The CIA admits to monitoring Twitter, Facebook, and blogs. This shouldn't surprise anyone, but the tools they’re using might be enough to get you to rethink posting anything personal or ever checking in...
What Higher Education Teaches us about Data-Driven Customer Retention
The U.S. higher education system has a problem: we need college graduates but students are dropping out at increasing rates. Over a third of all students drop out before completing their degrees and only half complete a 4 year bachelor's. In other words, colleges are failing to keep their students, and educators are scrambling for answers. Some forward-thinking institutions are asking, “can data analytics help?” Rio Salado Community College thinks so, and their innovative solutions have applications that reach beyond education.
This topic has far-reaching implications because it touches on one of the holy grails of business: customer retention. It's well known that it costs about five times as much to get a new customer (customer acquisition) as it does to retain existing customers (customer acquisition), so this is a hot-button across just about every industry.
For any college, losing students is a problem, but for Arizona's Rio Salado it’s even worse. Rio Salado is primarily an online university which means that dropout barriers are minimal, and their target market is high risk students who are mostly poor and first-generation U.S. citizens. It’s a risky plan, but by using data driven solutions Rio Salado is beating the odds and juicing their bottom line while helping students better their lives. In many cases it's just a matter of giving students support at the right times so they don't get discouraged and give up.
If you’re a student at Rio Salado, all your data is being recorded; when you log on, when you log off, what you view (or don’t), how long you're on the site, and so on. These metrics are used to build a data-based profile about you and determine the probability you’ll pass your class and graduate.
"In some ways, it's kind of to be expected if you look at online courses as a parallel to in-person courses," said dean of Instructional Design Michael Cottam. "If students show up, participate and do pretty well on the assignments, they'll be successful. But the difference is that now we have data that's tracked every day on what a student does in a course.”
8 days of data is all it takes to predict the likelihood a student will successfully complete their class. After 8 days, students are divided into three risk levels and monitored for changes from that point on. Armed with this data, the faculty can focus their attention on helping those at risk of failing. Some of this is automated, text messages about unfinished assignments and discussions they haven’t yet contributed to, but the personal touch is heavily relied upon to pull those students back into the class.
“We wanted to be able to identify at-risk students early because the earlier in a course you can predict whether a student is going to be at risk, the more time you have to put interventions in place--support structures, contacts, and so on that could help mitigate the risk.”
This predictive analytics solution is still in the testing phase, but so far Rio Salado has raised their retention rate for online classes to 68% a full 18% higher than the national average.
Interest in data mining and predictive analytics is a growing focus of higher education, and Rio Salado is just one example. Their student retention parallels what many other companies are trying to do with customer retention in many other industries.
Customer Retention and Segmentation
Return customers are the bread and butter for all businesses: the Pareto Principle applied here says that 80% of your sales come from 20% of your clients. The goal of customer retention analytics is to make sure you do whatever you can to make that 20% happy and give them what they need to succeed. Enable your best customers. At the same time, it's very probable that 20% of your customers are eating 80% of your costs, which means you'll want to make sure that it's worth it to spend your limited resources making those customers happy.
Step Zero is customer segmentation.
Rio Salado is currently optimizing their student retention through extensive and focused testing, and they’re finding some truly telling results. Because they’re a college, they automatically have some data other businesses have to work to discover: their customers come segmented and customer goals are explicit.
Each new student enrolls with a major selected, which designates the segment they belong to within the colleges offerings. This segmentation is valuable because it allows Rio Salado to treat its different segments uniquely. For example, Rio Salado tested an interaction method in one of their Psychology classes that worked marvelously and raised retention and completion. However, when they tried the same method in a math class, it had the opposite effect. With proper segmentation in place, Rio Salado is able to identify what methods work best for their different groups, put action plans in place to retain their students, and surpass their competition.
Customer goals dictate how they interact with a business and what they want from it. For Colleges it’s simple; students want to graduate. However, for companies offering services it’s more complicated. Your customers' goals will vary based on their industry, their customers, and their expectations. Knowing their goals allows companies to anticipate customer needs and move to satisfy them.
For most companies customer segmentation is much harder to discover, but the value is much the same. Once you know what segment a customer belongs to, you can tailor your interaction with them in order to provide the best possible experience. If you're lucky you can segment your customers easily, today. But if you're not this can be tricky, because most software packages in use today don't support the algorithms you need to identify these segments. Behind the scenes you'll need to be able to implement clustering algorithms, such as the popular k-means clustering algorithm. (Although there are others that are pouplar as well.)
If you’re interested in learning more about how this type of data mining is done, we’re putting the finishing touches on a new report on this very topic: Applied Data Labs Customer Retention Bluepritn. If you’d like a free copy when it’s released shortly, leave your email address below and we’ll send you a free copy.
For more data and analytics-related articles like this one, you should follow us on Twitter.
Or provide your email address below for free exclusive reports and commentary.