How to Conduct a Prior Art Search

A prior art search is the process of systematically looking for evidence — patents, published patent applications, and non-patent literature — that an invention, or something that would render it obvious, was already publicly available before a planned patent filing. It is the practical, hands-on counterpart to the legal concept of prior art defined under 35 U.S.C. § 102: rather than asking what legally counts as prior art, a prior art search is the work of actually finding it before an examiner, or a competitor, finds it first. This guide walks through what a prior art search includes, which databases to use, how to structure the search itself, and when a preliminary self-search is enough versus when a professional patentability search is warranted.

What a Prior Art Search Includes

A thorough prior art search is not limited to issued patents. It typically includes:

Issued patents and published patent applications — both U.S. and foreign, since a reference does not have to be a U.S. document to count as prior art against a U.S. filing.
Non-patent literature (NPL) — peer-reviewed journal articles, preprints, conference proceedings, theses and dissertations, technical reports, and grant abstracts, once publicly accessible.
Products, demonstrations, and sales — a product already on sale or publicly demonstrated (at a trade show, for instance) can be prior art even without any published document behind it.
The inventor’s own prior disclosures — a paper, poster, or talk the inventor themselves gave can become prior art against their own later application; see the prior art dictionary entry for how the one-year grace period under 102(b) works and where it does not apply.

Because the search has to cover both patent and non-patent sources, no single database is sufficient on its own — a real prior art search combines a patent-database search with a separate literature search.

Patent Databases to Search

Four databases cover most of a practical prior art search, each with different strengths:

USPTO Patent Public Search (ppubs.uspto.gov) — the USPTO’s free, official search tool, which replaced the legacy PatFT/AppFT systems. It provides full-text search of U.S. patents and published applications and supports both keyword and Cooperative Patent Classification (CPC) searching. Because it is the same corpus a USPTO examiner searches, it is the natural starting point for any U.S.-focused search.
Google Patents (patents.google.com) — a free, easy-to-use interface covering patent documents from roughly 100+ patent offices worldwide, with full-text search, citation links, and a built-in “Prior Art Finder” that suggests related documents and scholarly literature for a given patent or application. Its breadth and simple interface make it a good first pass, though it should not be treated as a substitute for a jurisdiction-specific database when precision matters.
Espacenet (worldwide.espacenet.com) — the European Patent Office’s free search tool, covering tens of millions of patent documents from over 100 countries, including machine-translated full text for many non-English documents. Espacenet’s classification search (built around CPC, which the EPO co-developed with the USPTO) is particularly strong for European and Asian filings that a U.S.-only search might miss.
WIPO PATENTSCOPE (patentscope.wipo.int) — the World Intellectual Property Organization’s search tool, covering published PCT international applications along with national collections from more than 70 participating patent offices. It offers full-text search of PCT applications from the day of publication and cross-lingual search that translates a query across multiple languages — useful for surfacing foreign-language prior art that an English-only keyword search would miss entirely.

Searching all four is not redundant: USPTO Patent Public Search and Google Patents are the fastest way to check U.S. and broadly-indexed international coverage, while Espacenet and PATENTSCOPE are more likely to surface non-U.S. filings — particularly relevant given that, as covered in prior art, most foreign patent offices apply an absolute novelty standard with no U.S.-style inventor grace period.

Searching Non-Patent Literature

Because journal articles, preprints, theses, and conference papers all count as prior art once publicly accessible, a prior art search that stops at patent databases is incomplete. Non-patent literature searching typically draws on:

Discipline-specific and multidisciplinary journal databases (e.g., PubMed, Scopus, Web of Science, IEEE Xplore, depending on field)
Preprint servers and institutional repositories
University thesis and dissertation repositories
Conference proceedings and trade-show materials
General web and Google Scholar searches, which Google Patents’ own Prior Art Finder draws on directly

Non-patent literature searching matters most in fast-moving research fields, where an academic publication frequently predates any corresponding patent filing — sometimes the inventor’s own.

Search Strategy: Keywords and Classification Codes

The USPTO’s own guidance for conducting a preliminary search — commonly referred to as the seven-step search strategy — combines two complementary approaches rather than relying on either alone:

Brainstorm search terms. Write a brief, accurate description of the invention and list synonyms and alternate terminology for its key concepts, features, and function — avoiding overly broad, generic terms like “device,” “system,” or “process” on their own, since those return unmanageably large result sets.
Run an initial keyword search across the chosen database(s) using those terms and their combinations, to get an early sense of the relevant technology landscape and to identify recurring terminology used by others working in the same space.
Identify relevant classification codes. Patent offices classify every patent by subject matter under the Cooperative Patent Classification (CPC) system, jointly developed and maintained by the USPTO and the European Patent Office. Because classification search groups documents by what the invention actually does rather than by the specific words used to describe it, it catches relevant prior art that a keyword search misses when different inventors describe the same idea in different language.
Search by classification and review results, looking first at each result’s abstract and representative drawing to triage a large results list down to a smaller set of genuinely similar documents.
Review the most relevant documents in full — specification, additional drawings, and especially the claims — for the set identified in the previous step.
Follow the citation trail. References cited by the applicant or by the examiner on any closely related patent frequently lead to additional relevant prior art that a direct keyword or classification search did not surface on its own.

