Intelligent Alerts put the power of Anomaly Detection directly in your inbox or on your phone. Being immediately aware of significant changes in your key metrics and segments means no missed opportunities.
Hello, my name is John Bates. I’m product manager for Adobe Analytics. During this video, I’m going to introduce you to a new capability being released in the fall of 2016 called Intelligent Alerts. Now as a marketer, a marketing analyst, you need to know about the important changes that are occurring within the data. You also want to make sure that any communication that you’re receiving is focused on the most important changes.
Intelligent Alerts helps the software deduce problems. To illustrate, imagine that you’re looking at your products within any particular report suite. In this case, I can see that I have three products right here that are being ranked by purchases or orders. If I select these and then right click, I can create an alert from a selection. I want to be alerted whenever there is an important change from our top three products. Immediately you then land within the alert builder. I’m gonna go ahead and give this a name. Top products alerts. We can then specify the time granularity. How often you would like to receive these alerts and how frequently they are being checked. This can be at hourly, daily, weekly, or monthly. You can then specify also the recipients. So in this case, I’ll put myself. And then you can put distribution lists, as well as mobile phone numbers beginning with the country code. So I’m gonna go ahead and put my number here.
You’ll also notice that we automatically apply an expiration date, which is one year from the date in which the alert was created. We will then notify you a few times before it actually expires to remind you that the alert is set to expire so that you can come back into the alert manager and renew that particular alert. As we move further down within the alert builder, you’ll notice that we have a number of triggers. Send an alert when any of these metrics are actually triggered. In this case, orders, when there’s an anomaly that exceeds the 95% threshold.
Now, alert triggers do not have to be based upon anomaly detection. You could also change this to other rules based approaches, like is above or equal, also is below or equal, or also changes by a percentage. You’ll notice here that anomaly detection can then also be applied, not just to daily time granularity, but hourly, monthly, and weekly.
In this case, we’re looking at orders when there’s an anomaly at the 95% threshold and the product equals Wasatch Xtreme Parka. Or, when there’s an anomaly in orders for the Bonneville shorts swimsuit product, or when there’s an anomaly in orders for the Timpanogos scarf products. I can also apply segments. So if I wanna look at orders for the Wasatch Xtreme parka and I want them to be mobile customers, I can then apply that segment right here. I can also bring in over total metrics. If I wanna bring in all revenue, as well, for this particular product and mobile customers, I can do that right here, as well. You’ll notice an additional capability here. The alert preview. So often times, you wanna know how frequently to expect any particular alert. You don’t want yourself or others that are set as recipients to be sure if they’re getting spammed or receiving too much communication. What we’d actually do, the system automatically goes back historically and models out this alert recipe as you’re building it out to let you know how frequently you should expect this alert to actually notify you and the other recipients. In this case, we can see that six times over the last 30 days this alert would’ve been triggered. As you dig down deeper, you’ll notice that for each trigger we actually identify which triggers are contributing to these six alerts that would’ve been triggered over the last 30 days. And let’s say you wanna actually adjust this and lower it just a little bit more. To do that, I can just come down here and you’ll notice you have four times that this alert rule for this product right here, the Bonneville short swimsuit, that this would’ve triggered a communication. If I switch my threshold to 99%, now this would’ve occurred less often. And overall, this entire alert recipe would’ve only triggered two times over the last 30 days. If I’m happy with the way I created this alert, I can then click save and now my alert is live. Once you receive a communication, it will come via SMS text, as well as email, you can create an alert, as well, from the components right here. I can just say create new alert and it’ll take you right into the uplink alert builder and you can begin to design how you would like the triggers that you would like to apply to an alert. You can also manage your alerts by coming up to components, down to alerts, and here you’ll be able to see all of the different alerts that you created. You can see who the owner is, tags applied, when it was last modified. You can also select alerts and copy them and change their definitions based on these copies. So on and so forth.
I hope you’ve enjoyed this training on Intelligent Alerts. We appreciate your time.
For more information on this feature, visit the documentation.