Dec 12, 2019 10:58 AM
When looking at how an email has performed, I've noticed that some of the clicks and opens are likely false positives because the email is sent, delivered, opened, and clicked multiple times, all at the same time. There have even been occasions where it says a contact both clicked an email AND hard bounced.
I spoke to support and the explanation was that this is likely done by a bot on the recipient's email server that first inspects the email, then routes it to the recipient. If the recipient doesn't exist, then that's a click and a bounce.
Since our business is B2B, a number of our target accounts seem to have servers with bots like this, meaning that the engagement information from every contact belonging to certain accounts is likely unreliable and I have to look at each contact who supposedly clicked to see if there was some space between when the email was delivered, opened, and clicked. There was at least one occasion where I thought an email got a lot of interest from a particular account, only to discover that all that engagement was likely just a bot.
Does HubSpot have any plans to filter out obvious false positives like this?
Our email performance metrics are less meaningful because of them. This is a serious problem because 1. we can't properly assess which emails do well for optimization purposes 2. I can't give an honest answer to what our actual open and click rates are and 3. we can't rely on email engagement for lead scoring.
Dec 12, 2019 1:30 PM
Jan 15, 2020 3:07 PM
Has this been resolved? We are coming across the same issue, and our numbers are skewed. People who want nothing to do with us are shown clicking an email 15 times in the first 10 min of opening the email.
Jul 23, 2020 5:25 PM
We're seeing the same thing for our B2B business and it's a real problem in determing how emails are ACTUALLY performing, as well as determining REAL interest in our business. We definitely would like to see HubSpot address this!
Jul 24, 2020 1:38 PM
I'm talking with a Product Manager on the team Bot filtering/ detection is a continued investment area for their team.
Right now at a high level, they use two approaches to bot filtering:
1. Blocking known IP ranges that correspond to bots or security software.
2. Behavior analytics-based approach. Looking at the characteristics of the events the systems make determinations on whether something is likely to be a bot.
Both of these the team is continually refining over time.
I hope this information helps.
Aug 20, 2020 5:00 AM - edited Aug 20, 2020 6:05 AM
I turned on bot filtering, but I still have the issue of emails being sent, delivered, opened, and clicked all at the same time not being filtered out. While it's plausible that an actual person might open and click an email the minute they receive it, if that all happens within less than, let's say, 5 seconds of one another then it's probably a bot (or an accidental click). On the user's end, the time stamp only goes down to the minute but I assume the devs can see seconds. Could that be factored into the behavior based approach?
edit: fixed typo