It's even worst than that. Most of the time the validity of the t-test that they are running is questionable. They are technically running an online t-test and as soon as they find significance they stop. This is fundamentally wrong, and not conclusive at all.