Intro.
A patent is only granted for something “new.” If the technology has already been publicly disclosed, already patented, or already described in a published paper, it lacks novelty and registration gets denied. No matter how confident a founder is that their idea is original, the patent examiner judges based on the database. The gap between confidence and data is what comes back as a rejection.
A prior art search isn't simply a matter of checking “does this already exist or not.” It's a strategic exercise: figuring out exactly what's original about your technology, which direction competitors are building their patent portfolios in, and how to design your claims.
Skipping the search creates three risks. First, you waste money filing for technology that already exists. Second, you launch a product that infringes a competitor's patent and get pulled into a dispute. Third, you file without ever capturing the core piece that actually needed protection in your claims. All three are losses an early-stage team can struggle to recover from.
02
The major databases used for prior art search are all free to access. Each tool differs in country coverage, search usability, and how you view the original document. Searching just one database leaves gaps — you need to run at least two in parallel.
| Tool | Operated By | Coverage | Free? | Notes |
|---|
| KIPRIS | Korea Institute of Patent Information | Korea, some overseas | Free | Optimized for Korean-language search. Full access to registered/published Korean patents |
| Google Patents | Google | 100+ countries | Free | Supports natural-language search. Built-in translation makes foreign-language patents accessible |
| Espacenet | European Patent Office (EPO) | 150+ countries | Free | Focused on European/PCT filings. Strong for tracking patent families |
| J-PlatPat | Japan Patent Office (JPO) | Japan | Free | Original-text lookup for Japanese patents. English abstracts provided |
| USPTO Full-Text Database | United States Patent and Trademark Office | United States | Free | Full access to registered/published US patents. Strong classification-code-based search |
Even if you're only targeting a domestic filing, using Google Patents alongside KIPRIS is the baseline. If you have a global market in mind, add Espacenet to check PCT filing status as well. Using patent classification codes (IPC, CPC) helps you work around the limits of keyword search.
03
Founders doing this for the first time often just throw search terms in with no system and call it done. Following the sequence below reduces gaps and gives you a documented result.
- Define your technology in a single sentence. Compress “what you're doing and how” into 30 characters or fewer, centered on the core verb and noun. This one sentence becomes the backbone of your search terms.
- Extract 3–5 core keywords. Pull one from each dimension — technical term, material, method, application area. List synonyms and near-synonyms in advance too.
- Run a first search on KIPRIS using Korean-language keywords. Skim the titles and abstracts of the top 30 results, and read the full claims for anything that looks similar.
- Run a second search on Google Patents using English-language keywords. Check for overseas prior art and PCT filings. If a patent classification code (CPC) turns up, run an additional search on that code.
- Organize your findings into a table. Capture patent number, applicant, filing date, key claim content, and how it differs from your technology, all in one row. This table becomes the foundation of the prior art search report you'll bring to your first meeting with a patent attorney.
Completing all 5 steps typically takes 4–8 hours. If the technical field is complex or there's a large volume of prior patents, it's better to split the work across two days. Reading claims while fatigued makes you miss similarities.
04
Most mistakes trace back to two root causes: casting too narrow a search net, and interpreting the results too optimistically. The table below shows how to handle each type.
| Mistake Type | Specific Situation | How to Fix It |
|---|
| Keyword tunnel vision | Searching only with the exact terms you use for your own technology, missing prior art described in different language | Search synonyms, broader umbrella terms, and English terms in parallel |
| Domestic-only search | Checking only KIPRIS and missing overseas filings | Running Google Patents and Espacenet in parallel is essential |
| Registered-patents-only check | Skipping rejected/withdrawn published applications — publication alone destroys novelty | Search both published applications and registered patents |
| Skipping the claims | Judging “this is different” from the title and abstract alone, when the claim scope is actually much broader | Always read Claim 1 in full for any related patent |
| Ignoring non-patent literature | Papers, technical reports, and open-source disclosures also count as prior art | Run additional searches on Google Scholar and arXiv |
| Skipping documentation | Finishing the search with no record and just concluding “doesn't seem to exist” | Save the list of discovered patents as a table and share it with your patent attorney |
Patent examiners find prior art the applicant never searched for. “I couldn't find it” is not evidence that “it doesn't exist.” The goal of the search is to reduce risk — not to earn yourself the confidence that “nothing's out there.”
05
It's most efficient to bring in a patent attorney once you've completed every item below. If any item is unchecked, go finish that piece of the search first.
- You've defined your technology's core operating principle in a single sentence.
- You've listed 3+ Korean-language keywords.
- You've listed 3+ English-language keywords.
- You've reviewed 20+ related patents on KIPRIS.
- You've searched English keywords on Google Patents and checked for overseas prior art.
- You've read the full Claim 1 for any similar patents you found.
- You've searched non-patent literature (papers, technical reports) in at least one database.
- You've organized your findings into a table including patent number, applicant, claim summary, and differences.
- You've written 2–3 lines describing how your technology differs from the closest prior art you found.
- You've prepared your findings as a document ahead of your first meeting with a patent attorney.
Going into a consultation with a patent attorney after completing 7 or more of these 10 items shortens the consultation and sharpens the accuracy of your filing strategy. The more research the founder has already done, the more the attorney can focus purely on drafting strong claims.
Summary.
Q. If I find prior art, do I have to give up on filing?
Not necessarily. Novelty is only defeated if the prior art covers your entire technology. If there's a difference from the prior art, you can capture that difference clearly in your claims and still file. The point of the search isn't to give up — it's to pin down exactly where your differentiation lies.
Q. Does a founder's own DIY search carry any legal weight?
A founder's own search is informal. The official prior art search report submitted to the patent office is prepared by a patent attorney or the patent office itself. That said, doing your own search before filing directly contributes to your strategy and cuts costs. It gives your attorney better information to work with and can help you avoid unnecessary filings.
Q. What should I do if I find a competitor's patent?
Check three things. First, is that patent still valid (term remaining, still registered)? Second, does your product fall within the scope of its claims? Third, is a design change available that routes around that patent? All three of these need to be reviewed together with a patent attorney. A competitor's patent existing doesn't automatically mean infringement.
Q. We're still at an early stage — is it okay to put off the prior art search?
The right time to start is once your product-development direction is set. Discovering a competitor's patent infringement after launch means the cost of redesigning your product is far higher than the cost of filing would have been. Government grant reviews are also increasingly checking whether applicants have an IP strategy in place. The earlier you start searching, the more options you keep open.
CTA
Before you fill out the IP-strategy section of your business plan, check whether the technical-differentiation narrative in your current draft meets review standards. OpenSeed provides an analysis report across 15 perspectives, including the technology section. Free during the current beta.
Get Your Business Plan Reviewed Now
Get your business plan's technical strength reviewed now — free during the current beta.
🔒 Free during beta · your submission isn't saved
Start Free AI Feedback →