I recently came across this external blog post by Courtney Grimes and thought it would be useful to share with the community (great post, btw, Courtney). Especially since there has been a lot of discussion/questions recently on false-positives of email opens and clicks:
I did want to comment on the guidance to combat this:
After testing this idea on dozens of different Marketo instances, the single most effective way I’ve found to track email clicks is a combination of two filters: Email was delivered + Visited web page: [web page linked to in email]
Typically, one would simply add these two filters to the smart list of the trigger campaign. When we tried that in the past, it didn't work for us since Marketo doesn't log this activity immediately as it happens - at least not in our case (especially the "visited web page" activity). We've seen up to a 45 minute delay. So we used an alternate approach: we included a "visited web page" choice in our "change program status" flow step. And prior to that flow step, we also included a 45 minute wait step. The only issue with this approach is if the user ever visited this page in the past, the choice will still be true. In other words, the choice step is not directly tied to this instance of the email click.
As Courtney states in her wrap-up, there really is no fool-proof solution. But if you had to pick one, this is probably the best approach.
Another solution that we've tried before is adding a small 1x1 white image pixel into our email templates that links to a landing page. The theory being that a real human will not see the 1x1 pixel, so all of the clicks on that pixel that go through to that landing page are bots.
Then in our smart lists, we add another filter that says "did not visit web page: "enter the bot landing page name" as an additional way to filter out the bot traffic.
Hi Courtney - there are two issues with this approach:
Dan this is exactly what we've done since realizing false clicks were such a huge problem, especially when counting them as a response. It cuts out a lot of the headtrash with deceiving email metrics. I'm pretty happy with the results of setting up smart campaigns in that way. Thanks for sharing!
Hi Chris - yeah, the chances of someone clicking the link in the email and having visited the landing page in the past is probably pretty small. So you should be OK in that regard. The other occasional issue I would sometimes come across is the dreaded red squiggly in the value of the landing page. Some of our users were simply copying/pasting in the URL value. But just like the "visit web page" filters/triggers, the http/https protocol is not included here. And since Marketo didn't throw an immediate warning (and allowed the campaign to be activated), it wasn't until a lead tried to run through this did we get notified that there was an issue with the campaign.
One other bit of advice is to also include a "remove from flow" flow step - especially if you have additional processing that takes place in the smart campaign (e.g., setting last touch channel, date, etc. values) - for those that don't qualify for the "visited landing page" choice prior.
Put your "visited web page" as filters in a separate smart list, with an "in past 46 minutes" constraint, then in your choice, replace them by a if "member of smart list" choice, and that will remove the false positive due to older visits.
This is a great catch Grégoire Michel!
I was blanking on how I could only include "new" visitors. Thankfully I am the only user in my company doing the back-end production but was totally blanking on how to differentiate old visitors to visitors who should continue in the flow.
Would you recommend adding the clicks link in email into that smart list as well? I want to try and template this smart list for future campaigns.
No need to add the "click link" since they already in the smart campaign trigger.
with this approach, there will always be the manual step of adding the email link URLs to the false positive detecting smartlist, correct? feels like a dumb question, but trying to write a process so need explicit steps for executing.
Dan, are you writing this because you've seen an uptick in this behavior recently? If so, you're not alone! I'm also looking for a method to split the difference between too much marketing suspend and too much false-positive clicking. Let me know what you find out! 🙂