Data Science — The Opportunity for Enterprises
Opinion
Enterprises have numerous problems that can be solved for cheap, and at scale, using Data Science. The opportunity is greater in larger companies, as there are plenty of areas to constantly improve on, and challenges apart, data can be scrubbed and made available — unlike in the case of many smaller companies and startups, where data is simple not there.
Why do enterprises have such a tough time embracing data science?
In times of rapid change, experience could be your worst enemy. — J. Paul Getty
Many companies fail to embrace data science as a core function. Here are some reasons from my experience:
They don’t believe in the Breadth of solutions that data science can tackle
Enterprises are notorious for their silos. Departments like to keep their best and brightest inside their functions, hide their problems, and work on them. They also believe that other teams don’t understand and cannot solve their problems. In the absence of a higher authority / leadership, horizontal problem solving teams don’t get formed. Data science may be one of the most intrusive teams in a sense, because they get to see the department’s data at a very granular level…a terrifying thought for many.
They don’t believe in the Depth of solutions that data science can create
Enterprises have traditionally struggled with implementing IT solutions. Implementations have been long drawn, involved many expensive failures, and custom projects have at times turned into technical nightmares. In the face of such history, they do not believe that they can put together applications that can rival those provided by enterprise vendors. For example, the CTO may find it difficult to imagine how an internal team is going to build and deliver a forecasting system better than one from an Oracle or an SAP. The new reality, thanks to advances in cloud computing and open source software, is that a dedicated data science team can arrive at excellent solutions across many areas, for the organization.
Leadership simply doesn’t know how to go about it
Leadership may simply not have a good approach for going about data science that works on the ground. Many a times, lofty goals get created and the show stops there instead of ending up with a well-defined strategy and roadmap of projects. When leadership is confused and yet expects results, projects often get divided up amongst functions, and chopped into multiple pieces that get allocated to IT or outsourced to consulting companies. Formation of a data science team and working on the foundations required to enable them take a back seat.
What’s The Way Out?
Leadership
All indications are that data analytics will lead to a major transformation in the way business is done. Leadership should enable and empower data science teams to succeed. A good data science strategy, proper place in the organization structure for the data science team, and roadmap for data science are solely the responsibilities of leadership. If the necessary talent doesn’t exist in-house, there are consulting companies and data science leaders that can help start your journey. However, leadership must come from within.
Internal Data Science Team
Many enterprises are wary of creating data science teams. Instead, they prefer to rely solely on external parties e.g. consultants, software vendors, etc. to undertake such tasks. While this approach may give quick results and may be a good complementary approach, we strongly believe that every company must build its own internal data science team.
Demonstrate Value Quickly
Data Science teams need to demonstrate real value to the rest of the organization quickly. To do this, they must be enabled with the right team composition and the right tools, and be self-reliant to a great degree. Just as going from ideas to prototypes fast is important, equally important is going fast from Prototypes to Production. Business functions like solutions fast, and like to repeatedly iterate upon them as they learn. The mantra for data science teams must be ship data products fast.
Embedded into Business
Data science teams need to be embedded into business teams, and not act like a centralized think tank group or as an extension of IT. This helps data scientists to focus on real business problems, quickly prototype ideas, and demonstrate solutions to potential users. Good data science teams, e.g. those at companies like AirBnB, Uber, etc. have the ability to not just demonstrate solutions, but to get prototype solutions directly into the hands of business end-users. Nothing delights business users like interacting with working prototypes quickly after solution discussions.
In our next post, we pick up the thread from here and discuss how to accelerate data science for teams.
Author: Ananth Krishnamoorthy
Please do write to us with your views, comments and data science stories. In case you a company/startup looking for help with machine learning, we’d be happy to help. Just drop us a line and we’ll get back.
