How benchmarkai works

Data sources and what they measure

Scoring dimensions and behavioural anchors

Each company is scored 0–10 on three dimensions. The anchors below define what distinguishes one band from the next, so a 6 and a 7 reflect materially different evidence rather than impression.

From signals to scores

Reading the AI Strategic Positioning Matrix

Each company is plotted on two fixed 0–10 axes — AI Integration Depth runs horizontally (left: Experimental → right: Embedded) and Quantified Business Impact runs vertically (bottom: Anecdotal → top: Measurable). Dot size encodes the third dimension, Workforce Transformation: small marks a score of 0–3, medium 4–6, and large 7–10. The quadrants are diagnostic rather than evaluative: a position describes a posture, not a verdict.

The grid below mirrors the matrix: the top row carries higher business impact, and the right column carries deeper integration.

Evidence weighting and confidence thresholds

Every company carries a confidence rating that signals how much weight its scores should bear.

Reading the Signal Strength bars

The Signal Strength bars at the top of every report show, per company, how much independent public evidence the analysis found — a measure of how well-covered a company is, not of its AI maturity. A longer, greener bar means the report's read on that company rests on more, and more credible, sources; a short red bar means the company is thinly covered, so its scores should be treated as tentative.

Each bar's length is the number of distinct sources drawn on for that company, scaled relative to the best-covered company in the same report — the company with the most sources fills the bar and the others are shown in proportion, with the raw source count printed alongside it. Its colour is that company's overall confidence rating (green High, amber Medium, red Low), assigned under the thresholds above, which weighs not just the number of sources but their recency, independence, and quality. Length and colour can therefore diverge — a company supported by several weak or dated sources can still show a long amber or red bar. Because the scale is relative to the strongest-covered peer, Signal Strength compares evidence coverage within a single report and is not an absolute count across reports.

Limitations and interpretive caveats

1. 1

   Benchmark granularity: All four institutional benchmarks publish sector-level data only; individual company scores are not available from them. Company positioning within a sector is inferred from live signal analysis, not directly measured.

2. 2

   Evidence asymmetry: Listed companies with active investor-relations programmes generate more signals than private firms or those with limited English-language digital presence, creating a systematic bias toward larger, more communicative organisations.

3. 3

   Temporal inconsistency: Live signals reflect the past 12 months while the benchmark sources span 2024 to 2026, so cross-source comparisons should account for this gap.

4. 4

   LLM non-determinism: Model outputs are probabilistic: two reports on identical inputs may differ by ±1–2 points. Reports should not be used to draw fine-grained distinctions between companies with similar scores.

5. 5

   Signal-volume constraint: Each report draws on at most 20 web signals per company — sufficient for orientation and hypothesis generation, but not a comprehensive audit.

6. 6

   Self-reported data: Many signals originate from press releases, earnings calls, and newsrooms, which carry promotional bias. The model weights third-party and analyst sources more heavily but cannot fully eliminate it.