring, learning agility, and cross-domain pattern recognition score high on fungibility across almost any transition. Professionals with strong meta-skills and moderate technical depth are more fungible than highly specialised technicians with deep but narrow expertise — with significant implications for how organisations hire and develop talent.

The most fungible professionals in the AI era are not necessarily those with the most AI skills. They are those with the learning architecture to acquire new skills quickly, combined with enough domain depth to apply them meaningfully.

Five Things Operations Leaders Should Do Now

• Map your talent pool against the FI framework. Understand which skillsets you have in abundance, which are moving toward sunset, and which of your current resources have high adjacency to the skills you will need most in the next 18 to 24 months.

• Identify your high-fungibility resources. Professionals with strong FI scores across multiple potential target skillsets are your most valuable development assets. Ensure they are being developed, not just deployed.

• Build FI analysis into your Make vs. Buy decisions. Before defaulting to external hiring for an emerging skill need, run the FI numbers. You may find internal development options you had not considered — or get a clear, data-backed case for external sourcing.

• Accelerate development in AI-adjacent skillsets now. The AI-related skillsets with the highest adjacency to your current talent are the ones to prioritise first. Waiting until demand is acute means starting development after the window for internal transition has already narrowed.

• Track Learning Velocity alongside FI. The Fungibility Index tells you how transferable a skillset is in the abstract. Learning Velocity — the rate at which an individual can actually acquire new skills, based on training history and learning agility — tells you which specific people are most likely to make the transition successfully. Both dimensions matter.

The Window Is Narrowing

When I published the white paper on Skillset Fungibility in 2020, the argument was largely forward-looking. The VUCA environment was the context; the urgency was real but not yet acute for most organisations.

In 2025, the urgency is acute. AI is not a future disruption to prepare for — it is a present disruption to navigate. The organisations that have already invested in understanding the fungibility landscape of their talent pools are better positioned to adapt. Those still managing skills as fixed, role-specific assets are carrying a risk they may not yet have fully quantified.

The half-life of skills will not get longer. But the quality of decisions organizations make about skill development, talent deployment, and capability investment can get significantly better — if those decisions are grounded in data rather than intuition.

Skillset Fungibility is not an HR concept. It is a business resilience strategy.

And in the age of AI, it may be one of the most important levers available to operations and delivery leaders who are serious about building organizations that can adapt and grow.

Want to explore the full Skillset Fungibility framework? The white paper is available on request — reach out via shaileshgoel.in or connect on LinkedIn.