Pete Fishman, Director of Analytics at Yammer, discussed analytics and providing cheap insights that businesses can act on to make the best decisions with theCUBE co-hosts Dave Vellante and John Furrier, live at the HP Vertica Big Data Conference.

Yammer aims to  have “as big as an impact on people’s working lives as possible.” As Fishman said, “we think that social is going to be a big part of the it going forward, and in enterprise social, Yammer is a big offering.”

When one is in control of their entire pipeline, they are able to make the decisions that make most sense for their business, Fishman explains. As an analytics team, Yammer wants to “touch the entire business, go through it horizontally.” The aim is to pull in data from CRM platforms, data from marketing platforms, as “you have a lot of the real value once you start to combine a few different sources.”

Cheap, Deep Insights

“A lot of our philosophy is around cheapness, we want to deliver insights cheaply. We don’t go too deep into a lot of our models, our visualizations, this is what gets you the answer in a sort of simplistic fashion,” Fishman said. Yammer’s tool chain is all about its workflow, how one can use the centralized analytics team to deliver insights.

A lot of the current consumer development is data driven development, Fishman pointed out. Developers should “figure out what your customers are actually doing, don’t fantasize about it. Figure out how they are using your product, what features are working or not within your product.” If they can marginally improve on that, and “essentially spend fewer cycles on what’s not working,” companies gain a competitive advantage in the long run. As an example, Fishman used poker where pros remember the hands they walked away from, and did not lose much money on it, while amateurs mostly remember winning hands.

“Our job is to make A/B testing as cheap as possible,” Fishman added,  and “making distributed analytics very, very cheap.” To ensure that, first Yammer needs to make sure anyone can do it by building tools that enable anyone at a company to assess how effective a feature is. “You don’t need a team of statisticians to tell you if something is statistically relevant.”

Commenting on their use of Vertica and how it is incredible for companies that grow up very fast, Fishman said that when they had first become a Vertica customer, they were asked to predict how much they would need in a year, after starting with a half a terrabyte license. The increase was “an order of magnitude over my ridiculously high estimate.” Data is flowing in from a number of different sources across the company within a platform where Yammer’s data analysts and scientists work with it, acting as data architects. As they work closely with customers, they know what columns they are going to be needing.