beyond hits

`Hits’ and `clicks’ are giving way to more sophisticated Website performance measurement tools

Just a year or so ago, most Websites measured their viability by “hits” – the number of times visitors request a particular file within a website. Today, however, most online retailers recognize that a large number of hits, particularly on sites with large, complicated designs, don’t always indicate a large number of excited visitors. The current industry joke, in fact, is that “hits” really means “how idiots track success.”

To measure Web success these days, “you really want to Measuring site performance has gone far beyond tracking just hits and clicks. Find out more about the new measurement tools savvy marketers are using to gauge their Web success close the loop between site traffic and e-commerce transactions,” says Paul Baudisch, a Needham, MA-based general manager with Circle.com, a unit of Snyder Communications that develops electronic customer relationship management solutions. For instance, a growing number of online retailers now look for measures that let them know how their sites perform in terms of the bottom line. Besides tracking who comes to their sites, these e-tailers are recording the promotions or ads that prompted visits, figuring out which visitors have potential to become customers, and determining which products draw the most attention.

For instance, shortly after Christmas, six staffers at Foofoo.com, an online purveyor of such luxury items as fine chocolate and trips to Bora Bora, locked themselves in a room. For three days, they not only counted up the holiday orders, but also pored over figures on page views, customer traffic patterns, shopping-cart loads, search engine information, and any other piece of data that measured performance during the crucial season. “We analyzed just about every report we could,” says Philip Hawken, director of operations with the Arlington, VA-based company.

For catalogers and direct marketers, this drive for research should sound familiar. “The direct marketing audience is among the savviest in terms of target marketing,” says Rich Coyle, spokesperson with Nielsen Media Research. Just as direct marketers have used source codes, page analysis, and customer research to measure a mailing’s effectiveness, so are electronic merchants now tackling the issue of who they’re reaching, and how well they’re reaching them. Moreover, e-tailers can call on measurement tools print catalogers never dreamed of – discovering, in effect, how long customers visit a site, what they’re looking for, and whether they’re likely to return.

Tools of the trade

For most online marketers, page views have now replaced hits as a basic first Website performance measurement. By calculating the number of pages visitors download over a given period, a marketer can “tell how much traffic the site is getting,” says Eric Richard, chief technology officer with net.Genesis, a Cambridge, MA-based provider of e-commerce intelligence solutions. “It’s a very high-level gauge of overall interest.” Views of product pages, moreover, indicate an active buying interest. Clearly, the more visitors look at merchandise, the more likely they are to make a purchase.

Similarly, the length of time visitors remain on a site (known as “stickiness”) also measures Website health. Typically, longer visits mean higher sales. Keeping a prospect on a Website for a while only works, however, if he or she keeps moving toward a purchase decision. In some cases, visitors may actually be hanging around for a while simply because they can’t find their way around the site.

In their holiday post-mortem, for instance, the employees at Foofoo.com found that their Website didn’t prove quite “sticky” enough in terms of getting visitors to buy. By looking at the number of product page views, the percentage of abandoned shopping carts, and the number of returning visitors, the company learned that half its visitors had left its site without viewing any products.

To remedy this problem, the staff decided to merge its previously distinct editorial and shopping areas. By distributing the product pictures and information throughout the site, including on the home page, Foofoo.com ensured that every product was no more than three clicks away from any page.

As a result, says Hawken, 75% to 80% of users now see at least one product page. “We want people to come to the Website and view multiple product pages every visit,” he says. While it’s too early to see any concrete results, Hawken says he does expect the Website redesign to boost sales.

Cataloger and online jewelry/tabletop retailer Ross-Simons, on the other hand, found that its site was too sticky for sales, according to Peter Howard, vice president of marketing for the Cranston, RI-based company. By tracking the paths visitors took through its site and watching exit patterns, the company learned that some customers, frustrated by the site’s navigation, abandoned shopping carts at checkout.

With this new information in hand, the company cleaned up its order page to make it easier and quicker to navigate. This summer, says Howard, Ross-Simons will improve the site’s editing function, so customers can easily delete an item or add gift-wrapping. The Website will also upsell customers by suggesting complementary items.

Efforts so far appear to be working. Ross-Simons’ Website revenue will hit $20 million this year, up from about $8 million last year, Howard says.

How’d they get here?

To gauge Website productivity, meanwhile, Foofoo.com and other e-tailers have also begun measuring the breakdown between their unique and repeat site visitors.

Like other marketers, Web sellers need to attract new visitors for top-line sales growth, while retaining repeat customers for bottom-line profits. “The more people come back to the site, the more likely they are to buy,” says Foofoo’s Hawken.

To identify and track the site’s visitors, Foofoo.com uses net.Analysis, an enterprise-level e-commerce intelligence application from net.Genesis. By slicing and dicing visitor data, net.Analysis lets merchants ascertain how often customers return to Websites, and what paths they take once they’re there. Installing the application, according to net.Genesis’s Richard, typically costs about $100,000, although some smaller e-merchants start at $30,000.

Tools like net.Analysis also indicate what brings visitors to a site in the first place. By measuring new-visitor sources, online retailers can find out which types of advertising, portals, links, or affiliate programs deliver tire-kickers, and which deliver potential buyers.

