Month: September 2016

Customer Data Mining in the 21st Century

Let’s say you’re at the supermarket checkout line and you hand the cashier your grocery store loyalty card. She scans it, and you watch the bill drop because of on-the-spot discounts on a handful of items. When you get your receipt, it includes a few paper coupons for a couple of items you saw this week and made a mental note to try next time. Data mining makes it possible.

Maybe you’re home on a Friday night, browsing Netflix. You’ve watched all the episodes in your favorite show, and decide to try something new. It turns out Netflix’s suggestion to you is perfect. Data mining makes it possible.

What is Data Mining?

Data mining is sifting through all of the data collected by a business or organization, searching for relationships so that reasonable predictions can be made with regard to behavior. This kind of activity is most common among businesses that deal with consumer behavior of any kind. With the results of a data mining project, a business can make evidence-based decisions when looking at things like sales and marketing efforts, customer retention, fraud deterrence, and more.

In nearly all cases, because of the sheer volume of data available, and the ability of a computer to analyze data with speed and efficiency, data mining software is used, rather than human staff. With specialized and powerful software applications, a business is far more likely to discover relationships and correlations among data points that might be unexpected.

For example, let’s say you’re looking at smartphone customers of a certain age, with minimum income levels, and at least a master’s degree. It’s possible to assume that these folks would be your best smartphone customers, but maybe your data mining project demonstrates that folks with only a high school diploma are actually your best customers in terms of loyalty, regardless of their income levels.

Or put even more simply, you’re assuming that when someone has an umbrella, it’s because they’re expecting rain, when they’re really planning on using it to keep the sun out of their eyes. Data mining helps you identify the difference, and then know when to sell that person rain boots or sunscreen.

Data Mining for Beginners

In general, the overall data mining process is relatively straightforward. There are four steps:

  1. determine what information to collect
  2. decide how your business will store, maintain, and access data
  3. choose and operate powerful software analysis
  4. use results in business planning

If anything, you probably already have a plethora of data that you are collecting or are able to collect. Here’s where a reliable software application comes in handy – people simply aren’t able to examine the sheer volume of collected data in a way that draws subtle relationships and correlations with efficiency, like a computer program can.

Data warehousing includes the storage, maintenance, and access of data by the organization that has collected it. Storage is relatively cheap, but businesses that collect consumer data still need to worry about privacy and security of this data, as well as the ongoing costs of keeping it safe and accessible.

Choosing the right software analysis program is perhaps even more important than choosing what kind of data to collect. You need something that’s robust enough to handle any queries you have with speed and accuracy. It’s analytical and reporting abilities must meet your needs and expectations, and evolve with your business’s changing needs, as well.

The reports that you receive can be used to support a wide variety of business planning activities, including sales and marketing efforts, retention activities, fraud detection, and more.

Data Collection

While some data collection points are pretty obvious, some might not be; it’s collecting those pieces of information that look like outliers where you may be most likely to find success in isolating unexpected relationships and associations. Those unforeseen relationships can be the most valuable. For example, naturally you’d want to collect names, addresses, and spending habits of your customers. You may also want to collect data about purchased items, days of the week and times of purchases, and more. There’s really no limit to what you can collect because you can’t assume anything is irrelevant.

Data Mining Software

There are several different kinds of data mining software available. Many companies use outside vendor tools. Other businesses use open source or free software options. Each type presents pros and cons unique to your business. A large part of the decision-making about this solution comes down the skill and knowledge of the people in your business who will be running the software, as well as what kind of computing power you need.

Data Mining Models

In broad terms, there are several kinds of models that can be developed from your data and used in decision-making, including descriptive and predictive models.

Descriptive modeling sorts data into categories and looks for sophisticated relationships among those categories. Key types of analysis include identifying anomalies, common characteristics, relationships (primary and secondary), and similarities. This kind of modeling gives you a picture of the past. With predictive modeling, data mining tries to make realistic guesses about future behavior based on complex patterns among known data, and probabilities of future occurences.

5 Things You Can Do With Data Mining

A reliable and powerful data mining program can have real results:

  1. use to avoid churn and detect fraud
  2. strengthen customer loyalty
  3. identify new target markets
  4. “market basket” analysis
  5. trendspotting

Predictive modeling can help you figure out when and why customers leave or commit fraudulent acts, taking the appropriate action to help minimize that behavior in the future. Data mining also helps companies strengthen customer loyalty by recognizing patterns and habits, allowing businesses to reward that behavior through incentives such as special offers, coupons, or discounts.

Thorough analysis can turn up unexpected relationships and lead to the identification and development of new markets for a company’s products or services. Market basket analysis allows companies to anticipate customer needs because if they purchase one group (market basket) of goods and services, they may be interested in certain related items. And finally, data mining allows companies to spot trends, putting them ahead of the curve instead of behind it.