How to use big data to stop customer churn
Your customer service representative answers a call from an irate customer. "This darn thing I bought just doesn't work!" he exclaims. "I've tried and tried to get help from your service folks, but they're always late and they can't fix it either. I've had it with you guys. I want my money back!"
There's silence as your rep calmly listens to this obviously unhappy customer--and pulls up a raft of information about him, ranging from a few years' worth of transaction data (from the data warehouse) and service call information (from the service department databases) to call history (from the CRM system) and what he's said about your company on Twitter, Facebook and the blogosphere. There may also be a stream from previous online chats or, thanks to cookies, a list of where he's been when searching your website.
All this information is compiled so the rep can see, through a visualization tool, that this is actually a good customer who's just having a bad day: He hasn't been troublesome in the past, he frequently Tweets and therefore has a high Klout score (which makes him a social media influencer, presumably with lots of followers), he gave you a Facebook "like" and he spends a fair amount of money with you.
"This is a significant advancement for organizations that, until now, had to rely on customers' frankness and candor to understand the issues."
This gives the rep the green light to offer this customer a refund, a free return shipping label and a coupon for 20% off his next purchase. The customer is happy&hmdash;and, even better, he's decided you aren't so bad after all. Case closed.
Big Data Holds Big Promise for Improving Customer Experience
Customer service reps red light, green light dashboard view quite yet, but for companies on the bleeding edge of big data analytics, the scenario described above is happening today, says Eric de Roos, senior director of product management for business intelligence vendor MicroStrategy.
"Big Data gives you a more in-depth understanding of what people are doing and how they are engaging with the organization. Ten to 15 years ago companies were just storing transactional data," he says. "Now we are tracking more behaviors. We give people logins, we store cookies. When they come back, we know who the customer is...what pages they click on and what they're looking for."
That data, combined with transactional information, gives companies a good picture of an individual customer's value, de Roos says. Add sentiment analysis from the IBM Infosphere Streams Twitter API and the MicroStrategy Wisdom engine, which tracks the 15 million "Likes" of some 65,000 Facebook users and their friends, and you start to get a much clearer picture of what that customer thinks of you and, just as importantly, your competition.
Customers Tweeting, blogging and liking things on Facebook simultaneously accumulate an "influence score" and engage with your brand, says Wilson Raj, global customer intelligence director for SAS. These "digital traces," in turn, can be harnessed to get a full view of that customer.
Since this level of customer experience analysis remains in its early days, only the most adventurous clients combining Twitter feeds with YouTube headers, aggregated sentiment analysis from Web-generated data sets and the blogosphere and wrapping it all in a natural language processing (NLP) engine in order to deem a customer happy or sad. As these technologies mature and corporate IT departments find the time, talent and resources, they'll catch up with this trend.
Analysis: Google's BigQuery Offers Infrastructure to Crunch Big Data
They'll have to, says Rita Sallam, Gartner's BI analyst and research vice president. "Advanced analytics must be more pervasive to deliver significant value and competitive advantage to an organization," Sallam wrote in her February report, Advanced Analytics: Predictive, Collaborative and Pervasive. "To date, use of tools and processes for building advanced analytic applications and deriving and consuming insights have been limited to a small number of highly trained and experienced statisticians, analysts and operations research professionals."
Moving these tools into the hands of line-of-business users may not be as hard as you would imagine, says IBM's Director of Emerging Technologies David Barnes. "The cost of getting started with this isn't that great. That's the cool part of these massively scalable systems. I can start on my MacBook, decide I like it, take that exact same code and spread it across 100 servers."
T-Mobile Using Big Data to Understand and Predict Network
You can also use big data on the back-end to make sure your offerings are what you say they are. T-Mobile, for example, continuously analyzes 2 PB of network performance data on its IBM Netezza data warehouse, loading nearly 20 billion rows and processing nearly 150,000 ELT jobs daily. In a five-minute span, says Christine Twiford, who works in T-Mobile's network engineering department, as many as 60 users may be executing load and query operations on the system in near-real time.
The main focus of network engineering is optimal network performance in the name of customer experience, she says. "We use clickstream data to calculate the speed of downloaded songs. This gives us a great proxy for understanding throughput speeds at the tower level, as well as handset speeds."
With this data at the ready, T-Mobile has been able to predict--and prevent--network outages that could have been caused by faulty Android applications, Twiford says. (After all, a smartphone user without network access might make a customer service call similar to the one described at the beginning of this article.)
Twiford and her team also use handset data to determine whether they may need to open the floodgates if, for example, they anticipate a large volume of video suddenly hitting a given geographic area--Boston on the Fourth of July, Times Square on New Years' Eve and so on. This is exactly the opposite of what happened at some Olympic venues, where usage overwhelmed the networks set up to serve both event organizers and fans.
These streams can be used to get in front of an emerging trend--or a presidential candidate.
Say a coffee chain using Wisdom to better understand its customers discovered they have an "affinity" (to use MicroStrategy's parlance) for a new type of chocolate hitting the market. If the company wanted to make a quick decision based on this preliminary but compelling insight, it could stock that chocolate in the stores where Facebook "likes" are highest, notes Warren Getler, MicroStrategy's vice president of Corporate Communications.
Journalists on the campaign trail have used Wisdom to track down Mitt Romney supporters at their favorite restaurants. How? If they "like" Mitt and they "like" Joe's Diner, reporters know they can stake out Joe's and get some interviews.
Big Data Will Make Big Strides
Big data, while intriguing, remains in its early days. Big data doesn't come easily--or, for that matter, cheaply, as Wisdom Pro starts as $25,000. Companies working to integrate the many silos of internal data with intelligence from the Web and sentiment analysis from NLP engines, then, are riding the curl of this wave.
Such firms are in the minority. Fewer than 1% of companies are currently using big data this way, says independent customer strategist Esteban Kolsky. However, that will change, he adds.
"Big data provides social data and other publicly available data that can be analyzed and used to understand the customers' sentiment and needs before they become issues or problems that lead to churn," Kolsky says. "This is a significant advancement for organizations that, until now, had to rely on customers' frankness and candor to understand the issues."