Article

Finding retail marketing insight from within your database

Extracting business insight from raw data is possible through the establishment of a sound analytics process.

June 20, 2010

By Charles Noland and Richard Bradley

Retailers today face any number of formidable challenges in acquiring and retaining customers. More choices for consumers, expanded technologies, growing competition and an unstable economy — all serve to bring dramatic changes to the way consumers shop and make purchases, as well as how retailers keep up.

How can retail marketers know what customers want, what their channel preferences are, how much they will pay and how often they will shop? The answers lie in extracting insight from within the marketing database.

Finding insight in data

Today, retailers seem to be replete with data. In recent years it has become standard practice for data to be collected at nearly every point of a transaction, which is a good best practice. However, the sheer volume of raw data likely is overwhelming, especially if a marketer doesn’t know how to connect the data to inform marketing decisions.

Extracting business insight from raw data is possible through the establishment of a sound analytics process and by professionals that are expert at data discovery, analysis and strategy.

Data discovery

While retailers generally are knowledgeable about their overall business, it’s not uncommon for them not to know nearly as much about their customers, how they behave — and why. There are retailers that have access to a great deal of data but don’t necessarily maintain a database. They may be able to report overall sales, sales by product and sales by period, but don’t know such critical data as sales by customer.

Without such knowledge, building targeted programs that will deliver marketing ROI is virtually impossible.

The first step in turning raw data into marketing insight is data discovery. The analyst begins with a high-level assessment of data sources, looking to answer such questions as:

• Who/what is covered within the data and for what period of time?
• How many records of each type are there?
• At what level are the records stored (customer or household, order total or basket detail, etc.)?
• What is the overall table structure for the data?
• How do the various tables relate and connect?
• What fields are contained within the tables and what do they mean?

Data analysis

The analyst first puts the data in a logical structure. An initial investigation is aimed at ensuring the integrity of the data as it has been pulled together. As an example, the customer view is compared to transaction data so that patterns are revealed. Both figuring out where and how to look for patterns and subsequently interpreting them are informed by past experience.

The analyst will examine the data to determine how many transactions match to customers in order to answer such questions as: What is happening to unmatched transactions? What are loyal members spending on average? In which departments and sub-departments are shoppers buying? What might be the opportunities for cross shopping?

The analyst will develop a marketing opportunity analysis by looking at sales, visits, baskets and coupon redemption — by customer — to determine an RFM (Recency / Frequency / Monetary Value) view. Based on both a prior knowledge of the retailer’s business, the industry and the patterns found in the data, groups of customers are then created. The analyst can then investigate behavior similarities and differences across groups with the goal of pinpointing actionable findings. One desired output is the identification of low-hanging fruit — profitable customers that can be cross sold, for example.

At this point the analyst is looking for answers to the following types of questions:

• How do groups of customers differentiate themselves – what types of customer segments do we have?
• What characterizes frequent shoppers – basket size and composition, spend, timing, seasonality?
• When should we talk to the different customer segments?
• How often should we talk to them and through what channels?
• Based on the shopping behavior of the different segments, how can we optimize the marketing spend against each group (profitable versus unprofitable)?
• How has past contact history impacted sales patterns and customer segment migration?

Data strategy

Such a systematic approach improves triggers and targeting and helps retailers find insight into what to market to whom and when. Consider that a retailer has two groups of one-time visitors. Some of those fall into the “will never see again” category. Others will shop with the retailer again. Through data analysis, it can be determined which customers have the potential to spend more. The resulting data will help a retailer create a strategy for determining how much a customer is worth and how much they will be worth over time. From there, differentiating communications and contact strategies can be created and marketing investment in different customer segments can be tailored to their potential.

Data is powerful and has the potential to provide tremendous insight that can be used to inform future marketing strategies and tactics. Savvy retailers would do well to mine their existing marketing databases to drive profitability in any economic condition.

Charles Noland is senior vice president of marketing analytics and Richard Bradley is senior consultant for Harte-Hanks. Harte-Hanks, Inc., San Antonio, TX, is a worldwide, direct and targeted marketing company.