ComScore follows 1.5 million web users to make sense of web numbers—and concludes that conversions, sales and average order are a lot more important than simply visits.
To Gian M. Fulgoni, chairman and co-founder of Chicago-based comScore Networks, the future of retailing on the Internet rests to a large extent on properly measuring the current trends in online shopping. Whether the issue is gaining continued support of senior retail management, extracting additional funding from venture capitalists, or mapping marketing strategies by understanding the competitive dynamics of the market, it is now clear to Fulgoni that the e-retailing industry needs better tools of measuring industry trends in online purchasing.
“The e-retailing business is at a point where it must prove its return on investment, and that puts a lot of pressure on the metrics,” Fulgoni told an audience last week at the E-Tail 2001 Conference in New York City. “We’ve moved from a world of where we measured eyeballs and traffic to a world where we have to measure who’s buying what, and the metrics have fallen down.”
Convinced that traditional electronic measurements of web traffic (visits and page impressions) and offline surveys of consumers produce incomplete and misleading data on actual online purchasing trends, Fulgoni helped to create comScore to obtain more accurate shopping data by electronically tracking every Internet move made by a large sample of Internet users. ComScore has signed up an impressive 1.5 million web shoppers for its electronic survey panel, agreeing to greatly enhance their Internet download speeds in exchange for the right to monitor all of their browsing activity and web purchasing transactions. ComScore does that by routing each panel member’s browser to one of its more than 250 servers around the globe, without replacing existing ISP or corporate network connections. It also uses a technique that leaves no software on office computers that might prove disturbing to corporate MIS technicians.
ComScore has been electronically monitoring every Internet move of its 1.5 million participants since last fall, and the data it has captured-summary versions of which Fulgoni presented to conference attendees-question conventional wisdom and open new windows on Internet shopping patterns. ComScore’s basic finding, noted Fulgoni, is that traffic to retail web sites is much higher than has been reported by other syndicated rating services based on consumer interviews. “Traffic to retail web sites has been grossly underestimated,” Fulgoni declared.
To make the point, Fulgoni noted how other syndicated rating services late last year had measured the monthly unique visitor count for Yahoo.com at about 100 million. ComScore pegged Yahoo’s traffic during the same period at 177 million unique visits by projecting the traffic of its sample group to the market at large. In a company press release distributed early this year, Yahoo revealed that its monthly volume was 181 million unique visits, or very near to the comScore number and nearly twice the number projected by other services.
Part of the reason for the understatement of other surveys, Fulgoni asserted, is that they fail to properly count the international traffic coming to U.S. web sites. “Fifty-six percent of English-speaking computer users are located outside the U.S., and that international traffic to U.S. web sites is not properly counted on a country-by-country basis,” he insisted.
This international traffic, Fulgoni said, heavily favors the retail web sites of the pure-play competitors and those with strong international brands, when compared to other competing retail web sites in the same market. For example, comScore data reveal that international visits to Beauty.com, Revlon.com and Drugstore.com account for 38%, 26% and 19% respectively of their total traffic. By comparison, Fulgoni said, CVS.com and Walgreens.com get 17% and 6% of their traffic from outside the U.S.
Other comScore findings reported by Fulgoni at E-Tail 2001 include:
Shopping From Work: Retail web site traffic originating from work-based computers accounts for 34% of visits compared to 58% from computers at home and 8% from school-based systems. Yet, sites that feature business-related services depend more heavily on work-based computers for their traffic. For example, comScore’s data show that Travelocity.com, Dell.com and UPS.com get 42%, 48% and 53%, respectively, of their traffic from computers at work.
Conversion Rates Vary Dramatically: “Valuing retail web sites based on the number of eyeballs they attract can be very misleading,” Fulgoni declared, “because that method was based on an advertising model, not a retailing model.”
In a recent month measured by comScore, Amazon.com had 27 million unique visits compared to 6.4 million for Kmart Corp.’s Bluelight.com, 5.8 million for Barnesandnoble.com, 3.6 million for Sears.com and 2.7 million for 1800flowers.com. But that traffic, according to the comScore data, did not translate into purchases on anything remotely resembling a proportionate basis. The data show that Amazon.com converted a healthy 6.3% of those visitors into buyers, while 1800flowers.com achieved a stunning conversion rate of 10% and Barnesandnoble.com recorded a respectable 4.7% conversion. Meanwhile, Sears.com achieved only a 1.9% conversion rate, and Bluelight.com converted a paltry 0.6% of visitors into buyers, likely a factor that led the chain to pull its stand-alone retail web business back into the corporate fold.
Other conversion rates reported by Fulgoni using comScore data include FTD.com at 10.5%, Spiegel.com at 7.2%, Cooking.com at 1.4%, Borders.com at 2.8%, Bedbathandbeyond.com at 2.4%, Bloomingdales.com at 1.9%, Potterybarn.com at 3.6%, Neimanmarcus.com at 5.3%, Williams-sonoma.com at 2% and Talbots.com at 9.8%.
Transaction Sizes: Transaction sizes also vary dramatically from one retail site to the next, but here the variance has as much to do with the merchandising mix as it does with the stickiness of the site. The fact that Neimanmarcus.com’s average order is measured by comScore at $187, compared to $107 for Bloomingdales.com, clearly reflects the former’s greater affinity to the carriage trade and high-priced designer fashions. Nonetheless, merchants with similar product lines can have very different average order sizes, which in turn can affect the relative profitability of their web businesses. Bedbathandbeyond.com, according to comScore data, rings up $84 of merchandise in an average order, compared to $44 for Williams-sonoma.com., even though the conversion rates of the two sites are comparable.