E-Commerce Innovations That Work

These days, return on investment is everything. It has to be big—and it has to come soon. To survive the economic slump, e-commerce companies have tightened technology budgets. Launching a sophisticated commerce site is costly enough—upwards of $52 million, according to Forrester Research. Sinking cash into flashy new technology is no longer a minor gamble: It could mean the difference between staying afloat and closing up shop.

Corporate technology spending has dipped from 11 percent last year to 7 percent this year, according to investment bank Adams, Harkness & Hill. Unproven technologies like mobile e-commerce showed up low on the priority list for the 300 businesses the firm surveyed.

Unlike last year, when companies invested in infrastructure and external systems, the current wave of tech spending has shifted toward technologies focused on keeping the best customers. This includes everything from analyzing and acting on customer preferences and installing 3D technology that keeps shoppers engaged to implementing fraud prevention solutions that sniff out crime before it happens.

Far from drying up, the online commerce market stands to snowball in the next three years: IDC predicts the worldwide market in 2004 for business-to-business and business-to-consumer e-commerce will total $3.1 trillion, up from $350.4 billion last year. There’s still lots of money to be made. Here’s how a handful of smart companies are using new technologies to get a bigger share of that spending.

Get Inside Your Customers’ Heads

Why do two-thirds of online shoppers click away in the middle of a purchase? Answer: Most companies don’t really know their customers.

Companies pay lip service to personalizing the online shopping experience, but very few use customer metrics or mine their data properly—if at all. A recent Jupiter Media Metrix report on customer loyalty uncovered some startling results. A majority of the companies surveyed—71 percent—don’t analyze customer drop-off rates or find out why shoppers leave after filling their carts. And while 60 percent buy plain-vanilla customer demographics from third parties, only 10 percent use predictive technologies to anticipate buying behavior.

On the other hand, Jupiter found that actively identifying loyal customers—and catering to their needs—can reduce acquisition costs by 27 percent and increase average order sizes by 60 percent. To that end, some forward-looking e-commerce shops are starting to use predictive modeling.

The technology goes beyond simply culling through HTML log files. “A site with 10 million visitors a month can generate 5GB of [log-file] information per day,” says David Daniels, a Jupiter analyst who specializes in e-commerce. “The raw data is often too hard to get a grip on, and needs someone dedicated full-time to trying to make sense of it while melding it with third-party demographic information. Log files follow where you click at the basic level, like how many people went to the Products vs. Services page. They don’t tell you how customers interact within [those pages], and they don’t get into target marketing.”

Find Out What They Want

New methods of data tagging, which bug items on a page with code, allow e-commerce companies to not only identify which pages, products, or microsites customers flock to, but also how long they spend on each page or with each product, especially when there are multiple products on a page. While tagging requires more time and planning at the outset—you must decide what gets tagged, for example—it paints a much more complete picture in real time.

Take Expedia.com. With $1.8 billion in travel bookings last year—second only to Travelocity, according to PhoCusWright, an online travel market research firm—the folks at Expedia understand the importance of keeping up with customers’ demands.

Before it launched an internal customer relationship management effort a year ago, the site tracked only how many tickets it sold. That changed after Expedia added a predictive modeling solution from data analysis firm SPSS called Clementine, which starts at $55,000. By tracking purchase habits and other variables, Clementine revealed that travel purchases follow certain cycles, and that customers expect more support during and after purchasing trip packages. “We’ve got a much clearer picture of when we expect people to call and what they will call about,” says Jon Zimmerman, Expedia’s director of IT and applied research.

Through retention forecasting, Expedia then discovered that it cost less to acquire and retain customers than it had originally guessed. “We thought customers were leaving and not coming back,” Zimmerman explains. Expedia learned that customers it thought it had lost after a single purchase were actually returning to the site several months later.

The system also shed light on why 44 percent of Expedia shoppers abandoned itineraries without buying. Click-stream data that Clementine tracks showed that browsers skipped off the site in the least likely place: the hotel directory. Customers got lost there and gave up. So Expedia redesigned the directory to make it easier to navigate. “We had a fairly quick payoff and it improved the customer experience,” Zimmerman says.

Brand Power

Clinching more sales online is not the only reason to add a customer-data–analysis solution to your site.

Since its launch in July 2000, Crayola.com’s main goal has been to reach its core customers—parents, kids, and teachers—and bolster its brand. “If we got everyone coming back twice a month to download content, we’d be happy,” says Justin Knecht, manager of Internet technology at Binney & Smith, maker of Crayola products. If visitors buy a pack of crayons or markers while they’re at it, that’s an added bonus.

