Bringing Data Science into the Organization

By on

In a nutshell, Data Science accelerates business growth. Sudeep Rao, offers solid evidence of this by stating that worldwide nearly $30 billion is invested annually for artificial intelligence (AI) and machine learning (ML) projects. A BI and analytics survey indicated that 94% of the survey participants reported “data and analytics” as important factors for the digital transformation of their businesses.

Industry insiders have offered these arguments in favor of transforming every business into a data-driven organization:

  • With data insights and data-powered competitive intelligence, businesses are able to make far superior decisions.
  • Product and service-related data in many forms — customer data, product logs, product review data, competitor product data, market analysis data, technology analysis data — help to create better products and services.
  • Operational and transactional data collected from everyday business functions help to make a business more efficient and cost-conscious.
  • Advanced data analysis, like predictive analytics, help to take advantage of market intelligence to arrive at accurate predictions or projections for the future.
  • Data-enabled tools also help to measure the success or failure of data-driven business decisions.
  • From the career point of view, the field of Data Science offers the best work and remuneration opportunities in today’s job market.

 So whichever way you choose to evaluate the role of data in business, data comes out the winner. Using AI and Data Science to Drive Worker Productivity  explains how Data Science enhances productivity in businesses, even when employees work remotely.

Common Barriers to Bringing Data Science into Organizations

According to Forbes, data and data technologies often function in isolation, distancing Data Management teams from mainstream business activities. A Deloitte survey indicates that 63% of executives of large organizations feel that their organizations lack a data-centric infrastructure, and very often, business teams have little connection with Data Management teams. The following factors have often been described as the primary barriers to introducing Data Science into an organization:

  • Lack of executive focus
  • Lack of overall organizational technology strategy
  • The cost involved in developing and maintaining in-house data teams
  • Data security and privacy issues
  • Siloed data across business functions (no centralized control)
  • Absence of talent, which of course, can be mitigated with Data Science

A recent blog post throws light on problems facing data-minded businesses who do not know where to begin. Another article cites a communication gap as the biggest stumbling block to achieving the desired business objectives through data. In many of these cases, in spite of having access to advanced BI tools, the business users use Excel or other spreadsheets to report results.

However, as the majority of the industry literature indicates, developing Data Literacy programs in organizations is probably the right approach to inculcate a “data culture,” so that all business users, irrespective of their job role or designation, starts solving problems with data.

Building Data Literacy in the Organization

In this excellent MIT paper, Professor Catherine D’Ignazio and research scientist Rahul Bhargava describe Data Literacy as the ability to “read data, work with data, analyze data, and argue with data.”

The typical approach in a medium or large organization is to dedicate Data Science teams to the most complicated and pressing business problems. Theoretically, this approach sounds right, but in reality, a more holistic approach, involving multi-function team members, usually yields better results. Here are some ways to democratize Data Science in the enterprise:

  • Develop excellent Data Literacy programs to improve communication and knowledge sharing between the Data Science personnel and other business users. When everyone starts speaking the same data language, it becomes much easier to solve business problems through data-driven insights.
  • Identify the core business risks or opportunities that can be resolved through data-enabled solutions.
  • Build working teams with one or more “citizen data scientists” in them. This should automatically happen if Data Literacy programs have been conducted successfully throughout the enterprise.
  • Distribute the limited number of data scientists across the developed teams, so that their talents are best used for the most strategic business needs.

The HRB authors cited in the above link use interesting analogies from the field of sports (NBA) to illustrate this point. While it makes excellent business sense to invest the best Data Science talents in strategic operations, it is equally prudent to nurture and develop Data Science throughout the organization — across functions and job levels to maximize the benefits of grassroots Data Science and collaborative decision-making.

Gartner projected that by 2020, 80% of organizations would implement Data Literacy in their organizations to take full advantage of Data Science in business.

Getting Your Organization Ready for Data Science

Though large organizations have been able to build their data centers and Data Science team to solve their business problems with advanced data analysis, the communication gap between the “techies” and the “non-techies” (the business users) has created a huge roadblock to achieving the expected results.

In many cases, the business users, while superior domain experts, have failed to communicate their business goals and expectations in the language of data — thus hindering the technical experts’ ability to realize the full business outcomes from the available data assets.

These measures can help an organization prepare for data-driven business outcomes across functions:

  • C-Suite Buy-In: All business decisions, technical or otherwise, start at the top. So, the executives and other business leaders running the organizations will first have to be convinced about the importance of preparing their businesses for data-driven decision-making.

