Your own personalized shopper

While shopping online for a computer a few years ago, Fred Riggins noted that three Compaq models offered the FireWire feature he wanted. The three were priced on a low, medium and high range, corresponding to which processor they used. While Riggins might have bought a more expensive model if that's all he saw, he nevertheless bought a cheaper one because it was staring right at him on the screen. For Riggins, an information systems professor who's been teaching courses on e-commerce since 1996, the experience got him questioning the effectiveness of product display and pricing online. In his working paper, "Using Personalization Technology to Reduce Product Cannibalization," Riggins looks at how online stores can segment and sell to high- and low-end customers without cannibalizing their best sales.

While shopping online for a computer a few years ago, Fred Riggins noted that three Compaq models offered the FireWire feature he wanted. The three were priced on a low, medium and high range, corresponding to which processor they used. While Riggins might have bought a more expensive model if that's all he saw, he nevertheless bought a cheaper one because it was staring right at him on the screen.

Most online shoppers aren't researchers looking into online pricing strategy, so few would have given much thought to the mental switch he made in his buying behavior. However, for Riggins, an information systems professor who's been teaching courses on e-commerce since 1996, the experience got him questioning the effectiveness of product display and pricing online.

In his working paper, "Using Personalization Technology to Reduce Product Cannibalization," Riggins looks at how online stores can segment and sell to high- and low-end customers without cannibalizing their best sales. Riggins finds that online stores should take an approach similar to that of grocery stores that want customers to buy high price cereals: Feature the premium products and make them easiest to find while making offerings that appeal to low-end shoppers less prominent and harder to find.

Slicing the pie

A well-worn maxim preaches that you can't be all things to all people. A company that tries — even a behemoth like Wal-Mart — would spread itself too thin. But that's not to say that a business should only sell one product for one price for one particular customer. A Ferrari dealer might stock other high-end cars, for example, and might have a product line-up of Ferrari models that range in price and quality.

Even Wal-Mart, known as a budget retailer, still sells a range of lawnmowers and TVs that vary by price and quality. The reason vendors offer a range is to prevent customers from walking away empty-handed because you didn't have the right product for them at the right price.

The challenge for sellers is how to offer a range of products without losing big ticket sales. For example, if a young, high-earning computer gamer has a $2000 laptop in mind when he visits the Dell web site, will Dell or the customer be happy if he sees the $500 econo offering and buys that instead? This is called cannibalization: when a low-price sale replaces a high-price one.

Brick-and-mortar stores already segment customers. Shoppers know by brand which supermarkets stock high-end foods and which carry just the necessities, and stores vary depending on the demographics of the neighborhood. Within each store products are further segmented. Consider the cereal aisle: if you want the expensive, brand-name cereals, they're displayed at eye-level or on an end-cap.

If you're looking for the budget boxes, you'll have to get down on your hands and knees to dig for the one you want. Similarly, salespeople can make some educated guesses about what you're looking for. Walk into an auto showroom in a suit and tie and you may be shown the 4-door sedan with leather seats but walk in with work boots and a flannel shirt and you may be steered towards the pick-up trucks.

Segmenting the online store

The challenge that Riggins is exploring is how sellers can best segment their customers online. Although online stores can't see if you're wearing a suit or a flannel shirt, technology can make the segmentation process much more efficient. Except when a retailer has data as to who it is that has just clicked in, Riggins recommends highlighting higher-end products. Getting online sales right, however, depends on personalizing the product offerings and sales experience to the customer.

That means converting a faceless online visitor into a person with characteristics that can be correlated to product preference. Online sellers can make rudimentary guesses about visitors based on the basic data transmitted by the browser, Riggins says. For example, a high-speed connection running the Firefox browser is a clue that the visitor might be more affluent than someone with a slower connection using Internet Explorer.

Web analytics data also reveals domains. A visitor coming through the online door from the Goldman Sachs domain might spend more than a visitor from a small company. City names and Zip codes are clues about the spending power of a customer, too. And if an anonymous shopper arrived through an ad on the Wall Street Journal's site, a retailer might presume that person to be a higher spender versus a visitor who clicked through an ad in People.

Online sellers have a growing number of resources that provide even deeper, more specific information. If you've visited the seller's site before — especially if you have an account — companies have scads of information about what you've bought, what you've looked at and what you've decided not to buy. Online vendors create a more complete picture of who a visitor is and what he or she does online.

The question is, what does the online seller do in response to having created this customer profile? The average Internet shopper doesn't think much of the absent-minded click-stream created when shopping for a new Dell laptop, adding features and subtracting others, but Riggins says you can be sure that Dell is very aware of all of those moves and is analyzing the data to draw a picture of who you are and what you want.

A personalized experience

How does all this data translate to a personalized experience? For Dell, it might mean hiding most of its products and only showing a visitor the few that best suit him. Instead of asking visitors to identify the product or price category that fits, these back-end technologies do the sorting automatically.

Currently the Dell site asks visitors to select Home user, Small/Medium Business, Public Sector, or Large Enterprise when entering the site. But, with a little information, the site might automatically display its high-end Alienware brand to a visitor based on his computer settings and the fact that he clicked on an ad running on a gaming geek site.

But what if the educated segmentation guess isn't right and the visitor is really looking for the cheap, bare-bones laptop? Riggins says it'll still be there, but the buyer might have to be a more dogged shopper to find and buy it. In fact, he says, the site might actually strive to make it a challenge to make it down the stairs to the bargain basement, particularly when the profile suggests a high-end customer is visiting the site.

On the other hand, segmentation may peg you as a low-end buyer, but a seller may still find value in selling to you by changing the price on a high-quality product rather than offering you a cheaper one. In this scenario, a visitor segmented into a price-sensitive bucket might see a special coupon code displayed on a site when he visits it, but the higher-end buyer would not.

Although there's little difference between this online tactic and a company sending direct mail coupons to less prosperous Zip codes, Riggins notes that an each-according-to-his-need selling philosophy can be a tricky one to negotiate. Customers don't react well when they discover that others paid less for the same product.

As far back as 2000, Amazon was trying to perfect its price customization and dynamic pricing tactics, but got into hot water when customers discovered the same products being offered at different prices to different shoppers. A 2005 study on Internet usage and buying habits from the Annenberg Public Policy Center found that 76 percent of respondents agreed that "it would bother me to learn that other people pay less than I do for the same products."

Thus, while vendors may see great value in personalizing their prices and storefronts, the blowback may translate to "seller beware." Ultimately, though, online sellers must address at least one important implication of their attempts to use personalization to fight cannibalization. The cannibalization problem can cause a seller to not want to serve low-end customers in a marketplace, Riggins notes.

"Because sellers will find personalization technology useful to better understand customer preferences, they may need to take less action aimed at fighting cannibalization. When this happens, there is less chance low-end customers would be shut out of the market." However, Riggins cautioned that "the impact this use of personalization technology has on customer privacy is a societal issue that must be addressed."

In the Classroom:

• This topic is covered in Riggins's "Competing as an E-Business" program in the Masters of Science in Information Management (MSIM) program as well as the undergraduate "E-Commerce Strategy" course. He has been teaching courses on e-commerce since 1996.
• The course covers how online storefronts attempt to segment their marketplace and how they set prices in the online world.
• Practical takeaways from the course include an understanding of segmenting and online price discrimination and the importance and benefits of personalization.
• The class also discusses other selling and personalization offerings where informational web sites offer some content for free (a low-end offering) but hope to hook users so they'll upgrade to paid, premium content.
• Riggins uses the topic of personalization and data collection to explore the business and societal ramifications of privacy online.