[hrunt]

At a previous job, we had a product that was attractive with criminals. We were seeing some elevated chargeback rates and refund rates, so I dug in and determined that at least 25% of our new account purchases were fraudulent (unreported stolen credit cards). I put in some rudimentary fraud analysis, and injected a hoop that suspected fraud users had to go through before their account would be billed, simply to prevent the credit card networks from blacklisting our merchant account. The fraud rates dropped considerably with a small false-positive rate, but the new-user metrics slowed with it.

A week after I left my role, the company disabled it. They were sure that the system was a significant cause of some major userbase declines. Within three months, one of their credit card processors threatened to lock them out for elevated fraud rates, and they spent the next six months getting it resolved. Growth never returned.

No one was trying to be dishonest, but no one wanted to look bad, either. The entirety of the metrics showed something was off, but the growth narrative was so important, it was easy to ignore the questionable parts.

This was not a VC-backed business, so the pressure to perform was entirely internal. Nothing was faked, but the success wasn't (entirely) real, either.