In a world of algorithms, linear models, and neural network models, multichannel merchants often ask the same question: Now that we have data on our customers, how can we use the information to improve response?
Marketers can easily get caught in a trap of collecting data from customers just for the sake of collecting data, says Tony Chivari, vice president of marketing for New York-based general merchandise mailer Spiegel. Then they sort through the collected data, review the findings, and wonder what to do with them.
Often that leads to merchants’ asking for personal information such as age during the checkout process. Such information might not necessarily help them improve their target marketing — and asking for it can lead to higher site abandonment rates.
“You can tell a lot about your customers by what they are ordering, so you don’t need to ask them any questions that may come off as intrusive,” Chivari says.
What is actionable, then?
Information regarding recency, frequency, and monetary value of purchases — which makes up the warhorse of database marketing, RFM — is actionable data that’s already on file. So is the type of products customers buy and the channel they order them through. When it comes to taking action on the data, many experts believe that recency is king.
Arthur Hughes, vice president/solutions architect for Richardson, TX-based KnowledgeBase Marketing, says that 30-day hotline data are important because the most-likely buyers are the ones who have recently made a purchase.
“People get into a buying mood, and that’s when a cataloger has to hit them,” Hughes says. Catalogers need to treat new movers in the same fashion, since they have the same behavior traits of buying early and often.
“The behavior of new movers is important because they are most likely to think about buying new things,” Hughes says. New-mover lists open the door to more than just home-improvement merchants, he says, as these consumers not only have a new home to fix and furnish but also may be experiencing a lifestyle change such as a new job, marriage, or a baby.
Before you hit new movers or 30-day buyers with offers, though, you have to know what they respond to and how they shop. If, for example, you have a group of buyers who purchase only items that are on sale, you might want to omit them from most of your mailings but be sure to hit them with discount books and e-mails. Customers who buy only items that are new to your inventory should receive mailings and e-mails calling attention to what’s new and hot.
“Look at how I am behaving before you try to segment me,” says Anne Holland, president of Warren, RI-based marketing research firm MarketingSherpa. “The key for catalogers is to measure what can be told about their segmented buyers that you can react to. You should segment your house files to send your customer a more relevant item that may move him or her to buy. If I react to new items, send me a different catalog than my neighbor, who may react to less trendy items. You can tell this just by looking at your database and seeing what I order.”
Using its databases to break down customer purchase information, Chivari says Spiegel has been able to do well by cross-selling to consumers with its four specialty catalogs: Discover Spiegel Catalog (a smaller, “greatest hits” book especially for prospects), Travel & Swim, Home & Decoration, and Together (women’s apparel, targeting baby boomers). “Part of the success is getting people to cross-shop,” Chivari says. “As wonderful as a target shopper can be, your best long-term customers are the ones you can cross-sell to.”
Deciding which target catalog to send to which consumers requires looking at customers’ entire purchase history, not just what they’ve purchased in the past 30 days, Chivari says. For example, a customer who is on the record for buying a swimsuit a year would be sent the Travel & Swim catalog.
“You could send them a catalog with the next-closest related products, or one with apparel with a similar style or feel,” Chivari says. “Or if the customer appears to have only one interest, send them the main catalog with hopes it will spark an interest.”
Spiegel also uses purchase data to draw customers back for additional purchases. Say Jane Doe bought a green cardigan. Spiegel might comb its database to see what accessories were bought by other customers who also purchased the green cardigan, then e-mail Jane Doe an offer promoting those products.
Model methods
Predicting what items to upsell or cross-sell is as simple or as difficult (depending on your point of view) as modeling your lists, whether you’re using house files or co-op lists. Models do a great job of predicting likely behavior, says Jim Wheaton, cofounder/principal of Wheaton Group, a Chicago-based data-mining and decision-sciences practice. But he notes that their output is heterogeneous.
“For example, decile 1 of a model invariably will contain all sorts of different customers in terms of behavioral and demographic characteristics,” Wheaton explains. “This makes cost-effective targeting very difficult. In contrast, homogenous clusters by definition display at least some commonality across customers.” Finding the data your customers have in common, he adds, makes it much easier to target them in “resonating ways.”
