Database Reference
In-Depth Information
Gutierrez:
What data did you work with when you joined Nordstrom?
Shellman:
Our Stylists Suggest was the first project we tackled as a team,
and that dealt primarily with in-store transactional data from our 117 full-line
stores across the country. We started by examining what stylists were selling,
specifically the composition of shopping baskets
Now that we've expanded the recommendations work into a full platform we
work with data from many sources including clickstream and online purchases.
Gutierrez:
How do you mix together the venerable Nordstrom brand with
the innovative data science approach?
Shellman:
It's an ongoing challenge, but we've developed a couple strate-
gies that have been successful. The first is actually a company-wide open-
door policy that invites people to come by and see us, and they do that
quite a lot. Folks from all parts of the business will come and tell us how
they've been approaching a problem and where they think we can help.
Often they've been spinning their wheels, and are interested in our thoughts
on how to approach the problem. Other times people just stop to chat and
use us as a sounding board.
We also work hard to have good relationships with the people who are
ultimately responsible for getting our work in front of customers—primarily
the web team. So we keep in touch with them regularly and let them know
what is going on from our side. We also make sure that when they find bugs
that we respond right away—especially because in the enterprise world,
they're used to having to file a service request and it taking a week for some-
one to respond to them. So when we can respond with, “I just pushed the
change and in about ten minutes it will be live on the site,” they obviously
love that.
Finally, prototyping our products so that internal customers can use them
early on has been crucial for our success. It doesn't even have to be some-
thing fancy. For instance, our recommendations preview tool doesn't have
a particularly interesting visualization, but it's enough to show the result of
our algorithms. Now we can shoot off a URL to internal customers and it
allows them to sit at their desk and understand the behavior of our product
and experiment with it, and provide feedback way before we're talking about
getting it into production. This has been super helpful and has been a really
great way to get people excited about what we're working on. Building and
maintaining those relationships are just like anything else—you always have to
be working on them. Nurturing those relationships are part of the job, and if
you leave them unattended, they might not be there later.
Gutierrez:
How do you interface with stylists in your full-line stores?
Shellman:
Fortunately for us, Nordstrom loves to promote from within, so
a lot of our collaborators used to be on the selling floor or in the stockroom,
