Scoring the leaders: from patents to a ranking.

Counting patents is the obvious approach, and it is wrong. A large industrial files thousands of patents a year across everything it touches; a focused innovator may file thirty, all of them in one technology. A raw count crowns the conglomerate every time and learns nothing. Our scoring is built to avoid exactly that trap.

Three families of signals

Each company is measured in each technology along three complementary dimensions:

Specialisation — how much of the company’s own patenting, by count and by estimated value, is concentrated in this technology. This is the pure-play signal: it finds the company for which this technology is the core business, not a side project.
Contribution — how much of all patenting in the technology, again by count and by value, comes from this company. This is the footprint signal: it recognises that scale in a technology is real information, whoever brings it.
Quality — the average strength of the company’s patents in the technology, using multi-factor patent quality metrics from Patsnap that reflect, among other things, citations and family breadth. Thirty strong patents beat three hundred weak ones.

Specialisation and contribution deliberately pull in opposite directions — one favours the focused innovator, the other the established heavyweight — and the composite balances them, with quality as the largest single voice.

Compared with peers, not with the world

Raw ratios mean nothing across technologies: a 5% share of a crowded field like battery storage is a different achievement from 5% of a young niche. So every metric is standardised within its technology group — ranked, trimmed of statistical outliers, and converted to a score relative to the direct peer set. A company is only ever compared with the other companies patenting in the same technology, in the same period.

Aggregating without a size bias

Technologies roll up into subclasses and theme classes, and here a subtle problem appears: how do you combine a company’s scores across several sub-technologies without letting patent volume dominate? Our answer is a weighted consolidation in which the weight of each sub-technology score grows with the square root of the patent count behind it — so ten times the patents earn roughly three times the weight, not ten. Confidence should grow with evidence, but not linearly; the five-hundredth patent tells you less than the fifth.

The practical effect is that concentrated leadership survives aggregation. A company that dominates geological CO₂ storage keeps that advantage when scores roll up to carbon capture as a whole, rather than being diluted by fields where it is absent — while a company that is merely present everywhere, and excellent nowhere, is identified as exactly that. Market-share measures, by contrast, are recalculated at each level from actual patent counts, because market share is the one signal that should reflect size.

From scores to the portfolio

The dimensions combine into a single composite score per company per technology, refreshed quarterly on a rolling window of recent patenting. The top tier within each technology forms the selection universe from which the portfolio is built — after the usual investability screens for liquidity, listing region and sector exclusions.

Impact as the guardrail, never the signal

One layer remains, and its role is deliberately limited. Patents tell us who is building the technology; they do not tell us whether the company’s overall footprint helps or harms the transition it profits from. For that we use independent impact and screening data from Sustainable Platform — measured contribution to the UN Sustainable Development Goals — as a guardrail on the selection, not as a driver of it.

The exclusions are explicit. Companies rated at high risk of greenwashing are out. That rating does what the word promises: it tests a company’s environmental claims against what the company actually does, combining its exposure to environmentally controversial activities — fossil fuels, deforestation, commercial fishing and the like — with its record of environmental fines and regulatory risk. The precise recipe is Sustainable Platform’s and proprietary; the principle is what matters here. The patent record cannot be greenwashed, and this check makes sure the rest of the company’s story cannot be either.

Beyond conduct, companies falling below minimum SDG impact thresholds are excluded, however well they score on patents. So are companies with no impact coverage at all — no data is treated as a red flag, not a free pass. And where two companies’ patent scores are close enough to be statistically indistinguishable, the one with the stronger measured impact ranks first: a tiebreaker that re-orders near-ties without adding or removing anyone.

Note what we do not use: conventional ESG ratings. Those blend governance box-ticking, policy disclosures and peer-relative opinions into a single letter — a measure of reporting diligence as much as of anything real. Our interest is narrower and harder: what a company actually contributes, measured against the SDGs. Innovation picks the portfolio; impact keeps it honest.

The scoring itself remains deliberately blind to everything else: no analyst conviction, no story, no meeting with a charismatic chief executive. Those things have their place; measuring innovation is not it.