Skills-Based Pay: What the Research Supports
(and What It Doesn’t)

Skills-based pay is being promoted as the future of rewards: pay people for what they can do, not just the job they hold.

The direction is understandable. Work is changing faster than job architectures can keep up, organisations want more internal mobility, and leaders want pay decisions that support reskilling rather than rigid roles. But the evidence is clear on one point: skills-based pay only works when “skills” are defined, verified, and governed — otherwise it increases inconsistency and perceived unfairness.

This article summarizes what the research actually supports, where skills-based pay tends to fail, and a more defensible model that separates capability, competence and skills.

When rewards reinforce development (clear skill growth paths and recognized proficiency), organisations can drive broader capability-building — particularly in operational and technical environments. Academic evidence from a plant-level study found that a skill-based pay program was associated with improved plant performance outcomes (productivity, labour cost per part, and quality outcomes). Academy of Management Journals

In practitioner literature, skills-based approaches are commonly presented as a pathway to improved agility and internal redeployment — provided the organisation can define skills consistently and apply proficiency standards in a way that holds up across teams. Deloitte describes this as moving toward a skills-based operating model that helps organisations align people to changing work demands.

Mercer focuses on the “skills foundation” required for these practices, including taxonomy and market alignment; its Skills Library uses large-scale data collection that targets millions of online profiles and job descriptions, with updates throughout the year to add emerging skills and remove obsolete ones.

Common Design and Implementation Risks – Most failures are not about intent — they’re about measurement, evidence, and governance.

Many organisations are now using AI tools to help identify skill requirements and build skill profiles at scale. This includes extracting skills from job descriptions and learning content, scanning internal documents, and inferring skills from work history or project records (for example, using enterprise copilots and similar tools to summarize experience and suggest relevant skills).

Used well, AI can accelerate two activities:

• Skills discovery and taxonomy building  – AI can help surface emerging skills, standardize language, and identify likely skill requirements from job content.
• Skills inventory and profiling – AI can assist in creating an initial view of what skills people may have based on role history, training records, project experience, or self-descriptions.

However, AI-generated skill profiles should be treated as a starting hypothesis, not a verified basis for pay or progression.

AI can infer that a person is likely to have a skill.  It cannot reliably confirm proficiency, judgement, or performance in context without validation.

For skills-based pay, the risk is straightforward: if AI inference replaces evidence, it can amplify the same problems as self-reporting — inconsistency, inflation, and fairness concerns.

Practical rule: use AI to accelerate identification and mapping, then apply proficiency standards and verification where the skill affects pay, safety, quality, or risk.

A skill is typically a task-level ability. Pay systems, however, need to reward reliable contribution in context, not just capability in theory.

In the I/O psychology literature, skill-based pay is usually defined as a system that provides additional pay only after employees demonstrate the skills/knowledge/competencies being rewarded — i.e., it assumes formal certification or verification, not self-declaration. SIOP

This distinction matters because in real workplaces—especially regulated or high-risk environments—what differentiates competence is not “having a skill” but applying it correctly under real conditions.

Skills should be rewarded when:

• they materially affect risk, safety, quality, or productivity
• the skill can be verified (observation, test, evidence record)
• the organisation can define proficiency levels clearly

This aligns with how skills-based pay is commonly framed in research: additional pay is earned after demonstration/certification of the rewarded capabilities.
SIOP

OECD research on skills mismatch links misallocation of skills to productivity outcomes and indicates that improving skill allocation toward best practice can be associated with measurable productivity gains (the report provides a quantified example for New Zealand).

The practical implication for pay: if you can’t trust the skills data, you can’t trust the basis for pay decisions

If you’re reviewing skills-based pay or skills frameworks for 2026, these resources may help:

Capability, Competency & Skills — What’s the Difference?
A practical explanation of how capability, competence and skills fit together, and why the distinction matters for pay, performance and mobility.

Skills Based Hiring

Competence Debt: The Hidden Risk Building Inside Organisations
Why frameworks become outdated or unverified over time — and what to fix first.

Competence Debt Diagnostic (10 minutes)
Identify whether your organisation has usable competence data — and where gaps are building across roles, assessment and evidence.

Skills & Capability Inflation
Why skills profiles can look stronger than reality, and how to reduce inflation through proficiency standards and verification.

Internal Mobility Is Broken — And How Capability Data Fixes It
How role clarity and evidence-based competence data improve internal movement, career ladders and succession decisions.

Performance Management: What Effective Systems Really Measure
Why performance should be anchored in role responsibilities and evidence — and where competencies and skills fit (and don’t fit) in appraisal.

Request a walkthrough
If you’d like to pressure-test a skills-based pay approach against your current role and competency structure, we’re happy to review what will be defensible and sustainable in your context.