My issue is that a large share of what he tests are Amazon products with alphabet soup brand names, where QA is likely nonexistent and the conclusions are often based on a sample size of N=1. Even if you wanted to buy the "winner", the exact same product may be sold under a different name a week later.
I also find his testing methodology inconsistent. In some cases he takes manufacturer specs at face value without actually verifying them, in others he goes out of his way to comprehensively measure things that don’t matter much (to me anyways), while skipping things that seem genuinely important (self-discharge of jump starter packs for example).
That said, he's doing this with his own time and money, and makes it available for free to anyone. A lot of this also comes down to personal preference in what you value in a test.