From hypothesis to evidence record
An evidence-of-use search starts with a known product and selected claims. The researcher looks for attributable technical sources that support or challenge each claim limitation, then records the source, date, product version and strength of the connection.
Source quality varies
A product manual may directly describe a function. A marketing page may be ambiguous. A teardown may apply to only one model revision. Good work distinguishes these source types and does not hide the inferential steps.
The output is not an opinion
The resulting chart helps technical experts and counsel focus their review. It does not decide claim construction, infringement, validity or liability.
Three assessments that should remain separate
Supported means the cited material explicitly describes the selected feature for the relevant product. Inference means a technical reasoning step connects the source to the limitation. Unresolved means the scoped research has not found adequate support. These labels describe the evidence for a limitation; they are not legal conclusions about the product.
Worked example: automatic temperature regulation
Our fictional manual states that a controller receives a sensor temperature signal, directly supporting that selected feature. A second statement says it maintains a selected temperature automatically. That may motivate a hypothesis about a comparison with a threshold, but it does not disclose the implementation. A remote-command limitation has no supporting source. A useful chart shows all three outcomes instead of making every row appear conclusive.
When public research reaches its limit
Internal software logic, firmware behaviour and hidden components may not be established from public documentation. The report should say where research stopped and which further material or investigation could address the gap. Testing and reverse engineering should be separately scoped rather than assumed to be included in every EOU engagement.