In practice, keyword and classification search are not alternatives — a thorough search alternates between them, using early keyword hits to identify the right classification codes, then using classification search to catch documents that use different terminology for the same concept.

Preliminary Self-Search vs. a Professional Patentability Search

An inventor or a technology transfer office (TTO) can, and often should, run a preliminary prior art search themselves before deciding whether an invention is worth pursuing further. A self-search using the databases above is useful for:

An early, low-cost sanity check on whether an idea appears genuinely new before investing further time or committing patent budget
Informing the initial triage decision a TTO makes when reviewing an invention disclosure
Surfacing obviously close prior art early, before any filing cost is incurred

A preliminary self-search has real limits, though. It is rarely as exhaustive as a professional search — a self-search typically covers fewer databases, is more likely to miss foreign-language and non-patent literature, and is performed by someone without the specialized search-and-claim-analysis training a professional patent searcher or patent attorney has. A formal, professional patentability search, usually commissioned once an invention has cleared initial triage and before committing to the cost of a full non-provisional filing, is typically warranted when:

The institution is preparing to commit real filing cost — drafting and filing fees, and potentially international filing — and wants a documented, professional opinion on patentability first
The technology area is crowded or fast-moving, where a self-search is unlikely to be exhaustive enough to be reliable
A formal, written patentability opinion is needed to support a licensing or investment decision, or to support the institution’s internal Bayh-Dole election-of-title decision
The search needs to be conducted by, or reviewed by, a registered patent attorney or agent, whose professional judgment on how a claim would likely be examined against the search results goes beyond what a database search alone provides

A professional patentability search is not simply a more thorough version of the same keyword search — it is typically conducted by a specialist with access to additional commercial search platforms, and is paired with an attorney’s or agent’s substantive analysis of how the located references map onto the likely claims, something a preliminary self-search does not attempt to do. Institutions should also be alert to firms that market expensive “invention promotion” or “patent search” packages to individual inventors with little real analysis behind them; see invention help companies for how to recognize a predatory invention-promotion firm.

What to Do With the Search Results

The output of a prior art search feeds directly into the decisions that follow it. If the search turns up nothing that anticipates the invention or renders it obvious, that supports moving forward — typically with a provisional patent application to secure a filing date while a full non-provisional application is prepared. If the search turns up close references, those results inform how the claims should be drafted to distinguish the invention from what is already known, or, in some cases, whether the invention is worth pursuing at all. Either way, the search results and the reasoning behind them are typically documented as part of the invention disclosure record a TTO maintains, and, for federally funded inventions, feed into the timing pressure described in the prior art entry between a planned publication date and the institution’s internal Bayh-Dole election-of-title deadline.

Frequently Asked Questions

What does a prior art search include?

A prior art search includes both patent literature (issued patents and published patent applications, U.S. and foreign) and non-patent literature (journal articles, preprints, conference papers, theses, and products already on sale or in public use). A search limited to patent databases alone is incomplete, since non-patent publications and public disclosures count as prior art too.

Is Google Patents enough on its own for a prior art search?

Google Patents is a strong, free starting point — broad coverage, full-text search, and a built-in Prior Art Finder — but it should not be the only database searched. USPTO Patent Public Search, Espacenet, and WIPO PATENTSCOPE each have coverage or classification-search strengths Google Patents does not fully replicate, particularly for non-U.S. filings and non-English-language documents.

What is the difference between a prior art search and a patentability search?

The terms are often used loosely, but a useful distinction is that a preliminary prior art search is the fact-finding step — locating relevant references — while a formal patentability search typically adds a professional’s substantive opinion on how those references would likely affect the patentability of specific claims, usually conducted or reviewed by a registered patent attorney or agent.

Should an inventor do their own prior art search before filing?

A preliminary self-search is a reasonable and low-cost first step, and many technology transfer offices expect one as part of an invention disclosure. It should not, however, substitute for a professional patentability search before committing significant filing cost — a self-search is rarely as exhaustive, and does not include a professional’s analysis of how the results would likely be applied against specific claims.

For the broader technology transfer landscape this topic sits within, see the tech transfer pillar page.