The call for an audit of Facebook’s metrics comes a week after the social network acknowledged inflating its video metrics.
(Page 2 of 3)
Personalization technology itself takes up to five forms, and retailers need to decide how many of them they need for targeting and serving customers, experts say. "There are a lot of different ways you can go with personalization, it depends on what you have in your arsenal and what you want to get accomplished," says Adam Sarner, analyst with research and consulting company Gartner Inc.
He lists five basic methods of personalization, covering a mixture of marketing and merchandising strategies:
l Rules-based, where a retailer uses a personalization software application from a company like BroadVision, ATG, E.piphany Inc. or Blue Martini Software Inc. in which particular online shopping actions trigger follow-up personalized e-mails that lure customers back to a web page configured to their interests;
l Collaborative filtering, a technique mastered by Amazon.com, which presents cross-selling opportunities in real-time as customers reveal their interests online;
l Statistical modeling, considered a more sophisticated "black-box" method of personalization from companies like Touch Clarity Ltd. that use algorithms to figure what consumers with particular shopping histories will like to see in merchandising displays and marketing messages;
l State-based personalization, which is designed to respond to changes in a customer`s personal state of shopping behavior. If a regular customer suddenly stops returning to a favorite web site, for instance, that could trigger an e-mail message with a discount that links to a web page populated with merchandise the customer has been known to favor;
l Configuration options, such as virtual modeling or web pages that allow a shopper to self-configure a personal computer.
Prices for personalization systems vary widely, especially if they`re implemented as part of new complete e-commerce system. A software package, for instance, can range from $500,000 to millions of dollars, ATG says.
One thing at a time
"One personalization technique doesn`t necessarily work best in all situations," Sarner says. If a retailer isn`t sure where to begin, he adds, it needs to first determine who its customers are and what they`re looking for. "The mistakes made in the past with personalization occurred when retailers tried to do too many things at once--cost savings, bringing new people to a site, cross-selling," he says.
A pure-play online retailer offering a broad range of merchandise, he adds, may benefit most from collaborative filtering to encourage shoppers to add a variety of products to their shopping carts.
But a multi-channel retailer that wants to focus on a strategy of pushing its highest-margin products to its highest-value customers, he adds, might do best with a statistical modeling version of personalization. "Statistical modeling helps the retailer to decide where to put its resources to target segmented customer models," Sarner says.
Personalization strategies are beginning to overlap with efforts to use web analytics and marketing optimization tools that monitor online shopping behavior and provide reports on customer preferences.
"Personalization and marketing optimization are blurring together," says Todaro of ATG, one of the pioneers in personalization technology. "Today, you really can`t have a successful marketing program without tools that segment what you deliver."
Wine.com, for instance, uses web analytics from Omniture Inc. to identify customer preferences for particular merchandising offers, a strategy that helps it tailor merchandising and marketing campaigns for the many segments of wine and gift customers, Shaffer says.
Wine.com must sort through customer preferences for its more than 10,000 products, including gift baskets, that can also be affected by state laws that restrict out-of-state shipment of wine. By using web site cookies that record a customer`s home state as well as buying preferences, it can tailor follow-up e-mails to visitors, offering promotions that induce them to click to web pages configured to their interests. "As long as a cookie shows that, for instance, you`re a customer that buys only $15 bottles of wine, I can serve up a shopping experience personalized to your interests," he says. "The sooner I can get you looking at what`s relevant to you, the sooner I can get you to close the deal."
Effective personalization, experts say, requires a broad stretch beyond just using web site cookies to track a shopper`s web site traffic in order to respond in subsequent visits with a web page adorned with a personal greeting and displays of items related to past online shopping visits. Getting maximum mileage out of personalization, they add, requires combining shopping behavior with a shopper`s personal data in a CRM application and with the shopping behavior in all the retailer`s channels of customers who have shown similar preferences.
"Early personalization received a black eye because retailers didn`t integrate merchandise displays, and personalization was more around individual touch points, the web site or call center," says Tom Johnson, managing director in the CRM solutions practice at BearingPoint. "The next step of personalization is to take the experience end to end." That means gathering customer shopping interests as expressed in each of a retailer`s selling channels, then responding in each channel with merchandise displays and marketing pitches that recognize a shopper`s interests by brand and groups of products, he and others say.
Don`t forget the rules
In addition, effective personalization requires retailers to apply, on top of information on consumer shopping behavior and demographics, business rules that support their retailing goals--pushing high-margin products to consumers most likely to buy them, for example, or pushing sales of certain products in a way that will smooth out levels of inventory.
Effective personalization also requires merchandise and marketing expertise as well as personalization technology, which must be supported by adequate levels of data storage and analysis capabilities, experts say. With all that underlying support, they add, a personalization strategy can be designed to reach consumers in more effective ways than simply greeting them by name on a slightly personalized web page. "Retailers are just now beginning to combine a customer`s personal information and shopping history, affinity across products, and retail business rules, helping them to better understand customers and motivate them to buy," says Craig Stevenson, IBM`s worldwide marketing manager for WebSphere Commerce B2C and multi-channel retailing solutions.