The site logs about 10 million page views per month, Knecht says. And 80 percent of registered users opt to receive Crayola’s biweekly e-mail newsletter. Still, says Knecht, “we didn’t have the sense that we were actually meeting our consumers’ needs.”

Realizing it could learn a lot about Crayola.com’s 800,000 registered users by running market research with customer data updated every 24 hours, in June the company hired digiMine to handle the job. Every day, digiMine sorts through Crayola.com traffic logs, screens the files for specific information, segments customer behavior, and overlays demographic and transaction data.

“The standard log analysis we were using was kind of like throwing darts,” Knecht admits. “It would know that someone was looking in our activity book category, but wouldn’t know what activities within that group were most popular. We wanted to peel back the layers.”

Crayola is already gaining new insights. For example, digiMine’s analysis showed that teachers send more electronic greeting cards from the site than the company realized. Knecht suspects digiMine’s service, which starts at $25,000, will pay for itself in short order. “Considering what we would spend on traditional marketing analysis, the return will be very high,” he says. “DigiMine will allow us to quantify how much it will cost us in conversion per product per point per group.”

Instant Feedback

But what if getting updated customer data every 24 hours isn’t often enough?

That was the issue for MuseumShop.com, which has agreements with museums worldwide—from the Metropolitan Museum of Art to the Louvre—to sell museum gift-shop products and to promote and fund exhibitions and programs.

MuseumShop.com CEO Mitchell Massey didn’t want to wait for detailed information about shopping patterns. Massey’s goal was to test products on the fly to see whether new features he introduced on the site actually drove sales. In June he started using Buystream Merchant, a real-time data-tagging analytics solution for online retailers, starting at $30,000 for licensing or $5,000 a month for the ASP model.

Massey tagged every page of the Web site, including microsites within MuseumShop.com. He even tagged artwork in the 3D virtual museum tour. Log files told him customers stayed on the site longer after taking the tour, but he couldn’t tell if it was driving customers to buy. So he tagged the artwork. “We’ve found we have work to do. We’re not getting enough purchases from [the site],” Massey admits. Based on the results, he plans to redesign the tour and make it easier for users to find the 3D portions of the site.

At the same time, Massey discovered a way to make streaming video pay. Using Buystream’s technology, he tagged videos in the Exhibition Spotlight portion of the site to track how long people were using the educational tool. Once he discovered the videos drove traffic, he wanted to see if they would boost sales. He embedded a gift related to the Exhibition Spotlight in a corner of the pop-up video, as well as a click-to-buy button. It worked. And thanks to Merchant’s data tags, Massey says, “now it’s clear who goes in and how many units are sold.”

Customer Do’s and Don’ts


  • Track drop-off rates to see where and why customers abandon your site.
  • Use predictive modeling—which monitors customer interaction with your site and uses that info to anticipate their behavior.
  • Tag pages, products, or site features to find out what entices customers to buy.



  • Assume massive log-file data will tell you exactly what you need to know about your customers.
  • Rely on generic third-party demographic data to get to know your unique customers.

Online Shopping That’s Better Than the Mall

For some things, nothing beats shopping at a real store. With furniture, for example, you want to sit on it, plump the cushions, open the drawers, and touch the fabric. The Web has long promised a virtual shopping experience that meets or surpasses the live act. Finally, rich media and 3D technologies are starting to deliver on that promise.

Until now, one of the biggest barriers to delivering reality over the Web has been transferring large, high-resolution images to customers over slow Internet connections. Dial-up connections will continue to outnumber broadband Net links this decade. A site that takes forever to download is one of the surest ways to send customers packing. To solve the problem, the companies developing the latest 3D Internet tools have devised compression algorithms that can turn massive files into manageable bytes—without compromising graphic quality.

Last spring Gifts.com, a Reader’s Digest company, tried rich media as a test during its Mother’s Day promotion of a pendant. The goal was to drive more sales of the gift item by capitalizing on people’s feelings toward their moms through a streaming 3D video presentation powered by RichFX, which compresses video 10 to 100 times more than standard techniques.

RichFX, a 3D technology provider that has created virtual tours for MuseumShop.com and Neiman Marcus, sets up licensing agreements ranging from $5,000 to $100,000 a year. The technology uses its own media player (a free downloadable plug-in) or Real Player so consumers can view interactive presentations. The idea is to keep customers on your site longer—four to six times longer, according to RichFX.

In its month-long test of the technology, Gifts.com was looking for a measurable return. It sent two promotions via e-mail to 50,000 customers each: one with a two-dimensional JPEG image of the pendant, the other with a link to the 3D video. The result? For every $1 the site spent on the RichFX promotion, it notched $2.77 in sales, compared to a 1-to-1 ROI for the 2D promotion. What’s more, says Dan McManus, VP of marketing and business development for Gifts.com, the customer conversion rate for the 3D campaign was seven times higher.

