17 Sep Becoming an Analytics-Driven Organization: 3 Simple Activities That Organizations Can Undertake to Start That Journey of Transformation

As we’re building TenPoint7, I’ve been engaged lately in many conversations with organizations and consulting firms alike on how TenPoint7 can help with their advanced analytics initiatives.

What often strikes me is hearing that most organizations believe in the power of data science yet often they get stumped creating a sustainable analytics culture: a culture that utilizes data to support most of their critical decision making; from product feature selection and better targeted marketing campaigns to broader organizational performance measures. The vision and strategy is often well articulated at the top by leadership but execution and implementation is either a non-starter or early initiative attempts are failing to meet initial expectations.

Unfortunately, we’ve all heard this play out many times before.

That classic quote from Jim Collins in Built to Last comes to mind: “Building a visionary company requires one percent vision and 99 percent alignment.”

So what then can good-intentioned organizations do to generate inertia toward building such a culture?

While culture transformation and recalibration is no trivial undertaking, there are some small yet impactful activities that organizations could evaluate to make that push toward being data driven.

Prior to co-founding TenPoint7, I product managed a social (media) analytics product that required strong collaboration across product, engineering, data science and business (such as sales) teams. There were insightful lessons that I learned through some of the practices we adopted that I think could be equally applied elsewhere.

Here are 3 such activities:

Building Prototypes:
When feasible, building a prototype model is often an effective method that not only helps secure buy-in (on value) and overcome doubts but also helps test out the operational readiness of the organization. This is already a fairly common practice at organizations that have some degree of data science adoption. Back at my previous company, the data science team would often build prototypes addressing a specific product feature such as topic modeling, ensemble methods to improve sentiment analysis, improving search relevancy of posts and other features. These prototypes would then be demonstrated to product and engineering teams.

Those demo sessions would often spur good discussions sometimes forcing us to re-examine the value of the product features being demonstrated and also helped upfront to identify implementation constraints that otherwise would only have been identified much later in the development lifecycle, and therefore proving more costly.

Key lesson here is that organizations need not only think big but they need to get the ball rolling by starting small. The prototype does not need to be perfect as long as it can demonstrate a quick success or win. Prototypes are an effective and safe method to accomplish this.

Embracing Failure:
Of the ever-growing list of critical skill sets needed to make a good data scientist, one of them is a sense of curiosity. Data Scientists are often encouraged to be self-empowered to conduct investigations and undertake projects that could lead to insightful discovery for the business.

Yet commonly this is mostly lip service given by organizations.

Most have a low tolerance of failure. And who can blame them? Business these days are conducted at the speed of … business.

What often ends up happening is the premature end of exploratory data science projects because of perceived lack of business value in a timely fashion. While analytic methods and statistical algorithms have been around for a long time, their application on large data sets and new use cases within functions, such as marketing and finance, is still relatively new.

To get to the state of being a data powered organization, it therefore is imperative that organizations truly and genuinely foster a culture of curiosity and risk taking that not only applies to their Data Science team but eventually to the broader organization as well.

With that being said, it is also equally important that lessons are learned from failures and learned quickly. Fail if you have to, but fail fast, learn, … and move on.

Abolishing the exclusive “club” perception:
While the promise of data science is compelling and being realized, it is still immensely intimidating to most within the organization. Data scientists are often looked upon as wizards or data sorcerers, and given the growing scarcity of this talent pool most organizations are willing to give them (discrete) preferential treatment.

While this might be one approach to retaining such hard-to-find talent, it nevertheless runs counter to an organization’s ambition of democratizing analytics and transforming to be a data driven enterprise. Organizations should at all costs avoid the perception that only a select few employees belong to this elite “club”.

There are few approaches worth considering that can avoid generating such perceptions.

One such approach is providing multiple learning opportunities on data science and what data scientists do. Back at my old company, the Data Science team would organize monthly brown bag learning sessions hosted over lunch. Popular data science topics and industry/research trends would be discussed with demos that would attempt to show business value and relevant outcomes applicable to our line of business. Invitations for these sessions were extended to all functions in our division.

Another effective approach we utilized was embracing a multi-disciplinary approach to the application of data science in our division. Sales, Customer Success, Marketing, and of course, Engineering, would all participate in product roadmap and feature discussions where the Data Science team would regularly explain and demo potential product features via prototypes. I realized that bringing together such functional diversity not only helped me in becoming a better product manager but on a broader scale further helped other functions to better understand the role of data science, and the business value that it delivers.

For organizations either just getting started with advanced analytics or have some existing degree of adoption, these 3 simple approaches are worth trying to further accelerate the journey to being data driven. This is a viagra natural receta fairly new journey for most organizations thereby posing inevitable challenges along the way. Whether these challenges are people, process or technology related, the good news is that they can all be addressed. The key, however, is getting started with simple approaches as listed here.

Good luck.

Shane Rai (Co-Founder, shane@TenPoint7.com)