When it comes to Bot filtering in Marketo may be confused by the setting options and how they work. Why does it matter? Bot activity is seen by Marketo as an interaction with the email asset. Whether it is an open action or a click action these are recorded in the person’s activity history. When you use a smart list trigger of filter for opens and clicks, Marketo is looking for these types of actions in the person’s activity. If these activities are the result of a bot and not the actual recipient, you can see how this can distort your program performance report. For example, the email link performance report may show very high click activity because it is including repetitive clicks in addition to unique person clicks.
The intent of this article is to try to explain the settings and how to run some tests to determine the best set things for you. A couple of things to first note is that there are two different types of filters which you can enable. The IAB List match and the Proximity Pattern match. Each is independent of the other so you can enable them separately. There is also an action setting for each to either log the bot activity or filter the bot activity. As you are away, the email interactions are captured in the person’s activity log, and it is from these logs that many of the smart list filters search to find matching criteria. These settings determine what happens when a BOT activity match occurs. Specifically, if either of the bot identification methods are enabled and Log Bot Activity is selected, the activity match details will be included in the history for it in your user’s profile/activity log. If choose to Filter Bot Activity, when the pattern is matched that activity is suppressed from the log. Therefore, you do not have the activity in the history to evaluate because it was never generated. Since the data is not stored, I recommend that you start with Log Bot Activity option and then later filter when you understand the patterns for your use cases.
You navigate to your Bot Activity Identification control panel in the Admin section of Marketo … Admin > Email > Bot Activity (tab) which will look something like the image below. The image below is from a Marketo instance that has been enabled for about a year which includes historical counts. You can see that both the IAB List and Proximity Pattern matching option are enabled and each is including the match information in the person activity history log.
Matching Options
IAB List Match
Open or click activity is compared with data from the Interactive Advertising Bureau bot list to match anything on the IAB UA/IP (User Agent/IP address) to determine that it is a bot activity. This is a data match of the registered bot which uses the agent data or IP address of the service to determine if it is a “known” bot action. These are directly recognized and actioned according to the selected activity action to either include or exclude the activity in the log.
Proximity Pattern Match
The proximity pattern match is a data analysis to determine if the actions are human-like or bot-like. Human activity has inherent pauses between actions, where bots are rapidly crawling and clicking at extremely high rates. When a Proximity Pattern recognizes many interactions within a second or two it can make the assumption that the patterns are not likely to be human or the true actions of the e-mail recipient. With this matching option you see one additional setting called “Duration Between Activities.” This is the time window you want Marketo to use to assume that 3 or more actions represent a bot activity. The value can only be the integer 0, 1, 2, or 3 second(s). The image below shows examples of click actions that would match this pattern for the four different duration options.
As you can see, probability of the 0 second pattern match being bot activity is high. A user can click a link multiple times per second, but the click action will likely take the user to a web page, and they would need to switch back to the email to click again, which creates a more human-like pattern. Care needs to be taken when selecting longer durations, because you can start to get more false-positives which for bot-matching is a negative. (Well … you get what I mean.)
I did a test with an email and set the pattern to a longer duration, and I was able to create a manual click pattern that match the Proximity Pattern.
Choosing the Best Setting (for Your Business)
The technology world is always changing, and we must make a decision that is best for our business today. This means we may need to revisit our assumptions and decisions over time. This is especially true when we are establishing settings based on today’s technology. The catch-22 in this case is the decision to log verses filter based on the pattern match because logging does not fix those bloated email performance reports, but filtering can cause your to lose real activity data. Also, whether to use one or both pattern match options.
Which Match Options to Use
I have seen most businesses use the IAB List match with success. The choice to filter is less risky because you are matching to a data set, but I have heard that some actual user activity can still be filtered. The Proximity Pattern match is commonly used, but the durations are typically kept to either 0 or 1 because the higher values begin to include more actual user activity. The real decision is driven by your primary objective. If you want to minimize bot activity from inflating your program performance data, then you need to consider the filtering with the tightest data match. If you want to have all the data but use specific filter conditions to exclude certain activity with conditional constraints, then logging is the better choice.
NOTE: Remember that email click and open data has a 25 month retention period. So even if you choose to log the data, it will be removed from the activity log at a future date and the filter constraints may no longer work.
Test for the Best Option
Since these decisions can impact program actions and reporting it is best to do some testing first and then decide which settings are optimal for your business. I would recommend that you start by enabling both match options with Log Bot Activity enabled, and I would use the zero (0) second duration for the proximity pattern. This will add the pattern match data to your current logs without losing any data. Then you can evaluate the data or add the bot condition to the smart list of email performance and email link performance reports which can exclude undesired activity. (see image below)
After enough emails have been sent and evaluated, you can increase the duration time for the proximity pattern to see if it can effectively identify more bot activity. There will be a point when user actions begin to match the proximity pattern, and you can determine which duration is optimal for your business. Put a reminder on your calendar to check the settings again in a few months to confirm the settings. If you determine that your filter settings are appropriate and you prefer standard reports to more accurately reflect user data, you can choose to filer the bot activity.
I hope this was helpful!
- Blane
.png)