News Stay informed about the latest enterprise technology news and product updates.

Booz Allen Hamilton's innovation blueprint

A peek at Booz Allen's innovation practice, what Etsy looks for in a data analyst and mathematical proof that vampires don't exist: The Data Mill reports.

Booz Allen Hamilton, a 100-year-old management consulting firm, decided that it needed to innovate or "we would not survive," said Karen Dahut at the recent Chief Innovation Summit in New York City. "We were great consultative problem solvers, and yet we had become more interested in the inspection of our current business instead of the introspection of the next big thing."

Today, Dahut is leading Booz Allen's innovation practice as the executive vice president of the recently created strategic innovation group. If big companies can't innovate because, as Dahut said, "they are not designed to innovate," maybe it's time to draw up a new innovation blueprint. Here are a few excerpts from her talk:

1. Room to fail. Bigger companies tend to value efficiency, but innovation needs room to grow. "We must give talented staff the freedom to experiment, to play, to have fun, to fail, and to try again," Dahut said. "We've been relentlessly focused on this."

Convincing her team to accept failure as a necessary part of innovation was not an easy feat, Dahut said. To help get over that hurdle, she created (and laminated) get-out-of-jail-free cards, which she handed out to her leadership team members with the instructions they try something new and, if they failed, to bring her one of the cards -- no questions asked. "It told them I had their back. Go experiment, try something, do something different, and I will cover for you," she said.

She also reinforced the cultural values of her innovation blueprint through programs that focused on key skills development, such as how to sell an idea concisely. "A five-minute pitch, not a 45-minute PowerPoint deck," she said.

2. Booz Allen Garage. Booz Allen is a global business, and enabling better collaboration for employees who don't work on the same campus is a challenge. Dahut decided to focus on Booz Allen's "internal ecosystem," which she defined as virtual tools that help source ideas, solve challenges and reuse code, data and data sources. It also includes virtual and in-person events to bring people together for "chance hallway encounters" and brainstorming sessions, which tend to be a natural part of startups.

An example of a new collaboration space is the Booz Allen Garage. "It's a mock[-up] of the Apple garage or whatever garage any startup may have started in, but it is a virtual tool," Dahut said. There, Booz Allen hosts competitions and posts challenges for anyone to work on. When a team based in Washington D.C. faced a client problem it struggled to solve and posted it in the garage, a team from Seattle created a fix that became one of five "selected by our client to be funded and to be taken to the next level," she said.

3. Partner network. Dahut encouraged attendees to consider the "broader global ecosystem" when innovating. "Companies that believe the best ideas only reside in their brick and mortar will not be successful," she said.

Booz Allen, for example, partnered with Intel, Microsoft and Allscripts Healthcare Solutions Inc., an electronic health records company, to develop Allscripts Wand, a mobile application built for Windows 8 devices.

What Etsy looks for in a data analyst

Need an updated job description for data analyst? At the recent Strata + Hadoop World in New York City, Nellwyn Thomas, director of analytics at Etsy Inc., the online vintage and handmade crafts marketplace, laid out what she looks for in an analyst beyond domain expertise. The four skills are as follows:

1. Analytical skills. The ability to understand the business problem and the opportunity, including how to break it down into logical steps, understand the goal and articulate the goal when others can't.

2. Math and statistics skills. These skills are especially necessary now, "when the scale of the data we're working with is so huge," Thomas said. It is vital for analysts to know what techniques to use and to understand what they're looking at, she said.

3. Technical skills. Etsy hasn't "stabilized into a place where everyone uses the same systems," Thomas said. So the company strives to ensure the analyst is as close to the data as possible. "The more layers of tooling added between an analyst and their data, the more likely you are to have misunderstandings about what the data is," she said. That's why analysts write in Scalding, a MapReduce programming language, directly onto the Hadoop stack to pull out the data directly. "That might change over time when we have tooling we have more trust in, but for right now, the immediacy between the outputs and the analysts are important," she said.

4. Communication. Thomas said it's important to identify meaningful patterns or nuggets in the 150 gigabytes of weblogs collected daily that will help the business make better decisions. It's also important to be able to communicate those findings clearly, as well as filter out what isn't important, she said.

Vampires and geometric progression

Speaking of using mathematical and statistical skills to understand what's out there … physicists Costas Efthimiou and Sohang Gandhi refuted the existence of vampires in their 2007 paper Cinema Fiction vs Physics Reality: Ghosts, Vampires and Zombies by employing the basic mathematical principles of geometric progression. The research was recounted at Strata + Hadoop World by Doug Bryan, data science services practice lead at the marketing firmRichRelevance, to explain how the term exponential growth is often misunderstood.

Efthimiou and Gandhi assume the first vampire appeared in 1600 A.D., when the global population was 536,897,911 (though they state that the date "has little bearing on our argument"). If vampires feed once a month (a conservative estimate), thereby turning humans into vampires who never die, then the progression will end up looking like this: in January, one vampire feeds, one human dies; in February, two vampires feed, two humans die; in March, four humans feed, four humans die. And so on.

Thirty months in, the human population would hit zero. "Apparently, whoever devised the vampire legend failed his college algebra and philosophy courses," the authors conclude. That, or we're all vampires now.

Welcome to The Data Mill, a weekly column devoted to all things data. Heard something newsy (or gossipy)? Email me or find me on Twitter at @TT_Nicole.

Next Steps

Previously on The Data Mill

What is your company's 'innovation identity?'

What's the secret to a disruptor's success? Question everything

What the Hortonworks IPO should signal to CIOs

Dig Deeper on Enterprise business intelligence software and big data