I thought it was a pretty interesting article, even though it only scratched the surface (hate those underresearched articles, grrr). And never mind Mr. Holloway's statistics, "20 to 50% are returned" - he might as well have said "0-100%" - wouldn't make a big difference.
E-Commerce Report
No Substitute for Getting Personal, if You Want the Perfect Fit
For all its innovations, the
Internet has yet to crack what is, for many, one of life’s most vexing
problems: how to find jeans that won’t make your backside look like a
tractor-trailer.
A new online business, Zafu.com, believes that it has made progress on that front. Unlike, say, Amazon
— which analyzes a visitor’s browsing and buying behavior and
recommends merchandise bought by others with similar behavior — Zafu’s
approach relies on users to do a little of the work.
On the site, which is basically a search engine for clothes,
visitors click through a questionnaire of about a dozen items, after
which Zafu determines the visitor’s body type and displays what it
believes are the best-fitting jeans to suit that visitor (it offers
only female styles for now). Each pair is modeled from several angles,
along with a link to the product page of retailers selling the item.
The company, which introduced its Web site in August, can already
point to a rapidly growing base of customers and merchant partners as
evidence of popularity. The company’s early success underscores the
industry’s slow but steady progress in personalization — finding ways
to match customers with their stated or implied product preferences,
and thereby satisfy what analysts say is a central consumer need.
“Online shoppers are control freaks, and the tools they like the
best give them the ability to customize something and do product
comparisons,” said Lauren Freedman, president of the E-Tailing Group,
an Internet consulting firm. “So I definitely see consumer appeal in
what Zafu is doing.”
Robert Holloway, Zafu’s chief executive, said the service had been
about two years in the making. Starting early last year the company
recruited women to come to its office in Emeryville, Calif., to test
its recommendation system, and try on jeans.
“At first, the accuracy was really low, about 50 percent,” as at
other apparel sites in general, Mr. Holloway said. About 20 percent to
50 percent of all jeans bought online, he added, are returned. “Slowly
but surely, we got it to the point where 94 percent of the women who
went through our process said the jeans fit them great.”
Since Zafu.com made its debut, Mr. Holloway said, the site’s traffic
has grown rapidly, to more than 100,000 visitors this month, with
virtually no marketing. The company makes money by earning a commission
of 5 percent to 15 percent on every pair of jeans sold on the hundreds
of retail sites with which it has agreements.
Ms. Freedman said one drawback of Zafu’s service was that it did not
weigh heavily enough a user’s brand preference. “I got a few hip
choices, but it also returned me some brands I wouldn’t buy even if
they fit me,” she said. “The label, for a lot of women, is as big a
factor as the fit. But it’s still a really good service.”
Elsewhere in the e-commerce industry, personalization technologies
are growing quickly in importance, analysts and executives said, as
retailers try to counter the effects of rising marketing costs by
squeezing more sales from first-time visitors and more repeat sales
from longtime customers.
Overstock.com,
whose quarterly earnings reported earlier this month fell far short of
projections, is turning to two types of personalization technologies to
help visitors find merchandise more easily from among the more than
700,000 items on the site.
According to Patrick M. Byrne, Overstock’s chief executive, the site
last week rolled out a “gift finder” service, in which visitors receive
gift recommendations determined by their answers to several questions,
including the age and gender of the gift recipient, and how the user
describes the recipient’s lifestyle.
The technology behind the service, which is provided by
ChoiceStream, a Web site personalization company in Cambridge, Mass.,
essentially assigns a series of attributes to millions of items
according to their brands, styles and prices, among many factors. The
more the service is used, the more accurately it can predict which of
Overstock’s items will appeal to people with a certain personality.
Mr. Byrne said Overstock also began using another service last
month, Aggregate Knowledge in San Mateo, Calif., that works in a manner
similar to Amazon’s recommendation system. Starting this week, he said,
the system would recommend products to visitors according to their
buying history on the site, or their browsing history there (which
Overstock tracks). In both cases, a user will be shown items that
others with similar browsing or purchasing behavior have considered.
Mr. Byrne said the site’s personalization effort was “one of many”
he was working on to improve the company’s performance. “I have a lot
of hopes for this,” he said.
Martha Rogers, a partner with the Peppers & Rogers Group, a
consulting firm, said, “We’re seeing more sites starting to do this
sort of thing, but it’s been frightfully slow.”
The number of technology vendors offering the service, including
specialized companies like ChoiceStream, Aggregate Knowledge and Sento,
as well as larger e-commerce software companies like ATG Inc., is
growing. And even though it is not particularly expensive or difficult
to deploy personalization technology, Ms. Rogers said, many retailers
“are stuck in the mindset where it’s just more efficient to treat
everyone the same way.”
Dennis R. Hernreich, chief operating officer of the Casual Male Retail Group,
which sells clothing sized for big and tall customers, is planning to
change the retailer’s site into one that is personalized to the tastes
of individual users. At the moment, the company uses a fairly common
approach: sending out e-mail with recommendations gleaned from a
customer’s past buying or browsing behavior.
That e-mail essentially leads the user to a customized Web page.
Next year, though, Mr. Hernreich said Casual Male would rely on
technology from ATG to guide each of the site’s visitors “right to what
you’d be most interested in.”
That effort, he said, is similar to one in Casual Male’s retail
stores, which offer merchandise according to the various lifestyles
prevalent in a particular community.
“Whenever we put relevant offerings in our stores, it’s considerably
improved the sales of that store,” he said. “This is no different,
really. We’re just adding a bit more technology to it.”