The lesson
Before you show a benchmark table, teach the scoring with a small made-up example, so readers can judge the ranking for themselves.
A 'Benchmark' label and date sit over the headline and the line 'We tested 11 de-identification tools plus Health Universe on 5,866 documents from 4 collections'. A plain intro defines de-identification, four stat tiles give the headline scores, and a short section names the two real and two synthetic collections. 'Two Numbers: Recall and Precision' walks through one invented note scored by Tool A and Tool B, then a character-by-character example of the counting. The 'Results' table ranks every system by F2, tags each as a cloud service or local tool, adds recall for each collection, and closes with caveat notes and a 'REQUEST ACCESS' form for the full results.
Don't copy Health Universe's scores, competitor list or datasets. Keep the order of headline stats, worked example of the metric, full ranked table with per-dataset columns, then caveats and an access form.
Tell us what you want to make. We’ll talk through how to put these ideas to work for your brand.