Patent Analysis
- Introduced by
- Francis Narin (patent bibliometrics) and the patent-analytics community
- Year
- 1994
- Type
- Document-based technological-intelligence process
- Family
- Scientometrics and technology intelligence
What does it do?
Patent analysis, or patanalytics, mines the documents and metadata in patent databases to generate technological intelligence.
Because patents are structured, dated, classified, and citation-linked records of inventive activity, analysing patent counts, citations, classification codes, applicants, and text reveals who is innovating where, in which technologies, how fields connect, and how the technological landscape is shifting—evidence that feeds competitive intelligence, R&D strategy, and foresight.
The intuition behind the method
Every patent is a small, standardised dossier: it records what was invented, by whom, when, in which technical category, and which earlier patents it built upon. Taken one at a time these are legal documents, but aggregated by the millions they become a vast map of technological effort. Counting patents in a field shows where activity is concentrating; tracking citations shows which inventions proved influential; reading the classification codes shows how technologies overlap; and clustering the text reveals emerging sub-fields. Patent analysis turns this paper trail of invention into charts and maps that let strategists and policymakers see the terrain of innovation rather than guess at it.
History
Patent statistics have informed economic and technological analysis since the early twentieth century, but systematic patent bibliometrics emerged in the 1970s and 1980s with Francis Narin and CHI Research, who adapted citation-analysis techniques from scientometrics to patents and demonstrated links between patent citations and technological and economic value. Narin's 1994 'Patent bibliometrics' consolidated the approach. The subsequent availability of large machine-readable databases—EPO's PATSTAT, Derwent, and later open resources like Google Patents and Lens.org—together with text-mining and visualisation tools, expanded the field into the broad practice of patent analytics surveyed by Abbas, Zhang, and Khan in 2014.
When to use it
Use patent analysis when you need objective, large-scale evidence about inventive activity—who is patenting what, where, and how a technology field is structured and evolving—to support competitive intelligence, R&D strategy, technology foresight, or policy. It excels when the technology is patent-intensive and well represented in patent databases. The method assumes that patents are a meaningful proxy for inventive activity, that metadata and citations carry technological signal, and that large aggregates wash out individual noise. It is less appropriate for fields where firms prefer secrecy or where patenting is uncommon (e.g. some software or services), for very recent activity not yet published, or where strategic and defensive patenting behaviour decouples patents from real innovation.
Limitations and considerations
- Patents are an imperfect proxy for innovation: propensity to patent varies widely across industries, firms, and countries.
- An 18-month publication lag and the gap between invention and application mean the data always trail the technological frontier.
- Strategic, defensive, and 'patent thicket' behaviour can inflate counts that no longer track genuine technological advance.
- Name disambiguation, family consolidation, and classification noise demand heavy cleaning, and errors there bias every indicator.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Draws on vast, structured, publicly available datasets covering decades of worldwide inventive activity.
- Provides quantitative, reproducible indicators—counts, citations, classes—that support objective benchmarking and trend detection.
- Reveals the structure of technology fields and the linkages between them through classification and citation analysis.
- Offers early, fine-grained signals of emerging technologies and competitor activity that feed foresight and strategy.
Applications
- Competitive technology intelligence, benchmarking a firm's patent portfolio against rivals.
- Technology foresight and trend detection, spotting fast-growing or emerging technology areas.
- Identifying technological white spaces and opportunities for R&D investment or acquisition.
- Supporting freedom-to-operate, licensing, and litigation through landscape and citation analysis.
- Science-and-technology policy analysis, comparing national or regional innovation performance.
- Mapping technology convergence and knowledge flows across fields via citation and co-classification networks.
How it works
How it works
Define the technology field and analytical question. Analysts specify the technology of interest, the entities or markets in scope, the time window, and what the analysis must answer—competitive position, white spaces, emerging trends—because these choices drive the search strategy and the indicators that follow.
Retrieve a relevant patent corpus from one or more databases. Using keyword, classification (IPC/CPC), assignee, and citation queries against sources such as PATSTAT, Derwent, Google Patents, or Lens.org, the analyst assembles the set of patents and patent families that represents the field, balancing recall and precision.
Clean and normalise the data. Applicant names are harmonised (the same company appears under many spellings), patent families are consolidated to avoid double counting across jurisdictions, dates and classifications are standardised, and irrelevant or duplicate records are removed so that downstream indicators are not distorted.
Compute bibliometric and patent indicators. Patent counts, growth rates, citation counts and forward/backward citation links, family size, technology-class distributions, co-classification, and concentration measures are calculated to quantify activity, influence, breadth, and the structure of the field, following the patent-bibliometrics tradition.
Map, visualise, and interpret the results. Indicators are turned into technology landscapes, patent maps, citation networks, and trend charts, and interpreted in context—triangulated with domain knowledge and other evidence—to draw conclusions about competitive positioning, emerging technologies, and strategic opportunities.
Common pitfalls
- Comparing raw patent counts across sectors or countries without normalising for differing propensities to patent.
- Skipping applicant-name harmonisation and family consolidation, so the same entity or invention is counted many times.
- Reading citation counts as direct quality measures while ignoring examiner-added citations and field-specific citing norms.
- Interpreting a single database as the whole picture, missing jurisdictions or technologies it covers poorly.
Frequently asked questions
Are patents a reliable indicator of innovation?
They are a useful but partial proxy. Patents capture codified, patentable inventions, but the propensity to patent varies greatly across industries (high in pharmaceuticals and electronics, low in software or services), firms, and countries, and much innovation is never patented. Strategic and defensive patenting can also inflate counts. Patent analysis is therefore most reliable for patent-intensive fields and is strongest when triangulated with other indicators such as publications, R&D spending, and market data.
What is the difference between a patent and a patent family?
A single invention is often protected by separate patent applications filed in multiple jurisdictions (e.g. the US, Europe, Japan). A patent family groups all these related filings of the same invention. Counting individual documents without consolidating families overstates activity and biases comparisons toward inventions filed in many countries, so serious analysis works at the family level and uses family size itself as an indicator of an invention's perceived value.
Which databases and tools are used for patent analysis?
Common data sources include the EPO's PATSTAT, Clarivate's Derwent World Patents Index, Google Patents, and Lens.org, alongside national and WIPO collections. Analytical tools range from Derwent Innovation and VantagePoint for cleaning, indicator computation, and mapping to general data-science and text-mining environments. The choice depends on coverage needs, the depth of citation and classification data required, and whether open or commercial access is preferred.
Related methods
Sources
Narin, F. (1994). Patent bibliometrics. Scientometrics, 30(1), 147-155.
Open DOI referenceCopied.Abbas, A., Zhang, L., & Khan, S. U. (2014). A literature review on the state-of-the-art in patent analysis. World Patent Information, 37, 3-13.
Open DOI referenceCopied.