(Chris Devlin) #1




HE PLAN IS TO bring customers
back more regularly in the fu-
ture. One of the company’s most
interesting efforts is called Way-
fair Next. Think of it as an R&D lab focused
on replicating or even improving upon the
terrestrial home goods retail experience in a
digital forum.
The lab has been working with virtual
reality and augmented reality for a few
years, and during my visit, the head of
Wayfair Next, Shrenik Sadalgi, lets me try
an app on the Magic Leap AR platform
that makes it possible to pick and place
virtual furniture into an actual room. A
user can capture the pattern of her own rug
or wallpaper and insert it digitally next to
a Wayfair couch to see how they jell. They
demo a custom-built scanner that enables
them to create 3D-printed models of furni-
ture and a haptics application that makes
the screen of a tablet feel like the texture
of fabric. These gizmos are medium-to-
longer-term bets. They could turn out to be
the equivalent of concept cars that never
make it to market—or they may be the keys
to building loyalty among millennials as

increased spending. “We’re taking a discipline that is traditionally
quite subjective and applying a lot of math and science,” says VP of
marketing Bob Sherwin, who oversees the media-buying, measure-
ment, and ad-tech teams as well as 130 engineers and data scien-
tists. It comes as little surprise that he has no marketing experience.
He’s ex-McKinsey.
Wayfair’s marketing team has created a proprietary bidding
algorithm for paid search and a machine-learning program to
predict the value of approximately 20 million keywords. They built
a dynamic retargeting platform to analyze browsing history and
advise how Wayfair ads should follow you across the Internet, and
it’s being repurposed to improve product recommendations. “One
of the things we’ve learned is that leveraging information about you
is always a good thing. We see improvement every time we do it,”
Sherwin says. “But our ML platform will also determine the ideal set
of products to show any new visitor. Historically, that kind of thing
was done manually, based on intuition, but that wasn’t scalable.”
When it comes to media buying, the team uses custom algorithms
to scour for “arbitrage opportunities.” For example, they noticed that
rates drop every January, when media-buying agencies—they sur-
mised—are in contract negotiations with clients, so that’s when they
swoop in to get more value. “We do media buying like a portfolio
manager makes stock-buying decisions,” he says. “When the market
is behaving irrationally, we don’t chase the bids.”
Daniel McCarthy isn’t sold on any of this. He’s an assistant profes-
sor of marketing at Emory University and a coauthor of a widely cit-
ed paper, “Customer-Based Corporate Valuation for Publicly Traded


Non-Contractual Firms.” It explores various
methods for gauging customer long-term value
and scrutinizes customer acquisition costs
(CAC) of a few companies, including Wayfair.
The thesis is complex, but the takeaway is
bearish. In short, he says Wayfair spends way
too much and that things are only getting
worse. “CAC rose to new highs this quarter,
about $88 to bring in every new customer,” he
says. “They’ve been making these investments
for four years, and their margins have actually
been deteriorating. The last quarter was the
worst in their history.” McCarthy also questions
Wayfair’s ability to replicate its U.S. model in
From Sherwin to Shah, the Wayfair
response is to scoff at the simplicity of CAC
metrics. “The calculation assumes all of our
marketing dollars go against new customers,
but more of our ad spend is going to existing
customers that are loyal to us,” Sherwin says.
What’s more, customer behavior doesn’t track
neatly to quarterly earnings. While home-
furnishings purchases can be highly sporadic,
he points out that the average customer made
1.85 purchases over the past 12 months.

FRIENDS FROM COLLEGE: Shah and Conine back in the really early days—as
undergrads who bonded over data at Cornell University.