21st Century Marketing for BSCs
The way companies do business really is changing. In an occasional series on the revolution in marketing, we'll take a look at what's new and different and how you can profit. This article looks at new ways to identify prospects.
By Donald E. Tepper
Marketing techniques for building service contractors will change radically in the next few years. It's not because anything different is happening with building service contracting. Rather, the changes are occurring as the science and art of marketing changes. But they will revolutionize the way you seek out and develop new business. In fact, although you may not have noticed, they already have.
These new marketing techniques do two things differently than "the old way."
First, they try to understand why a person--a customer, a consumer, a purchaser--behaves the way he or she does.
Second, they try to specifically target your potential "best" customers.
But you might say, "That's what I've always tried to do. I try to understand my customer. And I try to reach my best prospects. So, what's different about this new marketing?"
Quite a lot. Some of it has to do with computers. Some with new discoveries. And some with new ways to accomplish old tasks, such as printing and mailing.
Let's get specific.
Data Mining-Prospecting For Gold
In the Gold Rush of 1849, miners sifted through tons of rock for a few ounces of precious gold. In the Marketing Rush of the 21st Century, you (and your competitors) are sifting through piles and piles of printouts to locate a few nuggets of valuable information.
The one thing you need to begin is data-lots of data-on your customers . . . and probably your prospects and ex-customers as well. What you're trying to do is find new patterns and trends. Actually, the patterns aren't new. They've just been hidden-buried under all the other facts and figures. Once you find those patterns, you act on them.
Here's an example.
Let's say you decide one morning: "I want to find more good customers."
The next step is defining what you mean by a "good customer."
The definition is up to you. Some building service contractors would define a good customer as one yielding the highest profit over the length of the relationship. Let's accept that definition.
If you've been collecting data on your customers-by account, by job, by location, and so forth, you should be able to determine who your most profitable customers are.
But that's not really data mining.
What you want to do is draw a profile of those customers.
In both retail and business-to-business situations, you can start by asking such questions as:
- In what geographic areas are our customers based? How are they distributed across the areas we serve?
- What is the average size of our customers?
- How long have our customers been customers?
- What is the average response to our various marketing offers?
- How much money do our customers spend with us?
The answers to these and similar questions will help paint a picture of your customers. The next step is to profile your "good" customers-those who, over the length of your relationship, are the most profitable.
Then compare the profile of your "good" customers to your entire customer base. Also compare that profile to your prospect base.
For example, prospects whose profiles are similar to your "good" customers-size of company, nature of business, location, services required, and so forth-may be worth pursuing more actively than those prospects who don't resemble your "good" customers. You can even use your profiles of "good" and not-so-good customers to predict how much revenue you'll receive from those prospects.
It's likely that you or your salesmen already try to "qualify" prospects. But are you sure you're asking the right questions?
Also, by examining who responds to your various marketing offers (telemarketing versus direct mail, for instance), you can see which marketing technique is most likely to attract "good" prospects.
Data mining can also be applied to existing customers. Companies in other industries have used data mining to identify their most desirable customers, and then concentrated their marketing efforts on those prospects with the most "average lifetime value." That's further boosted the income from those customers.
And finally, data mining can identify customers most likely to "defect"-to change building service contractors or to consider in-house contracting. What patterns-what hidden patterns-are there to your lost customers? Wouldn't it help to be able to identify them well in advance and perhaps prevent their loss?
Tools You'll Need
Data mining can't be performed using "gut feel." After all, the very definition of data mining is finding new patterns and trends by sifting through data. You'll need the data, and you'll need the tools to "crunch" the data. It can be done, albeit crudely, with computer spreadsheets like Excel and Lotus. But if you're serious, you'll probably want a specific statistical or data mining package such as developed by SPSS or SAS.
Collaborative Filtering (CF) has been described as automating "the process of 'word of mouth' by which people recommend products or services to one another." From the end-user's standpoint, collaborative filtering typically resembles the following process:
- An individual provides a profile of likes/dislikes, typically within a particular category (books, music, etc.).
- The individual is provided with a listing of other items within the same category that he/she is expected to like.
For building service contractors, the CF's value is to recommend additional services to a customer, based on what other similar customers have ordered.
But it's not a simple process. Here's why:
Collaborative filtering begins by registering or recording the preferences of a large group of people. Their preferences may be based on surveys or actual buying patterns. Let's say you've got a lot of information on a variety of services your customers have bought: office cleaning, carpet cleaning, relamping, parking lot cleaning, recycling, and so forth. You also know what frequency of cleaning is needed. And you have other demographic information-type of company, size of company, location, and so on.
The second step is to gather as much of the same information about your target-let's say a current customer.
You match your target up with the portion of the larger group that most resembles your target.
Let's say your target is in industry A, uses services B and C, and is in geographic area D. You match your target up with a group from your entire customer base that most nearly matches his profile.
