Data Management and Predicting Customer Behavior

You store tons of data about your customers. You know their shopping habits, their preferred method of receiving marketing materials, and how many times they've called with a complaint. But how deeply do you look into that information to predict what they'll do in the future? A new Teradata study, "Customer Valuation Gap Survey," conducted by BuzzBack Market Research, shows that only 56 percent of 192 responding senior executives at companies with revenues above $500 million can easily analyze such data to understand the value of their customer base, what their customers value, or what they will want next. About 45 percent can predict when a customer is about to take his business elsewhere, according to the survey. Sam Gragg, Teradata's assistant vice president of marketing and customer management solutions and the survey's cocreator, says that at first he was surprised to learn executives didn't want to have more information. Despite the wealth of available analytical technology, companies say they have too much data and don't know what to do with it, according to Gragg. The questions a company needs to ask are: Is the data accessible? Is it of sufficient quality? And, how can tools be applied to it? Guy Creese, managing principal at Ballardvale Research, says predictive analytics is going through an evolutionary process and "still won't happen quite as fast as some vendors hope." In the past, according to Creese, companies just wanted to get their products out the door. Then, they started storing data and looking at it over a week, a month, or a season. They started getting up to speed on general queries like how much sales increased from one quarter to the next, and then started storing more and more information. Creese remembers a time when employees debated what data they were willing to part with if they wanted to store a large file. Now, "it's cheaper to save everything than...to have the meeting about what to toss out," he says. "If you were hungry for information, now you're drinking from the firehose." Companies must cull what's important. Budget limitations, lack of continuous agent training, and data repetition are three obstacles to predicting customer behavior. Survey respondents cited budget limitations as the primary reason for not doing more with data. Lack of analytical skills among front-line agents, who need to learn how to delve deeper into understanding the data in front of them, also hinders growth, but, Teradata's Gragg says that "more people are training their own people, or at least have the desire to." "A lot of companies have multiple copies of the same data," Gragg says. "It's a schizophrenic view of the business. Nobody's wrong, but who's coming up with the right answers?" By using the same data throughout a company, Gragg says, executives can decide which customers are worth investing in early to turn them into valuable, profitable customers by "encouraging them to interact with you in a way that's beneficial to both of you." Another topic addressed in the Teradata study is how to predict churn. According to the survey, 86 percent of executives listed knowing what a customer will likely want next and when they are close to, or likely to, defect as being important for them to identity and easily access. By contrast, only 52 percent and 45 percent, respectively, of respondents said they actually could accomplish those tasks. One executive responded that what would help most would be predicting "the likelihood that a customer might churn, and the types of complimentary products that can be bundled to help retain that customer." According to Gragg, there are three levels to customer loyalty: good, disloyal, and defected. Although it's easy to determine a defected customer simply by the fact that purchasing has ceased, the disloyal ones who are already declining in loyalty are the ones who need to be watched. "If companies can predict that behavior, they can intervene and maybe they can save the customer, maybe not," Gragg says. Still, "companies have to crawl before they can walk," Ballardvale's Creese says. "We get fascinated by the technology and forget people are creatures of habit, and don't want to change too quickly. You can have this splendiferous tool that does all kinds of great things, but if you don't have buy-in from high execs, [you won't move forward]." Gragg agrees that the concept of analyzing data is "a moving target," with some companies way ahead of the curve and others constantly struggling to catch up. Each company needs to progress at its own pace. "It boils down to everyone [having] too much information. It's all about organizing it and making it accessible, then being able to use it in an intelligent way. It's a maturing process." Related articles: Along Data's Journey Into Might Transforming customer information into actionable insight. Prescriptive Analytics Delivers Right-Time Information Eliminate surprises, identify opportunities, and maximize operational effectiveness by having the right data at the right time. Don't Manage Relationships, Optimize Them One CRM industry executive's opinion is that now, the time has come for CRM to evolve through customer relationship optimization.