1. We would find evidence that someone was doing experiments on furin cleavage sites in coronaviruses.
2. We would find that the furin cleavage site matched a known sequence.
We did not find 2, which should have been easy to find. Eventually, we found 1, since a rejected DARPA grant proposal affiliated with the lab in Wuhan emerged.
Then, we found 2!! with a plausible explanation of why no one had found it earlier -- you had to turn on "patented sequence search" on NCBI blast, which is turned off by default. An easy mistake to make. As a professional biochemist for a decade I had no idea that blast had a patent sequence search, much less that it was turned off by default (which, btw is a sensible setting).
The scientific method says given a model it should make independent predictions (ideally unlikely) which are later verified. The evidence for a lab leak should be considered relatively strong.