Customer Service Analytics: Adding Intelligence to the Service Experience
Posted by Jeffrey Henning on Fri, Sep 12, 2008
At the Gartner CRM Summit 2008, Gartner analyst Gareth Herschel discussed Customer Service Analytics: Adding Intelligence to the Service Experience. He outlined four best practices:
- Define scope of analysis: Is your organization analyzing the service process in its entirety or the IVR/call-routing system or the call center staff?
- "Good, inexpensive call" may be an oxymoron: A call that is too long costs more and annoys the customer; a call that is too short may be missing an opportunity for engagement with the customer and may not resolve the issue at hand.
- One size doesn't fit all: Don't judge all calls by the same criteria. Categorize calls by the purpose of the call and the value of the customer, and then develop metrics for this segmentation.
- Tailor results to different views: Too often analytical reports don't really meet the needs of anyone in the organization. Some reports need to focus on "exceptions not averages", and should include alerts and triggers so that the company can take action to intervene on behalf of dissatisfied customers; some reports need to focus on the big picture, providing "context not summaries" and should try for advanced visualization. Too often dashboards have become just a collection of metrics; the best practice for dashboards is to compose them of "related metrics that paint a whole picture when presented adjacent to one another."
Gareth made the point that companies should pay special attention to "bad news". He quoted Despair.com's service department: "We're not satisfied until you're not satisfied." He said, while it was tongue-in-cheek for Despair.com (a publisher of "demotivational" posters), it was painfully true of many firms, who - too often - make providing negative feedback an arduous process, "putting the complaining customer into a penalty box". Negative feedback can often lead to the best insights.
For capturing market insights from service interactions, Gareth displayed the following quadrant analysis:
|
Mature Analysis Strategies & Techniques |
Immature Analytical Expectations & Best Practices |
| Data We Need to Collect |
EFM |
Blogs |
| Data We Already Collect |
Enterprise Data Warehouses |
Customer Call Recordings |
Blog and forum feedback can be "unreliable, prone to manipulating, and unquantified". Gareth suggested using it to identify potential service issues that were then quantified using enterprise feedback management.
Analysis of customer-service issues can be applied in three different ways:
| Concept |
Tactic |
Role of Analysis |
Analytical Techniques |
| Pre-emptive: Customer Issue Avoided |
Identify & Resolve Causes of Problems |
Identification of High Cost Issues |
Cost Allocation
Call Categorization
Root-Cause Analysis |
| Pro-active: Customer Notified Issue is Being Addressed |
Resolve Issue on Corporate Schedule |
Identify Treatment Strategy |
Issue Detection |
| Re-active: Customer Calls to Resolve Issue |
Divert Issue to Self-Service Channels |
Divert Issue to Self-Service Channels |
Intelligent Call Routing |
Gareth concluded with a complex slide showing categories of service analytics vendors. He pointed out that no suite solutions exist yet. Over time, different technologies will become bundled together and consolidation will happen. He said that enterprise feedback management and text analysis were a natural fit, and he expects more bundling of such solutions in the future (Clarabridge and Vovici have been doing just that since July). He doubted that 10 years from now the dozens of technologies he showed would be bundled into a standard customer-service analytics suite. He said, if it did happen, there would be another 20 new categories of analysis that would have emerged and would not be bundled in the suite!