Working with search

How to brief a prior-art search so you get a useful result

Most disappointing search results are not a searcher problem — they are a brief problem. A few minutes spent framing the request properly changes what comes back more than almost anything else. Here is what an experienced searcher actually wants from you.

1. Lead with the inventive concept, not just the field

"Search for prior art on our battery management system" is a field. The inventive concept is the specific thing you believe is new — for example, "balancing cell charge using a predicted temperature curve rather than a measured one." Searchers build queries around the concept, so the sharper you can state it, the closer the results sit to what matters. If you have an invention disclosure, claims, or even rough inventor notes, send them.

2. Say what the search is for

A novelty read before drafting, a validity hunt against an asserted patent, and a freedom-to-operate clearance are three different jobs with different standards of "done." Naming the purpose lets the searcher calibrate depth, date ranges, and how hard to chase non-patent literature. (If you are unsure which you need, this comparison of patentability, FTO and invalidity may help.)

3. Define scope — including what to leave out

Tell the searcher the boundaries: which features are core versus peripheral, which variants to include, and anything explicitly out of scope. A good exclusion ("we don't care about purely mechanical implementations") saves as much time as a good inclusion and keeps the report focused.

4. Name the jurisdictions and the dates that matter

Coverage is usually global through the major databases, but the jurisdictions you care most about shape where effort goes — and for validity or FTO work, dates are everything. Give the priority date, publication date, or product-launch date the search has to anchor to, so the searcher knows what counts as prior art and what does not.

5. State the output format up front

Do you want the closest 5–15 references ranked and annotated, or a full claim chart, or a feature-to-patent risk matrix? Do you have a house template? Saying so before work starts means the deliverable drops straight into your workflow instead of needing a reformat.

The one mistake that costs the most

Over-broad briefs. "Find everything related to X" produces a haystack, not a needle, and forces the searcher to guess at relevance — which is exactly the judgement you wanted them to apply with your guidance. A tight, well-bounded concept beats a sprawling one every time.

A good brief is three sentences of concept, one of purpose, one of scope, the key dates, and your preferred format. That is genuinely enough to start.

If you would like a short brief template to reuse, or want to talk a search through before committing, send a line — the first matter can be a no-risk pilot.

Hand me one real matter. See the work before you commit.

A free or fixed-fee pilot search on a case of your choice — documented strategy log, closest art, your report format. No retainer, no lock-in.