Expand my Community achievements bar.

Announcement: Calling all learners and mentors! Applications are now open for the Adobe Analytics 2024 Mentorship Program! Come learn from the best to prepare for an official certification in Adobe Analytics.

Anomaly Detection Alert


Level 5


Can we get an Email when Anomalies are detected!


Has helped in so many ways, but having to makes it less effective.







For anyone who voted on this issue, can you add a comment on this thread about how this would impact your business (and your ability to drive your business using Adobe Analytics)? The use cases will help us with our research in this area. Your help is very much appreciated. Thanks!


Level 1


The way I see it, there are two core benefits to this feature:


  1. Operational - when something breaks either on your site or in the implemented analytics code, many times it will cause an anomaly in the data somewhere. As an example, one of my site's primary success events dropped ~21% in a single day due to an implementation issue. Unfortunately, dashboards and reports were not so granular to highlight the drastic loss in data and seasonal cycles hid a lot of the impact for weeks on end. Having an e-mail alert for this anomaly would have helped us identify the fact that there was problem faster. Addressing the problem sooner would have helped us have a more complete data set. Now we have a huge ugly calendar event for many weeks to indicate the drop. There's the added benefit of having higher overall confidence in the data if we can readily identify and diagnose any anomolous metrics.
  2. ROI - my organization's business objects revolve around content engagement and managing costs. Alerts for anomalies arising out of specific campaigns can help identify opportunities to optimize spend. We have too many campaigns to do this analysis manually. On the other side, drops in content engagement due to a shake up in publishing frequency could help us quickly react and retain our audience.


Level 4


Hey Ben, 


The primary use case would be to catch a broken metric that makes it to prod as quickly as possible, so that it can be fixed right away.


In an ideal world, this would never happen, but in reality, I would bet that every analyst has experienced it at least once. This would be the saftey net.


For example, our share click tracking is quite sensitive--I would set up analomaly detection for share rate with a bunch of segments (browser, content type, device) and have SiteCat notify me if share rate dipped unexpectely.


The alternative is to do this with Observe Point or in-house automated testing scripts that run frequently, but this would be even better.





Level 5


this could save a ton of investment in reporting anomaly detection development as the key thing missing in Adobes is the ability to notify us. this makes adobe a more serious player in our QA operations if the alert functionality was there.