If you’re looking at customer acquisition efforts, says Kelly Hlavanka, director of Milford, OH-based loyalty market services provider Colloquy, either a logistic regression or neural network model is a good fit. The categories of data that need to be considered are demographic overlays, direct mail responsive scores (including presence on multiple list sources), physical attributes such as distance to nearest retail location, and financial scores from credit bureaus.
Hlavinka says an upsell campaign would be one of the best reasons to use a neural network, since you’re combining qualitative information such as customer acquisition rates with responsiveness, and creating more than two categories.
But for sales campaigns, Colloquy’s senior statistical consultant Norbert Schumacher suggests building a two-stage model: the first to predict the probability that a consumer will respond to a catalog; the second, using data from the first-stage responders, to predict the level of spend.
Tune into the channel
The channel used to make a purchase, along with the media that drove the buyer to that channel, are also highly actionable for crafting a contact strategy.
“If you’re a woman’s apparel catalog, and you’re targeting between the ages of 35 and 55, you’re going to have a certain segment that will respond better to e-mail offers than print offers,” MarketingSherpa’s Holland says, “and you should be able to figure that out through your database.”
Because the channel that drove the purchase isn’t always the same channel through which the purchase is made, merchants should make a point of tracking the promotion code as well as the response code. This tells you how your business works, says Linda Huntoon, executive vice president of consumer brokerage for Greenwich, CT-based list services firm Direct Media.
“If your business is 30% Web, it doesn’t mean your customers are responding to Web offers,” Huntoon reiterates. “You need to keep a close eye on the promotion code as you do the response code, because it will give you a better understanding as to how your customer is behaving.”
“Knowing your response allocation is extremely important,” agrees Jeff Hassemer, director of product strategy at Broomfield, CO-based database services provider Abacus. “You need to find out what marketing campaign drove your online transaction. This is not to say a cataloger needs to collect a key code, but you need to know if a mailing drove a customer online, or if he found you through a search engine.”
For instance, you can review your Website logs to call out the customers who searched for products by item number; they were most likely are driven to your site by a print catalog and therefore should receive more mail offers.
Information please
Though you can learn a lot about your customers through what and how they order, as well as by purchasing demographic overlays and the like, sometimes the only way to get specific, actionable data is to ask for it. The challenge is to do so without alienating the customer. If a prospective buyer feels that what you’re asking is overly personal or that you’re wasting his time by asking for too much information, he’s going to think twice about completing his transaction.
One way to avoid that is to wait until the purchase has been completed before asking for information. To encourage the customer to stick around and answer a few questions, many companies offer an incentive — a discount on the customer’s next purchase, for instance.
Even here, though, you need to know what segment of the audience will respond to which offer. Not everyone is motivated by a discount, notes KnowledgeBase’s Hughes. He recalls a gifts cataloger that e-mailed him an offer for 10%-15% off his next purchase; Hughes says he was insulted by the merchant’s “stupid” mistake.
“I’m not motivated by a 10% discount, I want to make sure the right gift gets to the right person,” Hughes says. “If they segmented their customer list by income, they would be able to offer those with a lower income a discount and those with a higher income something else, maybe a service like free overnight shipping. They could also use that segmentation to deliver a great quality catalog and a discount catalog.”
Even if you are offering an incentive — and the right type — be sure that you don’t ask for too much data and that you can use the information you expect to receive.
Hughes cites the Website of general merchant Sears as one that does an excellent job asking customers for relevant information. “I bought an appliance online, and Sears asked me three questions: How technical are you? Do you buy top-, medium- or bottom-of-the-line items? And what is your household makeup?” Hughes says. “Right then they know if they can send you offers about electronics, if you’re going to be motivated by sales or straight pricing, and what offers they should send you regarding which departments.”
The most important bit of data you can get from your customer, he adds, is the e-mail address. Without that, you miss out on the obvious chance to hit customers with targeted e-mails that can draw them directly to you e-commerce site. “Once a customer gives you permission to send them e-mails,” Hughes says, “it opens up a whole new relationship between the marketer and the customer.”