Larry Goldman makes some interesting points in his article on alerts - Red Alert! All Hands Battle Stations! - and it made me think about alerts in an EDM context.
In the article Larry asks this
Organizations should ask themselves the following questions to see if alerts can benefit them:
- What would I do if I knew a customer was trending downward?
- What would I do if I knew a customer was having a bad service experience?
- What would I do if a customer starting using our product/service differently?
If you can think of innovative strategies for any of these questions, you may be a candidate for trigger and alert technology.
If you are taking an EDM approach to decision-making, then you are likely to come across alerting in two areas - the responses to the questions Larry poses when the volume of these alerts is very high or where a real-time, as against quick, response is needed and the actual determination of whether these things are true, or not.
Let's take the first situation. If you have these kinds of alerts but have them in large volume, you probably cannot have someone call in each case. Perhaps an automate system makes more sense - write some rules about what to do, perhaps combine these with some predictive analytics that help guide the decision, and then embed that decision behind some kind of outbound engine - probably email or text messaging. That way when an alert happens, something is guaranteed to be done immediately, not just when some person happens to be available.
For example, say a customer starts sending more text messages than usual, perhaps your system should figure out (predict) how many they are going to send a month at that rate and then select a suitable plan (given the rules about the plans and their existing non-text message usage) and send them a text offering them a great price on it. All done without manual intervention and in a way likely to make the customer feel good ("We want you to be able to text without running up a big bill") but without harassing them with a phone call.
How about the second situation - figuring out when to actually trigger the alerts. Again, if your volumes are high then you don't want to rely on someone looking at BI reports and then sending the alerts. Indeed Larry correctly points out that many of these systems can generate reports flagging things only in certain circumstances. Few of them, however, offer the flexibility of an Enterprise Decision Management approach. In an EDM world I can build models to predict things about my customers, such as likely reduction in use of services, and combine these executable analytics with business rules about what makes sense as athreshold or what combination of data and predictions is significant. These automated decisions can then be used to trigger either manual or automated actions. In addition, the use of business rules technology allows for the rules to be changed. Perhaps I want to be able to increase or decrease my sensitivity to churn risk depending on my business priorities. Perhaps I want to change the various offers or actions that are being taken. All of this is easy to do with business rules.
Lastly we should not forget the "E" in EDM - Enterprise. Having built these models and written these rules to drive alerts, consistency of treatment of your customers can be guaranteed by embedding those same models and rules in the other systems with which your customers interact - the call center, the website, billing etc.
Alerts - another great use for EDM.
