Recently my colleague Mark Smith wrote about Splunk and its latest technology supporting analytics for IT on machine data and providing operational intelligence. I wasn’t familiar with the company, which has focused on IT users and improving the performance of a company’s networks and IT systems. From a customer management perspective, these are of little interest unless they impact the customer experience; for example, if the website is down or the online banking system is unavailable. But in a follow-up briefing I learned that Splunk is increasingly playing in the business analytics market and has several user cases that relate directly to customers.
I was intrigued by Splunk’s interesting use of the term “machine data.” I am a self-confessed technophobe, so I thought all data was machine-generated. Not so, according to Splunk, which can process data into three areas including:
Splunk focuses on processing the last two types and integrates with the first to extract contextual data; for example, it can use an email address to look up additional customer data in a CRM system. When I heard this bells began ringing in my head, because I have written that a 360-degree view of the customer should be more extensive than most people think. The common perception is that it should include all available transaction data concerning customers (in Splunk’s terminology, business application data), but I think it should also include all available information about customer interactions (human-generated data) and any event data (machine data) that might impact the customer experience – for example, whether a cell within a mobile network is down, preventing customers from making calls, which in turn is likely to result in lots of calls to the contact center. I also have observed that while some vendors are getting close to producing a 360-degree view, no one is there yet.
Splunk has an advantage here, because machine data is structured and often unstructured, and it has developed techniques that allow it to identify specific locations within a record as a defined field; for example, starting at location 100 in the script captured as a customer completes an online purchase, the next 20 characters represent the code of the product purchased. These definitions can be created automatically by analyzing multiple records, or users can define fields using tools within the product. These fields can be used to index records and thus to search for specific records to include in the analysis. Splunk’s product can also use fields as indexes to information from records in a business application; for example, a cell phone number could be used to look up the corresponding record in a CRM system to obtain additional customer information that can be included in the analysis. Analysis can also include data from multiple types of transactions, such as email, call recordings and Web scripts, allowing users to see, for example, all the different types of interactions a customer has had with a company – a technique that is often called cross-channel analytics. The analysis occurs in real time, and so could be used, for example, to show a contact center agent all the previous interactions a customer has made just prior to the current call.
Splunk has user cases showing use of its product to track customer downloads from a website, the use of mobile messaging and the steps customers take to complete an online purchase. Although Splunk is not a well-recognized vendor in the customer analytics space, such applications show that it has the ability to combine business, interaction and machine (event) data to provide operational intelligence. In this context, that allows companies to better understand customer behavior and thus to proactively improve the customer experience.
Regards
Richard Snow – VP & Research Director