Wednesday, October 27, 2010

Data Mining, Customer Service and More Success

Imagine you owned a gold mine with large nuggets lining the walls of the mine. But all you had to dig the nuggets out were your fingers. Yes, you might be able to scratch a few nuggets out but then you found you did not have a smelter to process the gold. Well, you could ask the college to help. You’d set up task forces. get committees going and some faculty would write papers for you on the history of gold smelters in the Gold Rush; the chemical properties of gold, gold versus human capital as an economic strength and of course gold as a metaphor for oppression in the poetry of the Victorian period. The committees and task force would produce why the college is not mining the gold and a recommendation that the university look into setting up another task force to evaluate gold mining tools and a request for release time in their reports that would be filed within a year or two.  

The reality is that every university, college, career and community college has just such a mine. A mine of great value and future wealth. A mine that could increase both enrollment and retention. A mine lined in millions of bites of data just sitting there waiting for you to use it. And unlike a gold mine, this data mine keeps increasing its raw materials every day, every hour, and most every minute as the school collects data on itself and its student body collectively and individually. And all a school needs to do is mine it. And there are tools out there just waiting to be used to bring the gold to the surface and then transform it into very useful material that will bring immense value to the school.

The tools fall into two categories of data miners and customer relationship management and the best of the tools do both to create usable information that will increase enrollment and retention.

Data Mining
Data mining at its simplest is using computer programs to create data and extract patterns from the data collected. For example, when a school enters al the information about students from their applications into an MIS system like Datatel or Banner, that is the start of creating the mine from which data can be collected. But like a gold mine there are not just nuggets of the valuable stuff, there is plenty of dirt, stones and even iron pyrite in among the gold data.

In a college data mine, there are millions of terabytes of information on the students, the institution and everything else entered into the MIS system. The valuable data needs to be sorted out from the piles of unneeded distracting data sets and terabytes of numbers and unwanted data.  For example, when a school wants to learn about the interests of incoming freshmen so they can try to focus on their interests during orientation, someone or something needs to sort through every bit of information gathered on student interests to be able to create a pattern that can inform the design of orientation. This can be done by hand of course and some schools are still doping that. Manually going through piles of paper to pull out information from the question “What do I do for fun?” Some schools may have put the question on a survey with five choices. Then the system can be asked to spit out the results of the survey but of course that limits students to pre-determined interests. The survey is a simple form of data mining but a limited one.

A real data mining system would be able to analyze every bit of information students might have entered into the system and then break it into patterns of interest say by gender, age, town of origin, major, minor, how many will be taking English 101 on Tuesday and Thursday, and so forth. Data mining pulls out all the gold and even sorts it into categories and patterns.

It has great uses in higher education in many areas such as admissions and retention. In admissions a person could use data mining to study application and admission yields from different geographical areas.  It could see that some towns or cities provide greater yields than others indicating that perhaps more time should be spent working that town to increase enrollments; or inversely greater effort needs to be out into turning applications from low yielding areas to enrollment. It could be used to see what towns produce the most applications and then which applications from those towns turn into shows.

When I was a college chancellor and we wanted to expand our admissions reach into new territories, we data mined our information. We determine the geographical, social and characteristics of the last three enrollment pools. We discovered that the students tended to be from working class background, towns of less than 36,000, suburban, families that earned less than 80,000 a year.  We also learned that our new students came from areas with mixed racial populations with prominent African-American population segments. Finally, we also discovered that our students were predominantly from areas in which college attendance was not very high so they were often the first to attend higher education.  With that information mined from our own data, we were able to target populations that matched our demographics to increase the probability of enrollment success.

Conversely, if we were seeking to differentiate our student demographics we would use the information to target populations and areas that could yield new students that were different from our current population.

We also used data mining to discover which advertising worked best to motivate student populations to contact the college. After identifying what our on-campus population was like demographically for business majors, we mailed out direct marketing to families in areas that fit the demographics. The date mining created target markets for us.  Our return from the direct market effort was 27% points higher than a random test mailing. .

We were also able to use data mining from a simple program named Leadwise to create predictive models for students from particular high schools in particular majors. This was part of a retention effort to admit students who came from areas that showed better success than those who did not. We were not simply looking to make the initial admissions numbers but to increase retention to graduation. What we found by using a sophisticated data mining program was that students from one high school who entered this major did poorly in composition and were prone to drop out of school within six weeks. That allowed us to first try to reinforce the developmental aspects of composition for these students. That worked a little but they still dropped out in numbers that were larger than any other high school group we studied through data mining. So we worked to direct these students into other majors in which they seemed to do better.

