How Big Data is Being Used to Stop Customer Churn
Companies have been working for years to get the much vaunted "single view" of their customers. Many businesses have made great progress in this area by breaking down silos of information and sharing them among the different departments and people that deal directly with customers face-to-face.
Today, however, “big data” (which some define as “any amount of data that's too big to be handled by one computer”) is adding another layer to the information cake. Most firms have plenty of data on their customers: how long they've been purchasing from the company, what they bought, who they are, age, geographic data, how much they've spent, etc. Larger businesses, with tools to help them, often also calculate customer scores like lifetime value. But what they don’t know much about is what their customers are thinking and feeling about them.
This is where Big Data is starting to make an impact. A whole World Wide Web of publicly available information is just waiting for savvy companies to include into their decision-making and customer retention processes. And companies large and small, from business intelligence (BI) vendor SAS to startups like Gnip, are just waiting to help them take advantage of it.
"Most times customers leave for multiple reasons, not just one," says Esteban Kolsky, a customer experience consultant and founder of Thinkjar.com. "Even if [the customer] can quote a last-time interaction or connection, they often accumulate poor experiences to make their decision.”
Instead of responding to anecdotal (and sometimes misleading) information, Kolsky says, “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. It can also be used to understand the reasons for past churn by looking for patterns and abstracting the information to reasons they are churn issues."
It is this use of how the customer feels verses what the customer did that is what makes these new insights, if used properly, such powerful stuff.
The not-so-distant future
Imagine this scenario: An irate customer calls into your contact center. Let’s pretend that upset customer is, say, you. Even though I’m sure you have never been upset by anything a company has done. Let’s just pretend.
The customer service rep (often called a customer experience rep or CSR these days) gets the call. On his status screen, which knows your phone number from purchase records, the CSR immediately sees your existing personal and transaction data, as well as your Klout score (a service that tracks people's online activity and then gives them a "influencer" score based on that activity; the higher your score the more people potentially listen to you); how many tweets you've sent out about the company (and its competitors); the sentiment around those tweets (net negative or net positive, as judged by a natural language processing engine that quantifies such things); where you've been on the company's website and what you did while you were there; your online chat interaction transcripts; and anything else the IT guys and CMO can think of that they'd like considered (think YouTube headers – seriously) while the CSR is working to resolve (or upsell) you over the phone.
Now, all of this information doesn’t appear in raw form. That would be way too much to deal with and take way too long to wade through while the CSR is also trying to sooth someone's bad day. Instead, much of that information (particularly the Web-based stuff) has been normalized into something like a "Customer Churn" indicator that tells the CSR that this customer is on the verge of defecting to the competition.
But, wait a minute, this is a valued and, more importantly, a valuable customer. The company doesn't want to lose her. So based on the churn indicator "score" the system kicks out three next best actions (NBAs) or next best offers (NBOs) the CSR can use to placate the customer. And, Voilà! the company keeps the customer; she leaves happy.
"In the case of churn," says Rita Sallam, a Gartner BI analyst and research vice president, "currently most organizations that calculate propensity to churn use structured transaction data, perhaps supplemented by demographic data that they can purchase.”
She adds: “However, what if you could also include sentiment from call interactions (analytics of voice and text) and sentiment and influence derived from social data in the churn model?"
How much is that worth to your company? How many times has this not happened? Just read the trades: Way too often, at way too many companies. The idea is not only to push decision making as close to the customer as possible – not a new idea, of course – but to also put context around the decision.
So, in the example above, if you know that the customer is of average means, a company rep doesn't want to offer him a coupon for a very expensive item he would never buy just because that’s the promotion Marketing is pushing right now. No: You want to offer him something based, say, on a recent round of purchases that may indicate the customer started, say, a big landscaping or painting project. The point is to give the customer something he actually needs versus something the company wants out of the warehouse.
This is where start-ups like Gnip come in. They, along with a few other providers just getting into the business, provide social media metrics firms with a fire hose of data that businesses can use to figure out things like customer sentiment about your brand, or what's trending on Twitter and other social media sites (including wikis, forums and blogs).
"Our customers, most frequently big social media monitoring companies, take the data and use it to create the dashboards and monitoring tools," says Elaine Ellis, Gnip’s marketing manager. "We don't do analytics. What we do offer is data enrichments, so customers can filter the data based on language, Klout score, number of followers, etc."
I remember a few years back, prior to the 2008 crash, when things were booming for the credit card issuers. One of my credit cards raised my rate without telling me (they said they sent me a letter but I didn't see it). I wasn't too happy about it given my stellar credit rating (no joke, it really is – being a journalist means being frugal I'm afraid!). When I called them and told them I was going to cancel the card and move to another company, the CSR's response was, "Go ahead." Seriously. So I did.
Today, that same company, without us even having to ask, offered to refinance our mortgage this past spring at a lower rate while shaving a few years off for good measure because we were "valued customers." My, how times have changed. My guess would be, outside of being extremely smart, the financial company may have merged a few databases on the back end and cross referenced those with, say, credit score data to find the folks that actually pay their mortgages. (What a rare breed we've become.)
The Wine Guy
But I digress.
Let's say you’re a wine distributor. A certain type of wine is a big hit right now in the blogosphere; it's getting tweeted about, and Facebook "Likes" are on the rise. Oh, and that's right, there's a wine and jazz festival happening next weekend.
How would you like it if your field sales reps, for example, could advise the wine stores to stock up on that wine? "How about we ship over a few cases, Mr. Wine Guy, so when festival-goers come looking, you have what they want?"
In this case, it's the Wine Guy that you're interested in keeping happy and prosperous because, obviously, the better he does, the better you do. He sells more wine, you sell more wine. But now he's even happier with you because you gave him a great tip that he wouldn't know about otherwise, and he potentially might have missed out on a lot of business. Smells like customer loyalty to me. Win-win, right?
This scenario is unfolding today. Right now, companies with the marketing and social media savvy are investing in just this sort of sentiment analysis and data processing power via predictive analytics engines powered by Hadoop, map/reduce, Hive, Pig, and R (to name a few of the open source tools being employed) to help the marketing and customer service teams do just what I've described.
This is a big step toward turning all that information you have about your customers into knowledge that can grow the business and create more loyal customers.
"You can do real time things that are not valuable to the customer," says SAS's Wilson Raj, the company's global customer intelligence director. "At the end of the day, it's all about personalization and relevance. What we are talking about here with all this data is to make sure that, through every interaction, the brand and the company is doing its best to be as personal and relevant as possible at the point of immediate need."
See also:
- Subscribe to RSS Feed
- Mark as New
- Mark as Read
- Bookmark
- Subscribe
- Email to a Friend
- Printer Friendly Page
- Report Inappropriate Content
