Not "perfectly good proof" since a sceptic might still claim it was just luck. It would need to be demonstrated repeatedly elsewhere.
And that's the problem with dowsing, so few convincing successful sequences of tests*, so many less-than-fully-structured demonstrations.
*None?