Data Sources & Methodology

1. Role Data

SkillShift covers 500+ roles across 12 sectors: Technology, Healthcare, Finance, Manufacturing, Construction, Professional Services, Creative Industries, Education, Energy, Retail, Logistics, and Government.

Roles are curated from established occupational frameworks including O*NET (US Department of Labor), ESCO (European Commission), and industry-specific job classification systems. Each role record includes:

• Title, sector, and job family
• Core skills and competencies
• Salary ranges (min, median, max) for three base regions
• Demand levels by region
• Growth projections
• AI impact classification
• Education and experience requirements
• Remote work viability (full, hybrid, or onsite)

Data is reviewed and updated periodically to reflect market changes, emerging roles, and shifts in skill requirements.

2. Salary Data

Base salary ranges (minimum, median, maximum) are benchmarked against publicly available compensation data from:

• Bureau of Labor Statistics (BLS) Occupational Employment and Wage Statistics
• Eurostat Structure of Earnings Survey
• Industry salary surveys and published compensation reports

Three base regions are used: United States, Europe, and Asia-Pacific. Each region provides its own min, median, and max salary figures per role.

Location-specific salaries are calculated by applying regional multipliers to base salary data. These multipliers reflect cost-of-living differences and local market conditions, and are derived from cost-of-living indices (e.g., Numbeo, ERI Economic Research Institute) and regional compensation surveys. SkillShift currently supports sub-region adjustments for 8 US states and 10 European countries.

All salary figures are estimates and may differ from actual offers. For the most current compensation data, we recommend consulting primary sources directly.

3. Demand Levels

Each role is assigned a demand level per region: Very High, High, Moderate, Low, or Declining.

Demand classifications are based on published growth projections and workforce supply/demand analysis from sources including:

• BLS Occupational Outlook Handbook
• World Economic Forum Future of Jobs reports
• LinkedIn Economic Graph insights

Regional demand adjustments reflect local industry concentration and sector-specific conditions. At the sub-region level, sector demand adjustments are applied to account for differences in industry mix between, for example, California and Texas, or Germany and Spain.

4. AI Impact Classification

Every role is classified into one of three categories based on how artificial intelligence is expected to affect the work:

• AI Resistant — Core tasks require human judgment, creativity, physical presence, or complex interpersonal skills that current AI cannot replicate.
• AI Augmented — AI tools enhance productivity, but the role itself persists. Professionals who adopt AI tools gain a competitive advantage.
• At Risk — Significant portions of core tasks can be automated with current or near-term AI technology.

Classifications are informed by task-level analysis against AI capability research from McKinsey Global Institute, World Economic Forum, Stanford HAI (Human-Centered Artificial Intelligence), and the Brookings Institution.

5. Skill Gap Analysis

When you run an analysis, SkillShift compares your stated background against the requirements of a target role and produces a structured report covering current skills, role requirements, gaps, transferable skills, a phased learning path, and recommendations.

The flow:

• Inputs. Your free-text background (resume, paragraph, or pasted CV) plus the target role. If you paste a job description, it takes priority over the generic role template.
• Role context. The target role is matched against our internal role library (500+ roles), which contributes typical skills, salary, demand, and AI exposure.
• Generation. Inputs are sent to Anthropic's Claude with a structured prompt that requires a strict JSON schema. Every field must reference your actual background; generic role-theory statements are explicitly disallowed.
• Auditability. Each skill gap includes an evidence trace showing what was (or wasn't) found in your input and what the role expects. This makes every flagged gap inspectable instead of asserted.
• Recommendations. Courses, certifications, books, and projects are generated alongside the gaps. Where a gap matches our internal course catalogue, curated links are surfaced; otherwise Claude proposes options.

Outputs are estimates produced by a language model from limited input. They are designed to surface a useful starting point for self-direction, not to be used as professional career counselling. We recommend cross-checking against current job postings, hiring-manager conversations, and primary sources.

6. AI Readiness Score

The AI Readiness Assessment produces a score from 0 to 100 based on a five-dimension model:

• Usage base (0–25 points) — Current AI tool usage frequency and breadth. How often and how widely you use AI tools in your work today.
• AI awareness (0–20 points) — Familiarity with AI concepts relevant to your sector, from large language models and image generation to automation and predictive analytics.
• Task-level adoption (0–25 points) — The gap between tasks you perform regularly and how much you use AI for those tasks. Higher adoption across more work activities yields a higher score.
• Mindset (0–20 points) — Attitudes toward learning, experimentation, and AI adoption, measured through self-reported comfort with new tools, willingness to experiment, and openness to AI-driven change.
• Context bonus (0–10 points) — Adjustments for team size, years of experience, and the breadth of work activities performed. These contextual factors influence how likely adoption is to scale.

The maximum possible score is 100. Sector-specific benchmarks are research-based estimates of typical adoption patterns by industry — not measurements of SkillShift users — and comparisons against them are labelled as estimates throughout the product.

7. Career Path Computation

Career paths between roles are computed using a graph-based algorithm (breadth-first search) that traverses connections between roles based on skill overlap.

• Skill overlap is calculated using Jaccard similarity between the skill sets of two roles — the number of shared skills divided by the total number of unique skills across both roles.
• Transition difficulty accounts for skill gap size, salary change, education requirements, and typical transition timelines.
• Stepping-stone roles are identified when a direct transition is not feasible. The algorithm finds intermediate roles that bridge the skill gap, creating multi-step paths from origin to target.

Each path includes shared skills at every step, new skills to learn, an overall similarity score, and an estimated timeline in months.

8. Funding Programs Database

The Free Training Finder includes 91 programs across 17 countries (US, EU-wide, plus 15 individual European countries).

Data is sourced from official government program documentation, legislative texts, and agency websites, including:

• US federal programs (Department of Labor, Department of Education, Veterans Affairs, Department of Energy)
• US state-level workforce development programs
• EU-wide initiatives (European Social Fund Plus, Erasmus+)
• Country-specific programs across Europe

Eligibility criteria and benefit amounts reflect published program guidelines. Programs are periodically reviewed; discontinued programs are marked inactive and excluded from screener results.