I really like this article's structure for showing benchmarks. each benchmark has its own heading and a paragraph explaining the reason for it and analyzing the results.

I was a bit surprised, though, with how it ends right after the search benchmark. This seems like the perfect setup to talk about search/replace, which seems like the best use case for non-local edits that I can think of, and therefore would be a great opportunity to show how the algorithms compare in something the ropes should be better at.