If your products and services don't serve the data science community, but you're using big data analytics in your offerings for a competitive advantage, you're in a challenging situation with customers.
I call this using data science as a supporting strategy. For instance, Graze incorporates data science into its snack business to develop and deliver the next box of goodies their customer will get. But Graze is in the snack business, not the data science business. In this type of situation, I recommend keeping your data scientists as far away from customers as possible. Below are strategies for leaders using data science in this way.
In short, buffering is structuring at least one organizational layer between your data science team and your customers. Contrast this to leaders using data science as a core strategy — selling products and services to other data scientists, like RapidMiner, Kitenga (now part of Dell), and Cloudera; in that case, it's a great idea to put your data science team in front of customers, because like attracts like.
Let's assume Graze's clients have no interest in data science, so in a case like this, keep the analytics out of the conversation and have customers interface with other people in your company who are like them. Or, if you're in the business of wearables for athletic people, put a layer of athletic-minded people between customers and your data science team. A good friend of mine is a triathlete that runs analytics to help other triathletes compete. Although he's an analytic, he wears his triathlete persona when addressing customers; he's a one-man shop, so that's his only choice. In a larger company, this concept should have a sales and marketing layer comprised of athletes, not engineers.
An important job of the buffering organization is to translate what the data scientists are trying to accomplish into terms customers understand. The reason why you don't put data scientists in front of non-analytics is that they're typically difficult to relate to. Imagine a group of pro football players showing up at Comic-Con; the first time a Trekkie introduces himself to a linebacker in Klingon, there will likely be a problem. Before a product or service is introduced to customers, it must be sanitized from its analytic underpinnings.
For example, when Progressive talks to clients about its Snapshot device, there's no discussion about analytics. The company's marketing department may allude to the scientific prowess that goes into its product for effect; however, in practice they call it usage-based insurance because most drivers understand that term. You'll quickly lose those customers if you start talking about behavior-based digital profiling using a synthesis of regression and machine-learning algorithms.
It may take multiple layers within the organization to successfully translate your analytic-based competitive advantage into a customer-facing language; I've worked with organizations where the developers are three or four levels removed from the customer. When I worked with Visa, there was a product development group, a product function group, a business analyst group, and then developers and architects. Sometimes it takes multiple translations to get it right for the customer.
A special challenge the buffering organization has with its analytic brain trust is information overload; curating sifts through the piles of brilliance to extricate the golden nuggets that will appeal to customers. That's no easy feat.
Consider a museum curator whose job is to process archeological findings into a display of wonderment. Piles and piles of ancient bones, tools, and artifacts must be organized and displayed in a way that appeals to the masses. Curators do more than just translate — they manage and oversee their body of work, and interact with the viewing public.
In a similar fashion, your curators must own the body of work produced by your data science team. Whether or not you put your curators in direct contact with customers (both ways work), they should synthesize the wealth of information produced by your data scientists into a concise, attractive package that's relatable for customers. Even if you translate well, if you don't curate, you'll hit your target market with too much information and they'll find a competitor that's easier to understand.
If you incorporate fancy analytics into your products but customers aren't really jazzed by math and science, save the tech-speak for your in-house design team. As you structure your organization, ensure a buffer between your data science team and customers that can translate and curate their findings.
If I'm a Graze customer, I don't want a lecture on how to design the perfect meal — I just want a snack.
- Data visualization: When it's the wrong tool for the job
- Data visualizations need to present conclusions rather than artistic masterpieces
- Don't tune out user feedback when measuring data scientists' effectiveness
- Mohammad Sabah: Data chief. Product builder. Cricketer.
John Weathington is President and CEO of Excellent Management Systems, Inc., a management consultancy that helps executives turn chaotic information into profitable wisdom.