How to Analyze Customer Data for Data Driven Marketing - dummies

How to Analyze Customer Data for Data Driven Marketing

By David Semmelroth

Much of analyzing customer data in data driven marketing requires some advanced knowledge of mathematics and statistics. There are pitfalls to look out for and how to evaluate the results of advanced analysis. This article introduces you to a couple of the more common types of insights that can be coaxed from your customer data.

Group customers into segments in data driven marketing

Database marketing relies fundamentally on the process of identifying the target audience. This is essentially an exercise in finding groups of customers who share similar traits.

A great deal of the analytic work that goes on in marketing in general has to do with finding such groups, called segments. Such work is generally referred to as segmentation. Here are a few approaches to segmentation that are commonly used in data driven marketing.

Geographic segments

Perhaps the easiest to understand is geographic segmentation. Households in the U.S. are divided up into a hierarchy of geographic groupings. These groups are defined by the Census Bureau. They range from individual city blocks, to block groups all the way up to what are called Metropolitan Statistical Areas, or MSAs, which are essentially large population centers.

The reason that these segments are important to marketers is that they reflect the way mass media is distributed. The viewership of local TV network affiliates is closely aligned with these MSAs. The same is true of radio audiences and the subscription bases of local newspapers.

What’s more, these are the groupings that the Census Bureau uses to report census data. That means that marketers have access to a wealth of demographic and economic data on these areas.

Weather, culture, and even school calendars vary geographically. This means that you’ll often have reason to use geography in developing your marketing campaigns. Regional differences can help drive decisions about what products to offer, how to craft your message, and even when to communicate.

Demographic segmentation

Demographic data is a core part of your database. This data generally needs to be purchased if you want to have it at the individual customer level. There are a number of vendors out there who provide a wide range of customer level information.

Life stage traits such as age, marital status, and number of children all affect the product needs of consumers. They also affect the types of messages that will make an emotional connection. Income drives affordability and price sensitivity. Homeowners need a wide range of products and services that aren’t needed by renters.

As an ongoing part of your job as a database marketer, you’ll be involved in trying to define and refine a demographic segmentation scheme for your customers. Because every business is different, there really isn’t a one-size-fits-all segmentation scheme you can universally apply. By combining your customer purchase history with demographic data, you can develop a segmentation scheme tailored for your particular business needs.

Behavioral segmentation

Another common approach to segmentation involves analyzing transaction data. The idea is to group customers together based on the way they interact with you.

A standard behavioral segmentation in the credit-card industry involves looking at how customers use their cards. They’re grouped into transacters, revolvers, or inactive according to how much they use the card and whether they pay off their balance every month. This is a useful distinction, because the nature of the revenue these groups generate is very different. Marketing campaigns are dramatically different depending on which group is being targeted.

Behavioral segmentation is particularly fruitful with web data. When customers visit your website, they generate a huge virtual paper trail. You can tell what pages they’re looking at, what they’re searching and shopping for, and even how long they stay on your site. This data is extremely useful in developing relevant database marketing campaigns.

Build response models in data driven marketing

It’s possible to use the results of past campaigns to define target audiences for future campaigns. This exercise is known as response modeling. It involves looking at the various customer attributes of responders. Some of these attributes turn out to be closely associated with how likely a customer is to respond to a given campaign.

When advanced statistical techniques are applied, it’s actually possible to throw a bunch of different attributes into the mix all at once. These statistical procedures then tease out the combination of attributes that best predicts a customer response.

How to measure results in data driven marketing

Your job as a database marketer in data driven marketing begins and ends with the customer. Focus on the individual customer is the defining principle of database marketing. But there’s another defining characteristic of database marketing that’s a close second. Database marketing is measurable. Everything about it is measurable. Messages, offers, response rates, and even revenue can all be measured.

Not every one of these things can be measured for every marketing campaign. You need to pick and choose your battles, so to speak. But there is nothing about these campaigns that you can’t measure or test if you so choose.

There is an age-old adage among marketing executives that laments their inability to really understand marketing effectiveness. They know, so the saying goes, that only half their marketing is working. They just don’t know which half. This adage does not apply to the database marketer.

The secret lies in the fact that you know who is responding to your campaigns so you can tie responses back to individuals in your target audience. This allows you to essentially design marketing experiments.

These experiments can test the success of one offer over another. You can scientifically compare the response rates of different messages. And you can even confidently calculate the revenue that is specifically due to your efforts.

By properly setting up your campaigns as experiments, you can not only learn a great deal about what works and what doesn’t — you can also take credit for quantifiable contributions to your company’s bottom line.