Clicks by bots / spam filters
The main problems with email are spam, phishing, and the spread of malware. Of course, organisations aim to protect themselves from this. They do this using advanced filters, which judge all incoming mail. To block spam, these filters look at the content of the email, the use of images, and at technical aspects such as DKIM and SPF. To prevent phishing and malware, there is an increasing number of filters that follow the links in an email, to see what page they are directed to.
This is of course good practice, but our system records following a link as a click. This is not intended, because it does not concern human clicks. It is not advisable to include these clicks in the reports, or to use them as a startevent for an automation.
Therefore, our system tries to ignore these automatic clicks as much as possible.
How does Laposta recognize clicks from filters and scanners?
Since June 2021, we've tried to exclude clicks from scanners from our statistics. Our chosen approach did not prove effective enough, and thus we built a new system, which has been active since October 2023.
This system registers all incoming events (opens, clicks, subscriptions, unsubscribes, etc.) before they appear in the reports. Each event is judged on a number of aspects. The score for the individual aspects determines whether it concerns a human or a computer. Among other things, we look at:
- Rapid click sequences from the same email address;
- Rapid click sequences from the same IP address (possibly in combination with the user agent);
- The type of event;
- IP addresses that we have noticed are often scanners, or where we know they are not scanners at all.
These individual ratings are bundled in a final score, which determines if the event can be included in the results.
Influence on open ratio and click ratio
Before we implemented this system, automatic clicks were not filtered and thus included in the statistics, which was noticeable in the open and click ratios. These ratios may end up lower after the filter became active. This might seem unfortunate, but it does give you a better insight, as these were not human recipients in the first place.
Influence on automations
If you have set up an automation where an action needs to be carried out when someone clicks a (certain) link, then you do not want this to happen mistakenly. Because of this system's implementation, follow-up actions in your systems are only carried out when someone has actually clicked, and not when a scanner does.
Still clicks from a bot?
Are you still seeing scanner or robot activity in your Results? For instance, many clicks within one or two seconds? Please let us know if you encounter this, it helps us to further improve our filter.