Methodology
1. Sampling Strategy
A stratified random sample of 9,408 companies was drawn from 793,980 active UK engineering firms registered at Companies House. Stratification variables include 12 UK regions (NUTS1) and 12 industry sections (SIC 2007 code groups), yielding 784 companies per region for ±3.5% margin of error at 95% confidence. The national margin of error is ±1.0%.
Population: 793,980 active UK engineering companies
Sample size: 9,408 companies (784 per region × 12 regions)
Confidence: 95% nationally (±1.0%), per region (±3.5%)
Stratification: 12 regions × 12 industry sections
2. Data Sources
The index draws on five independent data channels. Rather than scoring each channel separately, all channels are merged into a unified evidence set per company. This ensures that the same I4.0 term detected across multiple sources is counted only once, preventing data-volume bias.
Website Analysis
Automated crawling of company websites, covering self-reported capabilities, product and service descriptions, and technology references. Detected I4.0 terms are matched against a 250-term dictionary.
Social Media Posts
Analysis of company social media activity for I4.0 technology mentions, capturing temporal signals of ongoing engagement with advanced manufacturing technologies.
Web Search
Search of the open web for each company's I4.0 adoption evidence, covering news articles, press releases, case studies, and industry reports. Provides independent, citation-backed validation.
Patent Records
Patent office search for I4.0-related intellectual property filings, mapped to technology pillars. Patent families are deduplicated so each invention is counted only once regardless of jurisdiction.
Company Profiles
Analysis of company profile descriptions and taglines for direct technology positioning signals and self-identification with I4.0 technologies.
3. Evidence Aggregation
All unique I4.0 terms detected across the five data channels are merged into a single evidence set per company. This approach measures adoption, not data availability — a term found on a website and in a LinkedIn post is counted once, not twice. The two primary dimensions extracted from this merged evidence are channel count (how many independent sources confirmed I4.0 activity) and pillar breadth (how many of the 10 technology pillars have at least one confirmed term).
Step 1: Merge
For each company, collect all detected I4.0 terms from all five channels. Duplicate terms across channels are deduplicated, creating a single set of unique terms per technology pillar.
Step 2: Channel Count
Count the number of independent data channels (0–5) that detected at least one I4.0 term. More channels providing evidence indicates stronger, cross-validated adoption signals.
Step 3: Pillar Breadth
Count the number of unique technology pillars (0–10) with at least one confirmed term across all channels (union count). Higher breadth indicates diversified I4.0 engagement across multiple technology areas.
Step 4: Tier Classification
The channel count and pillar breadth are combined to assign each company to one of six adoption tiers (see Section 5). This replaces a single composite score with a more transparent, multi-dimensional classification.
5
Data Channels
10
Technology Pillars
6
Adoption Tiers
4. Key Metrics
The index reports the following metrics at national, regional, and industry levels. The adoption tier is the primary classification used across all headline reporting.
Adoption Tier
Six-level classification (No Adoption → Leading) based on channel count × pillar breadth. The primary metric for all headline reporting.
Adoption Rate (%)
Percentage of companies with at least one confirmed I4.0 signal from any channel. Reported at national, regional, and industry level.
Channel Count (0–5)
Number of independent data sources that detected I4.0 activity for a company. More channels = stronger cross-validated evidence.
Pillar Breadth (0–10)
Number of technology pillars with at least one confirmed term (union across all channels). The key dimension for measuring technology diversification.
Pillar Adoption Rate (%)
Per-pillar percentage of companies with at least one confirmed term for that pillar. Reported at national and regional level across all 10 pillars.
5. Adoption Tier Classification
Companies are classified into six adoption tiers based on two dimensions: the number of independent data channels providing evidence and the number of unique technology pillars detected (breadth). This channel × pillar approach rewards cross-validated, multi-source evidence rather than relying on a single composite score.
No confirmed I4.0 adoption signals detected across any channel
Single verified I4.0 signal from one data source, or limited evidence across two technology areas
Early cross-validated signals or broad single-channel breadth
Multi-source evidence across several technology pillars
Strong, well-evidenced adoption with broad technology coverage
Deep, multi-channel adoption spanning most technology pillars
Channels = number of independent data sources (website, social media posts, web search, patents, company profiles) that detected at least one I4.0 signal.
Pillars = number of unique technology pillars (P01–P10) with at least one confirmed term across all channels (union count, not per-channel).
6. Data Quality & Sample Notes
The sample underwent a comprehensive quality audit in March 2026. All 9,408 companies have been verified to possess valid website and LinkedIn URLs, and have complete data across all five data channels.
Regional coverage: All 12 regions have exactly 784 companies, achieving the full stratified target. Regional margins of error are ±3.5% at 95% confidence.
Transportation sector: Only 17 companies (target: 60) due to the very small population of engineering-related transport companies (167 total). Transportation findings should be interpreted with caution (MoE ±24%).
Ambiguous term validation: 42 terms classified as potentially ambiguous were validated through contextual analysis, resulting in removal of 2,224 false positive term detections across 2,404 companies. This ensures that generic terms (e.g., "API", "CRM") are only counted when used in an I4.0 context.
Patent family deduplication: Patent records are deduplicated by invention family. A single invention filed in US, EP, GB, and CN is counted once, not four times. All jurisdictions are kept (global filing still demonstrates innovation activity).
7. Technology Intensity Classification
All 9,408 companies are classified into three technology intensity tiers using a web-verified hybrid Eurostat framework. The classification extends the NACE Rev. 2 manufacturing technology intensity categories to include Knowledge-Intensive Services, providing a unified framework across all sectors.
Pharmaceuticals (SIC 21), Electronics (26), Telecommunications (61), Software (62), R&D (72)
Chemicals (20), Machinery (27–28), Vehicles (29–30), Mining (05–09), Engineering (71)
Food (10–12), Textiles (13–15), Furniture (31–32), Wholesale (45–47), Hospitality (55–56)
Multi-SIC resolution: Of 9,408 companies, 2,717 (29%) have multiple SIC codes. Among these, 560 had codes spanning different technology tiers. Three classification scenarios were evaluated:
Primary SIC only — Conservative approach using the first SIC code (misses legitimate high-tech)
Highest tier — Aggressive approach using the highest-tier SIC code (false promotions)
Web-verified (adopted) — Each of the 560 disputed companies individually researched to determine actual business activity
The web-verified approach was adopted as the most academically defensible. Every reclassification is backed by evidence about what the company actually does.
8. 10 Technology Pillars (250 Terms)
The I4.0 term dictionary contains 250 terms organized into 10 technology pillars (25 terms each). Detection uses case-insensitive word-boundary regex matching across all data channels. Terms detected across multiple channels are deduplicated per company.