Data Science: Making something out of everything
August 19, 2013 | Alicia Androich | Comments
This story was updated at 15:50 on Aug. 19, 2013
Data scientists can tell you everything from which customer you’re about to lose to how to best price your products. Using their unique and coveted skill set, they deftly mine golden nuggets from reams of data. A look at who they are and how they work their data-driven magic to help marketers
Much more than just a URL-shortener, Bitly keeps close tabs on what people all over the world are reading and sharing in the social sphere. Hilary Mason, recently named scientist emeritus at the company after nearly four years as its chief scientist, is a self-professed data nerd who was named one of Fast Company’s “100 Most Creative People in Business” this year. Mason went to Brown University and Grinnell College for computer science, and spent part of her early career as a software engineer. Most recently, she has joined venture capital firm Accel Partners as its data scientist in residence. Soon before moving to Accel, she spoke with Marketing to explain Bitly’s appeal to brands and why big data will never replace creativity.
On today’s big data buzz
There are a couple of reasons big data is popping up on most conference agendas these days, says Mason. While the work data scientists do with large data sets has been possible for ages, there’s been a massive reduction in the cost and an increase in the usability of the tools. The engineering aspect or analytics side of making data useful used to be professions in themselves (and in some cases still are) because that work was so complex, but Mason says it’s now become so much easier and more affordable that a data scientist can do the engineering, analytics and mathematics. Data scientists also do what Mason says is the most important part, which is either have the domain knowledge to make a good decision or the communication ability to sit down with their clients “and help them make better decisions based on data.”
Her time at Bitly
Mason joined Bitly in 2009 in the days when its whole team fit in one elevator. In her first year with the New York-based company her title was “engineer,” but she insisted they put “scientist” on her business card. “I thought it sounded cooler,” she says. As the company grew—there are now roughly 55 people there, including interns—so did Mason’s job. She progressed from being a hands-on coder to her chief scientist role, which saw her lead a team of six data scientists and had her working closely with the CEO and the board “to try and make decisions about where the business and products can actually use the data most effectively.”
Bitly in a sound bite
“Bitly is the set of links people are sharing with each other on social networks, so it’s basically a huge database of gossip,” says Mason. She jokes that people that joined her team of data scientists went through an emotional cycle that started with excitement about how cool the data is, “then you just get really depressed because you realize that so much human attention is going to Kim Kardashian and Justin Bieber and some sports guy who scored a point.” Next, though, comes the pivotal a-ha moment when Mason says the data scientists “realize that for the first time, we’re able to actually study human communication at the scale at which that communication is happening.” The Bitly team sees data from hundreds of millions of people every day and Mason says, “I still find that completely amazing.”
Finding who is flying under the radar
A lot of brands will tell the team at Bitly about who their customers are, but the company’s data scientists sometimes make unexpected discoveries once they look at their own data. “Most brand marketers are really smart about what they do, but sometimes we’ve been able to find little pockets of communities that weren’t really big enough to hit their surveys or their radar that are also engaging with their product.” After speaking to a brand that sells microwaveable breakfast food, for example, Mason says her team found a segment of single men in Texas that were into the product.
She’s hopeful that, in general, the data that’s becoming available will give marketers better information upfront for making decisions about who they market to.
A product that shows what the world cares about
Bitly’s Realtime, an attention-ranking engine, didn’t go through the typical product development process. Mason says it stemmed from a research goal—“Can we build a system that knows in real-time what people all over the world are paying attention to from the data we see?”—and grew into an application programming interface (API). “As nerds and engineers, we built it as an API and went around trying to talk about ‘Oh, we made this cool thing’ and nobody except other engineers thought it was interesting.”
So Mason’s team built a human-friendly interface – the one that’s found at rt.ly – that’s got a few different systems at work. On any topic, users can complete a query and get all the links ranked by popularity at that moment about that topic.
Or you can also ask to be shown, say, all the links about food being read in Brooklyn. “That’s one we do all the time,” says Mason, “because that finds us all the stories that our system thinks are related to food that are being consumed disproportionately by hipsters in Brooklyn, which means they probably have good taste and we find all kinds of fun stuff that way.”
Another query option is similar to Twitter’s “Trending Topics,” though it’s based on what people read, not what they say. Those curious about what’s happening right now can see the stories receiving disproportionate attention at any given moment. They’re not filtered by language or content—they’re simply the stories getting an unexpected burst in the rate of clicks per second, says Mason. Brands can use Realtime to find out which articles about themselves are currently popular across the social web, and also which articles are popular on specific domains, whether their own or those of a media outlet, for example.
They’re also using Realtime to find things to share with their audience, adds Mason. So if a brand wanted to identify with people who are into motorcycles or soccer, they can just monitor stories about those things and find good stuff to share.
Big data will never replace humans
Out-of-the-box sentiment analysis software or survey research counts the data and can provide numbers or graphs, but Mason says that data only represents a portion of the real world—not the whole thing. A data scientist’s job, she says, is to take the data that’s been counted then contextualize it, come up with a theory about why it appears to be the way it is, test that theory, and try to make better decisions.
“There is a lot of hype around how big data will replace creativity and, I know this isn’t the technical term, but I think it’s bullshit.” Although she hears lots of fear from people that data “is either going to replace them or somehow prove that they’re not actually good at their job,” she doesn’t think it’s justified. “The data should be a tool to empower people to be better at what they’re doing professionally, and I can tell you as someone who looks at data all the time that it is not going to replace creativity; it’s not going to replace a human connection.” As Mason sums up: “Data can tell you whether A or B is a better choice, but it can never tell you what A or B should have been in the first place.”
And if you want up-to-the-minute news coverage of marketing, media and PR news, subscribe to our twice-a-day free e-newsletters.
Update: The print version of this story identifies Mason as Bitly’s chief scientist. We have updated the story to reflect her most recent title at the company, as well as her new role at Accel Partners.