Get Ready for 3D

The number of sites that use 3D technology in various forms is expected to grow from a handful this year to more than 1 million by 2007—with more than a half-billion 3D-ready Web browsers on potential customers’ PCs by then, according to Jon Peddie Associates.

Specialty apparel retailer Eddie Bauer is already giving online shoppers an experience that rivals the real thing. Using 3D solutions from Viewpoint, the retailer lets shoppers view certain items as if they were right in front of them. Customers can watch clothing, shoes, furniture, and luggage revolve, zooming close enough to see the grain of the wood in an armoire or the stitching on a sandal in near–life size.

Viewpoint’s technology, which costs from $50,000 to $1 million a year, delivers its 3D effects using the Viewpoint Media Player and its own compression algorithms, letting customers see fine graphic details in real time without hogging a lot of bandwidth. One of Viewpoint’s latest technologies, called ZoomView, breaks the original image into sections. As a customer moves around an image, the software fetches from the server the corresponding pixels within that section, bringing just that area into focus. Previously viewed portions are cached, so the return view is instantaneous.

So far, these shopper-controlled features are working. Brian Walker, Eddie Bauer’s group manager of development services, is tight-lipped about dollar-figure ROI. But he will say that in tracking return customers, the company has found that those who use the 3D features spend more time on the site and more money at checkout. “There are no drawbacks,” Walker says of adding 3D to the site. “Either customers are seeing this stuff and it’s prompting them to spend more, or they are our best customers and are spending time using it.”

Rich media pays off for more than just consumer sites. Using Viewpoint’s technology, Dell Computer shows its corporate customers 3D views of complex server systems. Given Dell’s direct sales model—it sells $18 billion in computer components and services online per year—offering this extra makes selling high-end networking equipment via the Net a lot easier.

“People at trade shows come around the server equipment and say, ‘Open the cover and let me take a look.’ Viewpoint is the next closest thing,” says Kevin Libert, director of marketing and communications for Dell’s Enterprise Systems Group, which generates 20 percent of the PC maker’s revenue and itself saw 42 percent revenue growth last year.

Since adding 3D and other enhancements to its business-to-business site, the division’s customer conversion rate has jumped from 2 percent to 3.5 percent.

Online 3D Do’s and Don’ts


  • Use 3D graphics instead of flat JPEG images for special promotions.
  • Offer video, audio, and 3D graphics that let customers zoom in and out, pan around objects, and change colors.
  • Use compression technologies that keep images sharp but nimble.


  • Use graphics for every single product.
  • Put 3D everywhere customers click. Always test response rates to rich-media and 3D site enhancements.

Hook Them With Web Ads

If you’ve surfed the Web lately, you’ve probably already started to hate them: online “pop-under” ads that lurk on your desktop beneath the browser window. Walking a fine line between effective and obnoxious marketing, advertisers—many of which are online merchants—have searched long and hard for a way to get their message across without turning consumers off. While potentially annoying, pop-up and pop-under ads (known as interstitials) generate lots of traffic, successfully boosting brands, according to DoubleClick.

Web advertisers are starting to spend more on rich-media technologies such as audio, video, Flash, and dynamic HTML. Companies shoveled $8.2 billion into online advertising last year, says the Interactive Advertising Bureau, and less and less of that was spent on static banner ads. Although banners accounted for 47 percent of all spending, advertisers made a notable shift toward rich media and interstitials by the end of the year.

That begs the question: Do flashy new Web ads convert browsers into buyers?

Take X10.com, for example. You can hardly visit a tech-related Web site without encountering an X10 pop-under ad hawking home- and small-business–networking solutions. Between January and May, X10 ads reached 32.8 percent of all Web surfers. But according to Jupiter Media Metrix, 73 percent of the visitors who arrived via pop-unders left X10.com after just 20 seconds.

Yet proponents say the ads make an impact. The effectiveness of pop-under ads is “being confused with the arrogance of X10,” says Michael Tchong, president and CEO of Iconocast, an Internet marketing research newsletter. But X10 is also building its brand by leaps and bounds. “They aren’t getting very good conversion [rates],” he says, “but they don’t care. They are paying only for visitors who come to the site.”

Look! Over Here!

The trick is to get customers to notice your ad without ticking them off. AdReady, a Newark, New Jersey–based Internet advertising agency, has developed a solution called adPointer—a twist on the conventional interstitial. When you pause while browsing the Web, a matchbox-size window appears onscreen near your pointer. The ad, which may relate to the last site you visited, disappears if you ignore it. If you click for more info, the ad stays with you. You can move it aside for later viewing or expand the window, which generally includes pop-up graphics, a video, or other interactive content. Licensing fees come from a revenue-sharing arrangement, typically 20 percent of the media buy.