During the holidays, for instance, net.Analysis revealed that three types of messages on banner ads – last-minute gift ideas, free shipping, and “Foofoo for the holidays” – produced 10% to 30% more click-throughs than the company’s standard banner message, “For the fun and finer things in life.” (Foofoo ran the traffic-building ads on USAToday.com during the shopping season.) According to the net.Analysis results, “general branding messages didn’t work very well,” Hawken says.

Web experts caution, though, that click-throughs, like hits, have their drawbacks. As a traditional measure of advertising effectiveness, “click-through rate” refers to the percentage of people who click through to an advertiser’s Website after viewing an online ad. The problem, however, is that it doesn’t indicate how many and which people actually buy something.

Although a catchy banner ad may attract clickers, “there’s a big difference between an effective ad and a popular one,” says David Zinman, vice president of marketing with AdKnowledge Inc., a Palo Alto developer of software that analyzes advertising effectiveness.

Take Thomas Register’s experience, for example. The New York-based company, which maintains an online guide to manufacturing products and services, found that its visitor quality could vary dramatically depending on which banner ad visitors had come from. Using the AdKnowledge service to track click-throughs, the company discovered that only some of its ads attract a high percentage of visitors who register (and thus are likely to become customers).

Because of the discrepancy in advertising effectiveness, Thomas Register’s cost-per-registrant can vary between $10 and $150, says Susan Orr, Thomas Register’s director of Internet marketing.

Beyond advertising measurements, Web merchants also track visitors from affiliate links (Foofoo.com has links with 70 e-merchants, including Amazon.com), as well as from search engines. Even search-engine words can reveal crucial customer behavior.

For example, Foofoo.com – which receives about 15% of its traffic from search engines – found during the holiday that an unexpected number of those visitors had searched using the keywords “luxury travel.” To get those people to stick around, the marketer beefed up the travel section of its site.

Ratings services

There are even better Web measurement tools out there. For example, there are services that compare one online site to another, but they are still in their infancy. Like television rating services, these Web rating services measure site popularity by examining behavior from a sample group of users. For example, Nielsen\Net Ratings (a product of New York City-based Nielsen Media Research and Milpitas, CA-based Net Ratings, Inc.) tracks online behavior from randomly chosen Web users, indicating the relative popularity of different Websites among the overall Web population. Another firm, Media Metrix, Inc., in New York, provides a similar service.

“We’re seeing activity through the eyes of the user,” says Tim Meadows, senior vice president of marketing with Net Ratings, Inc. Both services charge between $25,000 to $100,000 per year, depending on the client’s number of reports, with typical fees running between $50,000 and $60,000 annually.

Some Web marketers, though – particularly in business-to-business – don’t always find the results of such site ratings meaningful. B-to-b marketer HardwareStreet.com, for example, says that 20% of its visitors convert to customers, according to its own figures. By contrast, however, one ratings service estimated such a small number of visitors to the site that HardwareStreet’s conversion rate would have been closer to 60%, says Barbara Coenson, vice president of marketing for the Reno, NV, firm. “We’re finding more people on our site than the services would suggest,” she says.

For all their new measurement tools, e-merchants still lag behind their offsite counterparts in some key research areas. Any TV producer, for instance, knows that a 20 Nielsen rating is a hit show – but there’s no such benchmark for knowing what makes a hit Website. “It’s hard to come up with standards that are meaningful,” says consultant Jim Sterne, with Target Marketing in Santa Barbara.

Even stickiness is hard to compare. An office supplies site, for instance, may measure success by getting customers in and out quickly. An online gift store, on the other hand, might want to encourage more browsing.

Whatever the Web lacks in benchmarks, though, it makes up for in information. All e-merchants today can generate an extraordinary amount of data – from hits to click-throughs to shopping cart usage – that they can review and analyze to improve their Websites, and ultimately, their bottom lines.

“In everything we do, whether it costs a little or a lot, we have to look at our return on investment,” says Coenson of HardwareStreet.com.

Ever see something in a store window you had to have, just as the store was closing? That’s the same frustration many online customers have. No matter how much they may want a product, if they’re having trouble getting to a site or downloading pages, they’re not likely to stick around long enough to make a purchase.

Measuring a site’s technical performance, however, is not an inside job. To learn where problems are, you need to analyze your site from your users’ perspective. Otherwise, you’re likely to miss problems that crop up outside your own system, such as damaged connections between your hosting site and an Internet Service Provider (ISP) in another part of the country.

Companies like Service Metrics (Boulder, CO) and Keynote Systems, Inc., (San Mateo, CA) spot performance problems much as a customer would. “We get the performance view from the end-user’s perspective,” says Bob Merlo, Service Metrics’ vice president of marketing.

By connecting computing devices to major ISPs throughout the world, these companies emulate flesh-and-blood Web users.

Once agents log on to a Website, they can measure page downloads and then run through and time a mock transaction, such as buying a book. Client companies can choose to measure performance at their own sites, as well as at their competitors’.

Both Keynote and Service Metrics charge monthly fees, which vary by the number of cities from which a Website is measured, and the frequency with which agents log onto the sites. Typically, client companies spend around $500 per month for each Web page measured.