What an evidence plan actually needs before you run the first test

A testing programme usually starts with a list of tests. Flexural strength, degree of conversion, wear, maybe biocompatibility if the timeline allows. The list looks thorough, the work gets done, and then somebody asks what the results prove. That question should have come first.

An evidence plan starts at the other end. Write down the claim you need to make, name the audience that has to accept it, and only then choose the measurements. A claim aimed at a regulator, a claim aimed at a purchasing committee, and a claim aimed at a journal reviewer are three different claims, and they rarely need the same data.

The second thing an evidence plan needs is a stated comparator. A material that performs well in isolation says almost nothing. Reviewers want to know what it was measured against and why that comparator was chosen. Picking it after the results are in is visible, and it costs credibility.

The third is the analysis, agreed before collection begins. Endpoints, sample size, and how the numbers will be handled. Reconstructing an analysis afterwards is the most common reason a dossier or a manuscript comes back.