Business Intelligence and Social Media
Continuation of "Remaking Business Intelligence" by Customer Inter@action Solutions
Crossing the Channels
The growth in Web and IVR/speech recognition and ATM/kiosk self-service, mostly at the expense of live-agent and retail interactions has created a need for companies to also deploy analytics across these channels to understand customers' behavior. The information obtained is being integrated with that pulled from contact centers and retail via BI to see and predict customers' actions.
View a 2-minute demonstration
of InetSoft's easy, agile, and robust BI software.
Crossing those channels can be challenging though. By definition self-service does not capture the customers' tone, which imparts invaluable information as effectively as speech analytics on live agent calls. Also, the information often rests in different silos. That runs the risk of inaccuracy: multiple entries and storage increase the odds of data errors such as misspellings and incorrect addresses that annoy customers.
"So how can firms optimize the use of BI tools and minimize data quality issues given the multichannel environment and limited direct exposure to customers?" asked Vuk Trifkovic, senior analyst, Ovum. "I'd recommend: a) data quality initiatives; b) master-data management (MDM (Modular Digital Multitrack) An audio recorder that mixes and records multiple tracks of digital audio. The two major MDM technologies are ADAT and DTRS. See ADAT and DTRS. ) initiative; c) a knowledge-management layer; and d) unified approach to customer intelligence."
BI and Social Media
If employing BI to process information and gain insights from customer interactions in existing channels: contact centers, IVR and Web self-service and retail was challenging enough, then social media threatens to knock managers for a loop. Social media is unlike any other channel in that it is not one-on-one but a community, one where customers not the companies control the messaging. Customers can therefore make or break a product or offering at the speed of light by their comments.
Tapping into and managing information from social media is also different from other channels. These rely on and BI apps pull from structured data whereas social media data is unstructured and flowing between multiple parties. Firms cannot just use standard text-parsing tools because the terminology is different for different industries and verticals and regions, and social communities. They must create custom applications to understand the meaning of extracted keywords that they can then integrate them into the BI applications.
"Anyone can write a program that would take an e-mail or IM and chop it up for keywords," explains Evelson. "That is not enough in social channels. It requires a much more sophisticated semantic entity fraction; to understand which keywords are the subjects, objects and actions. You have to 'train' these applications to understand the taxonomies that could be unique to each industry or business domain."
To tap into social media, Evelson recommends that firms tighten down their existing BI processes i.e. walk before you can run. They then need to understand what their problems are in the social channel, and set up goals and objectives and metrics in response.
While firms, chiefly their marketing departments are beginning to listen to the social channel and in some cases use it for brand awareness and respond to comments and issues. Yet a much more difficult task is to actually adjust strategy, tactics, campaigns and product offerings based on insights derived from social media.
"Even if I glean some interesting insights how do I tie that to back to the cross-sell/upsell ratio and customer satisfaction; it is not a one-to-one relationship," says Evelson.
|#1 Ranking: Read how InetSoft was rated #1 for user adoption in G2's user survey-based index
Ovum's Trifkovic says that BI tools per se can readily manage social channels through natural language processing built-in, taking advantage of the text-based nature of interactions. Also, social networking can be graphed to understand who is saying what and who are the leaders and follower on social channels via the relatively well understood concept of network analysis.
The challenges lie in normalizing data across channels, establishing social media participants' identities-which are often nicknames--and how best to monitor these interactions track what is being said about companies. These also include how to effectively flow through and incorporate loose unstructured data from that channel. Also, the volume of data pulled in from social media consumes is large; it may require firms to bolster their processing capacity.
"The BI framework is there but the analytics are not quite there yet," explains Trifkovic. "There needs to be more network and semantic analysis, and perhaps more parallel processing parallel processing to handle the information. While enabling BI for the social channel will require frankly a little bit of vision, case study but in general I don't think it is going to be an insurmountable problem."