Iconocast has already seen returns. After testing adPointer on its site for a month, it found that click-through rates rose from 2.5 percent to 5 percent. The average banner ad gets click-throughs of 0.4 percent.

But click-through isn’t everything. Cheap Tickets, a travel booking company, tested adReady’s technology this summer. Scott MacDonell, media director at Target Market Interactive, Cheap Tickets’ online ad agency, liked that adPointers offered “a new way to get in front of customers.”

For Cheap Tickets, however, advertising is only worth the expense if it can boost sales. “As far as going all the way to a ticket sale, it didn’t work out,” MacDonell says of the adPointers. “We got a lot of transfers [clicks]—a heck of a lot of transfers. But after that the numbers fell off significantly.”

But MacDonell says that doesn’t mean his agency won’t consider rich-media technologies for other clients. “It’s just got to work out from the ROI standpoint,” he says. “Sometimes simple is better.”

And of course, simple doesn’t have to mean boring. Since December, Nextel Communications has shifted 85 percent of its online marketing budget—$4 million to $8 million—to a single solution: Superstitials, a patent-pending technology from Unicast. Superstitials are the closest thing to TV commercials on the Net. The key advantage is that Superstitials don’t make consumers wait for their browser to load and buffer the video. Instead, the animated ad loads in the background and doesn’t appear until it’s ready to play. “Ads need to play instantaneously or they won’t work,” says Richard Hopple, CEO and chairman of Unicast, whose 300-plus advertising clients see an average 12 percent purchase rate—that is, 12 percent of the customers who click on a Superstitial make a purchase. Clients, which include American Airlines and Taco Bell, pay between $7,000 and $12,000 per ad for production costs.

Nextel had found that its static online banner ads performed three to four times worse than direct mail; Superstitials did three to five times better. Says Larry Everling, Nextel’s director of Web sales, “The worst thing an ad can be is forgettable.”

Advertising Do’s and Don’ts


  • Mix digital marketing strategies such as dynamic Web ads with e-mail campaigns.
  • Measure click-through and conversion.
  • Decide which is more important: boosting your brand or getting buyers.


  • Assume that traffic and eyeballs equal sales.
  • Annoy potential customers with gratuitous rich-media campaigns or pop-up ads every time they click.

How It Works: E-Fraud Busters

Online payment fraud worldwide could account for $15.5 billion in losses by 2005, up from $9 billion this year, according to Meridien Research. FraudShield GeoLocator, a new module in ClearCommerce’s popular payment verification system, tags potentially fraudulent purchases. ClearCommerce’s payment system starts at $85,000. GeoLocator costs $7,500. Here’s how it works.

  • A customer places an order on the Web storefront, supplying billing info, shipping address, and credit card number.
  • The ClearCommerce Engine checks to ensure it’s a valid credit card. Then GeoLocator determines the country where the order originated.
  • GeoLocator flags mismatched billing and shipping info on orders that come in from countries known to harbor fraudsters, such as Indonesia and Yugoslavia.
  • The merchant puts a hold on suspect orders and attempts to contact the customer and/or the issuing bank. If the order is bad, details go into a database to block future purchases.

Spy on the Other Guys

Think it’s the domain of Dumpster-diving mercenaries? Compete.com serves up the competition’s site performance info instantly. It’s clean—and legal.

Tracking customers’ behavior at your e-store is one thing. A snapshot of their habits at competing sites gives you even more to go on.

Compete.com blows open your competitors’ doors for this exact purpose. The service provider licenses clickstream data from ISPs to find out what more than 8 million active Internet users visiting 200,000-plus sites worldwide are up to. Subscribers pay $75,000 to $100,000 for a peek at the data.

For application service provider Intranets.com, the service offers two key advantages: a way to track the effectiveness of its own marketing campaigns and, by extension, a way to see what works—and what doesn’t—for its rivals.

After sinking lots of money into TV and radio advertising, Intranets.com needed to slash its customer acquisition costs. Compete.com’s analysis revealed that Intranets.com’s pricey TV and radio campaigns weren’t working. “We brought our cost of acquisition down from $50 in the first quarter of 2000 to $3 [today],” says Rick Faulk, Intranets.com VP of marketing.

Compete.com also shows Faulk select traffic information for competitors, which he applies to his own company’s marketing and customer outreach strategy. “I can understand where [competitors’] traffic comes from and where they are doing their marketing,” he says. “I can effectively tell the performance on a Web site based on where their traffic is. And I get pretty close to figuring out their conversion rate.”

By: Maria Atanasov

Leave a Reply