Why does it matter?
Every rate you report (conversion, engagement, click-through, cost per visit) is divided by a number of visits. If a large share of that number was never a person, every rate built on it’s wrong, in the flattering direction, for as long as it goes unnoticed.
How can you tell a visit wasn’t a person?
The strongest tests are physical rather than behavioural. A single visit can’t genuinely take place in two countries at once. A test like that has no threshold to argue about and behaves the same way on every site.
Behavioural scoring can catch more, but it tends to behave differently from one site to the next. And a measure that changes from customer to customer isn’t really a standard.
Does removing it cost you conversions?
It shouldn’t, and it’s worth checking. On the site above, removing a third of all visits removed zero conversions. That’s the evidence the test was catching the right thing.