Every inventor faces this decision early, usually with a limited budget and a strong hope that the free option is good enough. Sometimes it is. A self-run search is genuinely useful for killing bad ideas quickly and cheaply, and killing bad ideas quickly is most of what early-stage work is. What a self-run search cannot do is give you confidence that something is patentable, and confusing those two outcomes is where money gets lost.
What a patent search is actually for
Three different questions get called a patent search, and they need different levels of rigor:
| Question | Search type | Realistic DIY? |
|---|---|---|
| Has someone already patented this? | Patentability / prior art | Partly |
| Can I sell my product without infringing? | Freedom to operate | No |
| Is this specific patent still enforceable? | Status check | Yes |
The distinction between the first two is the one that trips people up most often, and it is worth being clear on before spending anything — patentability versus freedom to operate explains why a clean patentability result tells you almost nothing about infringement risk.
What you can genuinely do yourself
The tools are free and they are the same databases professionals use. The gap is method, not access.
The tools
- USPTO Patent Public Search — the authoritative US database, with full-text and classification searching. It has a learning curve, and a walkthrough of the interface saves an evening.
- Google Patents — the friendliest starting point, with machine translations and citation graphs. The step-by-step method covers how to use it properly rather than as a keyword box.
- Espacenet and WIPO Patentscope — essential for European, Asian, and PCT documents that a US-only search will miss entirely.
The method that separates a real search from a Google query
- Describe the invention functionally. Write what it does, not what you call it. Patent attorneys name things strangely on purpose.
- Build a synonym set. Your "clip" is someone else's retainer, clamp, bracket, or fastening member. Missing a synonym is the single most common reason a DIY search comes back clean when it should not — choosing keywords for a patent search covers how to build the set.
- Switch to classification. Find two or three relevant patents by keyword, note their CPC codes, then search the classification directly. This is the step amateurs skip and it is the step that finds the dangerous references, because classification does not depend on vocabulary. Searching by CPC and IPC explains the hierarchy.
- Follow the citations. Read the references cited on the closest patents, and the later patents citing them. Examiners already did work you can inherit.
- Read the claims, not the abstract. The abstract describes the disclosure; only the claims define the right, and a patent whose claims are narrow may be far less threatening than its title suggests.
Expect eight to fifteen hours for a competent first pass, and keep a log of every search string and classification you tried — it is the record that tells an attorney where you already looked.
Where DIY reliably falls short
- Non-patent literature. Prior art includes any public disclosure: academic papers, product manuals, trade catalogs, conference posters, archived web pages, even a YouTube video. Patent databases contain none of it, and examiners increasingly cite it.
- Foreign-language documents. Machine translation of a Chinese or Japanese utility model is often good enough to notice a problem and rarely good enough to characterize one.
- Classification blind spots. The most damaging references usually come from a field you would never think to look in — a medical device idea anticipated by a patent on agricultural equipment. Professionals search analogous arts systematically.
- Claim interpretation. Deciding whether a reference actually anticipates your invention, or merely looks similar, is a legal judgment. Inventors are systematically optimistic here, and it is the whole reason a search matters.
- Pending applications. Anything filed within the last eighteen months has not published. No search of any kind can see it.
What a professional search buys
A professional search is done either by a dedicated search firm or by a patent attorney, and typically delivers a set of closest references, a classification map, and an assessment of relevance. Rough US ranges:
| Deliverable | Typical cost |
|---|---|
| Basic patentability search and reference list | $500–$1,500 |
| Patentability search with an attorney opinion | $1,500–$3,500 |
| Freedom-to-operate analysis for one product in one market | $5,000–$20,000 and up |
| Design patent / product appearance search | $400–$1,200 |
The pricing detail is in what a professional patent search costs, and the FTO tier is a different animal entirely, described in what a freedom-to-operate search covers. Note the distinction between a search and an opinion: a search firm delivers documents, a patent attorney delivers a legal conclusion. The opinion costs more and is the only one that carries legal weight.
The sequence that wastes the least money
- Do a DIY search first. Two evenings. If you find your exact invention already patented and in force, you just saved thousands of dollars and the decision is made.
- If it survives, refine the concept around what you found. Knowing the closest art shapes a better invention and a better claim strategy.
- Then pay for a professional patentability search before authorizing drafting. Spending $1,000 to protect a $12,000 drafting decision is straightforward arithmetic.
- Add an FTO analysis before tooling, not before prototyping. FTO matters when you are about to commit capital to production or enter a market, not while the design is still moving.
Reading the resulting report is its own skill — the relevance codes and the reference ranking carry more meaning than the summary paragraph, as how to read a patent search report explains. And if a close reference does turn up, it is rarely fatal — most of the time the answer is a design change rather than an abandoned project.
The honest bottom line
DIY is excellent at proving an idea is taken and poor at proving it is free. Professional search is the opposite: it costs money and it substantially reduces the chance of an expensive surprise during examination. Run both, in that order, and treat neither as a guarantee — no search is exhaustive, and unpublished applications sit in a blind spot nobody can see into.
Projects House is an engineering firm, not a law firm, so the patentability opinion belongs with a registered patent attorney. What we do sits either side of the search: translating a rough idea into the precise functional language a search needs, and redesigning around a close reference so the product clears it and still manufactures at cost. If a search turned up something worrying, send the patent number through our contact form.