Before You File a Patent: What a Prior Art Search Finds and What It Can Miss

An inventor who files a patent application without checking existing patents risks paying filing fees for an invention that can’t be protected. A search before filing costs far less than an application that fails at examination. The search has limits, though, and knowing them matters as much as running it.

Why novelty is the test

To be patentable, an invention has to be new and inventive. Baxter IP, an Australian patent firm, explains that both tests are generally judged against documents published anywhere in the world before the application’s priority date.

Prior art is whatever falls inside that body of documents. Patents are the largest part, but articles, theses and product literature can count too.

What a search is for

A search before filing gives three things:

• An indication of whether the application is likely to be allowed
• A picture of the existing field
• Early sight of the documents an examiner is likely to raise

An inventor who sees the closest documents can narrow the claims, change the design or decide not to file.

Where the searching happens

WIPO’s PATENTSCOPE database is a free starting point. According to an Intepat guide, it covers 123.8 million patent documents as of 2026, including 5.2 million published PCT applications and national collections from more than 70 patent offices. Every PCT application appears on its publication day, and publication runs every Thursday.

The same guide lists the main search modes:

Mode	Use
Simple	Searches titles and abstracts by default
Advanced	Boolean operators, proximity and field codes
Field Combination	Narrows by applicant, inventor or date
CLIR	Translates a query into up to 13 languages
Chemical Structure	Structure, sub-structure and Markush searches

Searching in several languages matters because relevant documents are often filed in the inventor’s home country first.

What a search can miss

Three limits apply to any search, including a careful one.

The 18-month gap

Baxter IP notes that applications are published 18 months after their priority date. A filing from the past year is invisible to a search, so a clean result can still sit alongside a competitor’s pending application that predates the inventor’s.

Coverage varies

Intepat points out that smaller patent offices may supply only bibliographic data, and that full-text search of descriptions and claims is available for the US, Japan, China, South Korea and the European Patent Office. PATENTSCOPE also doesn’t show legal status, so a hit can’t be assumed to be in force.

No search is exhaustive

Baxter IP states that no international patent search is thoroughly exhaustive, because not every document published worldwide can be searched. Intepat adds that non-patent literature isn’t comprehensively indexed in PATENTSCOPE, and that a clean result shows only what the database indexed.

A workable sequence

1. Write a plain description of the invention’s core features
2. Pick keywords and classification codes, including alternate terms
3. Search by keyword, then by inventor and applicant names
4. Read the claims of the closest hits as well as the abstracts
5. Record what was searched, so the work can be repeated or extended

Recording the search matters when it is later handed to a patent attorney, who will check the same ground and may add sources the inventor missed.

A worked search, step by step

Suppose an inventor has designed a folding bicycle frame with a new hinge. The first pass uses terms such as “folding frame”, “hinge” and “bicycle”, then adds synonyms like “collapsible” and “foldable”. The second pass searches the names of known folding-bike makers as applicants. The third reads the claims of the 10 closest hits to see whether any describe the same hinge mechanism.

Even with a good result, the inventor should note the date of the search, since new documents appear every Thursday on PATENTSCOPE and the picture shifts.

Who should run it

Inventors can run a first search themselves and learn from it. A professional prior art search adds access to paid databases, trained use of classification codes and a written view of how the claims compare to the closest documents. For an invention with real commercial value, the extra cost is small beside the cost of prosecuting an application that doesn’t proceed.

What to do with the result

If the search finds a close match, the inventor has three options: narrow the claims to what is different, redesign around the match or drop the filing. If it finds nothing close, the application can go ahead with reasonable confidence, bearing the 18-month gap in mind.

Baxter IP lists a second route, an international-type search based on a provisional application, for inventors who prefer to file first and search afterward. The choice between the two depends on how soon a priority date is needed.