You take a look at what other services are bought by your customer group segment. You then recommend those other services to your target.
For example, after using collaborative filtering we find that others who are in industry A, use services B and C, and are in geographic area D also are big customers of service Q. Our target customer has never asked for service Q. It's never even occurred to him. But let's try and offer it. If he's like the rest of his group, he just might buy it.
What's the difference between collaborative filtering and data mining? Data mining uses a lot of data and looks through it all for new patterns. It's a one-way process.
Collaborative filtering is a two-way process. It begins by determining the preferences of a group of people. Then it looks at an individual: What does he or she want or like? CF takes the individual's preferences and forms an "on the fly" group of like-minded individuals. CF poses the question: "What else did this group like?" Maybe something the target individual hasn't even thought of. Finally, it takes the group's other likes and preferences and offers them to the individual.
A key concept is that CF is a filter. It's filtering out many options, letting through only those choices that appealed to the larger group.
Tools You'll Need
Collaborative filtering can be attempted somewhat crudely using either a spreadsheet program with statistical capabilities or a database program such as Access. For best results, however, you'll need a commercial program like LikeMinds developed by Andromedia-now Macromedia.
Geodemographic Cluster Analysis
Okay, so this one's a mouthful. Geodemographic cluster analysis (GCA) is most often applied to household and neighborhood marketing. It means analyzing geographic areas based on demographics. Residential cleaning would be a good service to use GCA.
GCA divides neighborhoods into groups based on similarities in income, education, and household type, as well as attitudes and product preferences.
Early cluster systems were based on U.S. Census Bureau data. However, because the Census Bureau isn't allowed to release information about individuals, the neighborhood-level systems can only offer generalizations. Newer systems add on independently-derived data such as lifestyle choices, media use, and purchase behavior.
For example, a company called CACI has developed the ACORN neighborhood segmentation system. It defines 43 types (or "clusters) of neighborhood segments. Each neighborhood is made up of a mix of these clusters and is defined by its dominant cluster.
Clusters range from the affluent (1A: "Top One Percent," 1B: "Wealthy Seaboard Surburbans," etc.) to the Upscale (2A: "Urban Professional Couples," 2B: "Baby Boomers with Children") to Retirement Styles ("4A: Retirement Communities," 4B: "Active Senior Singles") to Downtown Residents (8A: "Young Immigrant Families," 8B: "Social Security Dependents.")
Here's an example for Zip Code 22046 (Falls Church, Virginia). CACI identifies it as Cluster 1B ("Wealthy Seaboard Surburbans"):
This type is located along the eastern seaboard and in California. Residents are married, middle-aged professionals at the peak of their lifetime earnings. They rank among the highest for auto club membership and travel extensively. Media preferences include Metropolitan, Boating and Yachting, and the New York Times. This market's share of disposable income is relatively high.
Is this cluster a good one for residential cleaning services? Maybe. In some cases, the research companies have already compiled that information. In all cases, you would want to check your own records and find out where you've been most successful in marketing your services, determine the "clusters" most likely to buy your services, then market to other neighborhoods with similar characteristics.
Other marketing and research services have their own groupings. For example, Claritas divides 62 PRIZM clusters into 15 broader groups.
Claritas describes Zip Code 22046-the same one we considered above-using four groupings: "Winner's Circle," "Executive Suites," "Pools & Patios," and "Suburban Sprawl." Let's compare "Winner's Circle" to CACI's "Wealthy Seaboard Surburbans":
Executive suburban families. Age group: 45-64. Professional. Household income: $90,700. This PRIZM Cluster is most likely to: have a passport, shop at Ann Taylor, have Keogh plan, watch NYPD Blue, read epicurean magazines.
And Metromail-now Experian-offers US Mosaaaic, another geodemographic lifestyle segmentation system.
Tools You'll Need
Since this information is developed by research firms using a combination of Census Bureau data and other proprietary information, you'll need to buy information from one or more of the firms mentioned above, or one of their competitors.
These are only three examples of the new science of marketing. Others have been around for years, but are also being refined with new knowledge and new technology. These include content-based filtering (in which individuals filter their own data) and database marketing (in which firms collect data on individuals and households, then serve as links between the sellers and the consumers).
In future articles, we'll take a look at how the new marketing techniques are being used today and how they may be used in the years to come.
Donald E. Tepper is the editor of Services.
For More Information
CACI International Inc.
1100 North Glebe Rd.
Arlington, VA 22201
1525 Wilson Blvd., Suite 1000
Arlington, VA 22209
505 City Parkway West
Orange, CA 92868
SAS Campus Dr.
Cary, NC 27513
233 S. Wacker Dr., 11th floor
Chicago, IL 60606
School of Information Management & Systems
102 South Hall
Berkeley, CA 94720
(See "Collaborative Filtering Workshop" at www.sims.berkeley.edu/resources/collab/collab-report.html)
611 Mission St.
San Francisco, CA 94105