The predictive modeling we were able to do was significant to our show and retention success. We were able to maintain a show rate, the percentage of students who apply, put down the deposit and actually show by adding to our stitch in effort from some data mining. We knew what students from certain areas would likely need to be abler to make it to school and stay based on the patterns we were able to discover from mining the data. We also had a brilliant stitch-in manager who did an excellent job of working with students to get them to come to school. She was able to get a show rate that was often 90+% of all those who had applied. That was excellent.

Data mining would also allow a school to predict the class sizes of consecutive courses such as a foreign language. For example, the data could be mined of all students who took French 1-4, when they took it and how many dropped the courses, what their majors were, when they dropped and what their social demographics were. The model could show then that 100 students started French. Twenty dropped after the course. Ten were undeclared majors; six were majoring in science related areas and four were English majors. The remaining students were humanities, English, sociology, French, German and Chinese majors, with six in other areas. The data could then show that after French 2, thirty more students dropped out or did not re-enroll. Their majors could be plotted and so would those of the remaining students. The same for French 3 and 4. Let’s say that going into French 4 there were only nine students left and they were six French majors and 3 from liberal arts areas. In rough terms, the model would show that there was a 90% drop in enrollment from French 1 to 4.  This would be able to tell a school how many students are needed in French 1 to get a class of say ten in French 4 as well as predict how many students would be in French 4 based on majors. If this program were run for a grouping of cohorts, say five years of French 1 to 4, a very certain and sophisticated predicative model could then be developed that would show how many students need to start French 1 to get ten in French 2,3, and 4 as well as predict the class sizes based on majors.

Imagine the power of a planning tool that could do that? No need to imagine, get a data mining tool but get one that also has a CRM component to fully do the trick.

If we had a piece of technology that would have given us even more productive and predictive information we would have done even better. If we had a really sophisticated data mining and CRM combined tool like Hobson’s Retain, we would have done even better. And if we had used Leadwise fully to identify interests and student needs earlier, combined with its data mining capabilities we would have had a show rate in the 90% range every time.

Leadwise is a planned data mining tool that is customized for each and every client. It sets up a survey for students that asks them specific questions about the interests, motivation and goals that can be used to help with stitch in and show as well as building larger data mining scenarios. Leadwise asks students a set of questions that the school has worked on with the Leadwise representative; usually 10-15 of the questions.

Leadwise matches student interests from a questionnaire to a college’s catalogue and marketing materials to generate a fully personalized College Plan focused specifically on that student’s interests and goals. Leadwise simultaneously takes the details the student enters and sends it to the admissions department as a self-qualified lead in script form. The individual responses to the questionnaire are compiled in a Marketing Management Data Base to inform marketing decisions based on what potential students want and are viewing. 

Leadwise personalizes college catalogs and materials while also generating informed leads that convert to enrollments. The program digitizes college catalogs and marketing materials so that when students complete a brief questionnaire on their background, goals and interests, a fully personalized, on-line “college plan” is created and sent back to the student’s computer. The questionnaires include basic information such as name, etc but also obtains information on radio, magazine, movie preferences through higher tech information like pod casting or mobile webs. The pages all use the student’s name combined only with that information he or she wants. Each section is also personalized to the college using the pictures and welcomes from administrators, faculty and others.

Leadwise also generates a self-qualified lead in the form of “sales sheet” from the questionnaire providing the details an admissions representative needs to help the student decide to enroll at the school.  

All of this is great but the program also has a data mining section that helps target marketing for the school. The program adds the individual details from each student’s responses into a data base that the school can use to inform its marketing decisions. So for example, if a college decides to ask where the student learned about the school or what marketing piece the student recalls, that information is stored in a data base from which the college can data mine to better target its marketing. 

Leadwise™ is a flexible system that is customized and personalized for each school to integrate it into the college’s visual identity. It has been shown to increase applications and interest in enrolling by 14% and has cut admission’s representative time by as much as 34%. This allows admission representatives more time to follow up to increase show rate or enroll more students.

The system was developed by COREdataCenter in New York. Jerry Alloca is the award winning president of CORE and wrote the technical software for Leadwise. For information on Leadwise, contact Jerry Alloca at Core Data

If this piece had value for you, you will want to get a copy of The Power of Retention by clicking here NOW 


1 comment:

Potash said...

Is there any place that is really good at teaching college marketing?