How big can the distortion be?
On one enterprise site, a third of all visits couldn’t have been human. They came almost entirely from desktop, and not one of them had ever produced a sale. Every rate the team reported had been diluted by them.
Why does it change decisions, not just numbers?
Because rates are compared. Channels, pages and campaigns carrying more automated traffic look worse than they are, and budgets drift away from them. A team can spend a year fixing a “conversion problem” that was never a conversion problem.
Why isn’t this already handled?
Standard filtering catches the obvious cases. What slips through is traffic that looks ordinary visit by visit and only gives itself away in patterns no person could produce, like a single visit that turns up in several countries.
What is the first step?
Ask how many of last month’s visits could have been people, and what excluding the rest would cost in conversions. If the answer to the second question is nothing, the first number is the one you should have been reporting all along.