Why Organizations Are Struggling to Operationalize Data Science

The problems are, as the article notes, organizational and interpersonal rather than technological.

Too many tools, purchased by too many different units within the organization - and with not enough unified strategy. JL

Bernard Marr comments in Forbes:

Many companies are still struggling with operationalizing data science. Among challenges is “tool sprawl” – meaning that a company has many different methods of tackling a problem; (for instance) using 6.7 tools just to do analytics. Many of these tools might encompass multiple languages, multiple data warehouses, and multiple systems which don’t necessarily work very well together, (and) lack a unified approach to putting them to work.
In light of research which has shown that many businesses are still not getting value from their data science and analytics, it’s worth looking at some of the factors which differentiate those that are.
This is the focus of new research from Forrester published this week, which highlights how three different groups – tagged as “Insight leaders”, “The pack” and “Laggards” differ to their approach to collecting, storing and analyzing data, and their techniques for drawing value from it.
Forrester questioned 208 respondents in business leadership, data science and engineering arms of US organizations, ranging in size from 1,000 to 20,000-plus employees. By asking them questions about aspects of their data operations such as the different sources of data used and the structure of their data science teams, they were categorized according to the maturity of their analytical efforts.
Headline findings include the fact that those slotting into the insight leader category were four times more likely to create revenue growth exceeding shareholder expectations, and twice as likely to be market leaders within their industry.
Ian Swanson, CEO of DataScience, which commissioned the research, told me “What we are seeing in many industries are these small, agile disruptors that are changing the landscape.
“Uber, AirBnB - these companies that are much larger are tied up with legacy solutions, and can’t move as quickly as their more nimble competitors they are trying to keep up with.
“What was shocking was as we interviewed and surveyed over 200 companies all of them said we’re investing in data science. In many cases, they are doubling or tripling-down, and that’s exciting.
However among the pack and laggards group is evidence that many companies are still struggling to get to grips with operationalizing data science.
By “operationalizing”, this means putting it to work in a way that is achieving results – generally growing a company and generating revenue, or providing insights about their customers or prospects that can be leveraged.
“That’s one of the fundamental gaps right now in many businesses – they want to use their data but a lot of companies are investing in their data science teams and not seeing that value – not seeing the output,” Swanson tells me.
Among the other challenges highlighted by the report is “tool sprawl” – meaning that due to investment a company has many different methods of tackling a problem – probably including expensive software solutions – but lacking a unified interface or approach to putting them to work. In fact, 46% of respondents were found to lack a unified approach to their Data Science technology stack.
DataScience’s solution – in common with competitors such as Palantir - which operates most visibly in the governmental sector – is to offer a “platform-based” approach to data science.
It’s an approach which is certainly growing more and more popular. According the Forrester research, 88% of companies within the insight leader segment take a platform-driven approach – utilizing some form of data science Swiss army knife to avoid having too many disparate, disconnected solutions.
Statistics recorded in the survey show that companies are using an average of 6.7 tools just to do analytics. “Many of these tools might encompass multiple languages, multiple data warehouses, and multiple systems which don’t necessarily work very well together”, DataScience chief strategy officer William Merchan tells me.
The results are certainly exciting, and show that predictions made in recent years regarding the sharp increase in of adoption of analytics and data science were accurate. As more and more companies begin to get to grips with their own data initiatives, and take a platform-oriented approach to embedding analytics throughout their processes – from marketing to manufacturing distribution and customer care – the number bearing fruit is likely to increase, too.