Database Reference
In-Depth Information
Gutierrez:
Sounds like you have the math behind it and are figuring out how
to map it to real-world physical attributes.
Smith:
Correct. We have to be able to express what these variables represent
in communication with others. It's all about how we would be able to commu-
nicate users' taste variables so that the rest of the company would understand
that this is actually a good, accurate representation of someone's taste.
On a personal note, I'm really excited by the idea of finding more women like
me—almost like my body and taste doppelgängers. My new idea is that we
should be able to provide that because, once we can do that, it's nice to know
that there are other women out there who face the same issues with finding
the right dress.
Because the community is so nice and supportive, some of our customers are
very comfortable and expose personal details in their reviews. Things like “the
dress fit me oddly because I have implants” are helpful to the set of customers
that face the same issues. So being able to surface the right reviews for cus-
tomers to read that are based on people with similar bodies is a great thing.
I dream of customers being able to say, “Okay, I trust their input. They say it's
not too short, so it will be okay.” It's the whole collaborative system getting
to the right thing, being able to share with other people, and being really open
without any judgments.
Gutierrez:
Given the collaborative supportive community, is there a social
network in the product that is waiting to be teased out?
Smith:
That's one of the things other people have alluded to when I've given
talks. We haven't set it up that way, but I think that it could be a next step if we
wanted to make it even more interactive. We just launched Shortlist, which is
where you can put together a collection of dresses that you like that have a
theme. You can then talk about the collection with others, and this provides a
little more back and forth between people.
That said, even now, with just the reviews that people leave after wearing a
dress, our community of women are doing that. You can write a review, and
then someone can like it. Then you can go back and comment on it and go,
“Oh, that looked amazing. You look fabulous.” And they can respond back,
which is great since it's disconnected from any known social network so peo-
ple don't actually know who these people are. All they know of them is just
the picture they've seen of them. So it's nice how friendly everyone is to each
other while being anonymous.
Gutierrez:
As opposed to other web companies, there's a very personal
feedback mechanism that a lot of other data scientists aren't going to get.
Smith:
Exactly. Since I've worked here, though, I've become more of a power
user, so sometimes it's hard to imagine how the product and how we've
evolved is affecting people out there. We are constantly asking ourselves
