Anatomy of a Prior Art Search: A Worked Example From Idea to Report

Most guides to prior art search tell you what to do in theory. This one shows you what it actually looks like in practice, using one invention from start to finish, with the real search terms, the real result counts and the real judgement calls that got made along the way. Meet Priya, our example inventor, who spent a Saturday afternoon working out whether her folding bicycle pedal idea already existed. By the end, she had a shortlist of four references, a completed comparison table and a clear next step. Here is exactly how she got there.

Step 1: Define the invention in plain English

Priya's idea was a bicycle pedal that folds flat against the crank for storage, but locks itself rigid the moment you unfold it to ride. Existing folding pedals either needed a separate pin to hold them open, or wobbled slightly under load because the hinge had play in it. Her version used a spring-loaded catch that clicked into place automatically as the pedal swung open, and a small thumb lever on the pedal's outer edge to release it again for folding.

Before searching anything, she wrote one sentence describing what was actually new: "a folding pedal hinge that locks itself automatically on opening, released by a lever built into the pedal body, rather than a separate pin or manual latch." That sentence became her compass for the whole afternoon. Every reference she found afterwards got measured against it.

This step matters more than people expect. A vague idea produces a vague search, and a vague search misses things. Novelty has to be pinned down in words before you can look for it in a database. If you want a broader walkthrough of this stage, our guide on how to do a patent search covers it in more depth.

Step 2: Break it into discrete technical elements

Next, Priya split the invention into its separate working parts. Patents are usually infringed or blocked element by element, so search terms work best when they map onto real components rather than the idea as a whole.

She settled on four elements:

1. The folding hinge itself
2. The self-locking catch mechanism
3. The release lever or trigger
4. The pedal platform (the flat part your foot sits on)

Breaking an invention down like this, sometimes called element decomposition, stops a search from becoming one giant vague query. Instead, you get four smaller, sharper questions, each one easier to check against existing patents.

Step 3: Build a synonym table

Patent documents rarely use everyday words. A hinge might be called a "pivot," a "pivoting joint" or an "articulated connection." A lock might appear as a "detent," a "catch" or a "spring-biased latch." This is where synonym expansion comes in: writing down the patent-speak alternatives before you start typing search terms, so you are not relying on guesswork mid-search.

Priya's table looked like this:

This table alone doubled the number of useful hits she found later. Skipping this step is one of the most common reasons a DIY search comes back looking clean when it isn't.

Step 4: Run the keyword searches

With her element list and synonym table ready, Priya went to Google Patents and searched "folding bicycle pedal." That returned around 1,240 results, far too many to read one by one. She narrowed it: "folding pedal hinge lock" brought that down to 340. Adding "spring" and "release lever" together narrowed it further, to 58 results, a manageable number to skim by title and drawing.

She repeated the process on the USPTO Patent Public Search tool, which uses slightly different indexing and turned up two documents Google Patents had ranked lower. That is a useful lesson on its own: searching only one database, even a good one, leaves gaps.

As a final cross-check, she ran her one-sentence description through an AI-powered search tool built for this kind of task, which reads the invention conceptually rather than matching exact words. It surfaced one more candidate that neither keyword search had caught, a scooter patent using similar wording for its locking geometry but never mentioning "pedal" at all. That is exactly the kind of document a pure keyword search tends to miss, since the underlying idea can be filed under completely different vocabulary.

From all three passes, Priya triaged her results by drawings and titles first, ruling out anything clearly aimed at car pedals, industrial machinery or unrelated fold-flat furniture. That left nine documents worth a closer look.

Step 5: Find the closest CPC codes and browse them

Keyword searches only catch documents that use the words you thought of. Classification codes catch documents that don't. The Cooperative Patent Classification (CPC) system groups patents by subject rather than wording, so browsing the right code can surface things a keyword search never would.

Priya identified three candidate codes for her invention, illustrative examples of the kind of format CPC codes use rather than a guarantee of the exact registered classification:

• B62M 3/08 — pedals for cycles
• B62M 3/00 — crank and pedal gear more broadly
• B62K 15/00 — folding or collapsible cycles

Browsing B62M 3/08 directly, rather than searching by keyword, turned up 22 documents that her keyword searches had missed entirely, mostly because they used unusual phrasing like "collapsible foot support" instead of "pedal." You can read more about how this classification system works at the Cooperative Patent Classification site.

Combining the CPC browse with her keyword shortlist, Priya ended up with a final list of four documents worth reading properly.

Step 6: Read the claims, not just the abstract

This is the step people skip, and it is the one that matters most. An abstract or a drawing can look similar to your idea and still not actually cover it, because what counts legally is the wording of the claims. Priya read all the independent claims on her four shortlisted documents, one at a time, and made a call on each.

Reference A was the one that gave Priya pause. Its claim described a hinge that locks automatically via a spring-loaded pin engaging a slot when the pedal is unfolded, which is close enough to her own mechanism that a patent examiner could plausibly read it as covering the same thing.

Step 7: Make the judgement call

Having read the claims rather than just the pictures, Priya's conclusion was straightforward. Reference A looked like a genuine blocker for the locking mechanism as she had first imagined it. But none of the four references combined a self-locking hinge with a release lever built directly into the pedal body itself, rather than a separate pin, magnet or external clip. That combination looked like her one remaining gap.

As one patent search specialist puts it: "The DIY stage isn't about proving you're clear. It's about finding out fast whether you're obviously blocked, so you don't spend money drafting a patent for something that already exists." Priya's search didn't clear her idea entirely, but it did something more useful: it told her precisely which element still needed protecting, and which reference she would need a professional opinion on before going further.

Her next step was to commission a professional search focused specifically on Reference A and its patent family, before paying a patent attorney to draft anything. That is the sensible order of operations: DIY first to learn fast and cheaply, professional search second to confirm before you spend serious money. You can read more about typical costs at this stage in our piece on patent search cost.

Key takeaways

• Write your invention as one plain-English sentence before searching anything. It keeps every later step honest.
• Break the idea into three to five elements, then build a synonym table for each. This step alone often doubles useful search results.
• Search more than one database. Google Patents and USPTO Patent Public Search rank things differently.
• Browse CPC codes as well as keywords. Classification catches documents that use unfamiliar wording.
• Always read claims, not just abstracts or drawings, on your closest three or four documents.
• Score each reference clearly: blocks it, close but not identical, or not relevant.
• A clean DIY search is a green light to get a professional opinion, not a reason to skip one.

If you want to run this same process on your own idea, our tool for a prior art search follows the same element-by-element method described here, and our pricing page covers what a fuller professional check costs when you are ready for that step. For a deeper look at what actually counts as prior art and why it matters legally, see our companion piece on prior art search explained.

Sources and further reading

• Google Patents
• USPTO Patent Public Search
• Cooperative Patent Classification