Developing an interest for advanced technologies and understanding the role of such technologies in everyday business decisions is probably the best way to orient business staff toward data-centric business models.

  • Data Literacy Programs: The only way to close that undesirable communication gap between the Data Science teams and their peers is to endorse and implement strong Data Literacy programs.
  • Required IT Setup: Modern technologies such as the cloud, IoT, AI, ML, edge computing, and a lot of other related technologies have now made it possible for organizations, both large and small, to access data-enabled platforms and services.

Large organizations usually have data centers and Data Science teams, but even they are increasingly combining on-premise with hybrid cloud or multi-cloud architectures to extend the capabilities of their existing IT infrastructure. Moreover, medium and small businesses, who do not have proper IT resources in terms of infrastructure, personnel, or budget, can now buy service from cloud-based service providers at minimal cost.

The question here is for organizations that lack in-house data centers — can they build Data Literacy programs to make their business staff data literate? Maybe yes, as most training or educational programs are also available on hosted platforms.

  • Building Data Culture: This is perhaps a natural outcome of successful Data Literacy programs. After completing adequate training in Data Literacy, all business staff, regardless of department or designation, can be expected to communicate smoothly with highly technical staff such as data scientists, data engineers, or data architects. The communication or collaboration is one aspect; the other aspect of data culture is teaching every business user to solve day-to-day problems with data.

Building Data Science Teams in Organizations

President of the EDM Council — a global organization promoting DS standards, best practices, and training — John Bottega, said, “To some degree, building and maintaining a strong Data Science team is an art.”

The head of the enterprise Data Science team is sometimes the Chief Data Officer or CDO, a role that pioneered with financial services firm Capital One in 2002. A 2020 survey conducted by the data and analytics consultancy firm New Vantage Partners, indicated that 65% of its surveyed companies (85 large ones) had CDOs.

A strong and effective Data Science team will typically include data scientists — experts in mathematics, programming, statistics, and data mining; data engineers — experts from computer science or software backgrounds with special knowledge of data infrastructure issues like data acquisition, storage, and preparation; data architects — the overall “visionary” who builds the blueprint for data systems and data architecture designs; and finally the data analyst — who collects and maintains data, uses tools to interpret data, prepares reports for end users and frees up time for busy data scientists.

In the recent times, even AI and machine learning scientists and engineers are joining DS teams to strengthen the work of data scientists by offering training models.

The true Data Science teams also welcome business analysts, data visualizers, and citizen data scientists (the business user turned data expert), who play active roles in the everyday data-driven organization.

Here are additional pieces of advice for enterprises planning to build successful Data Science teams:

  • According to TechTarget, the best team members will demonstrate a mix of technical and business skills.
  • Fostering a culture of explorative learning is the best way to bring out the talent of team members while solving routine business problems.
  • Developing projects that promote collaboration between technical and business experts will enhance team spirit and collaborative problem-solving.
  • Mentoring programs will help the juniors move faster, and routine training camps may help all team members stay abreast of current technologies.
  • Developing “talent management” programs may help retain the best data scientists.

Data Science Best Practices

Here are some best practices usually adopted by successful data-driven businesses:

  • Focus on Data Quality to provide the best opportunity for AI and machine learning models to train on data.
  • Accurate labeling or tagging of data: Very often, unclear or inaccurate data labels hinder a training model’s performance and can potentially lead to inaccurate results.
  • Enterprise, humans-first Data Governance (DG) must be in place to facilitate defined Data Management roles, responsibilities, channels of communication, and controls for DG.
  • Advanced data security technologies monitor and manage data threats and security issues in real time.

Since the phenomenal growth of startups is clearly visible across the globe, no discussion about introducing Data Science in organizations can be complete without a mention of the startups.

Inducting Data Science into Startup Businesses

An Analytics India magazine article provides guidance to startups for building a Data Science team. This post aptly uncovers the current challenges of bringing Data Science into a startup business.

According to the post, regardless of sector, the factors that will help shape the Data Science setup are the maturity of the business and the current involvement of AI in product development.

Maturity of the startup is often determined by the operating budget, with future funding possibilities, current revenue, growth prospects, and users. These factors will help determine exactly at which growth stage the startup is at.

As most businesses today are already data centric, the next step of evaluating their data-friendliness can be determined by the AI involvement throughout their product lifecycles. The AI-involved startups can further be categorized as early stage or late (mature) stage.

The last category, according to the article, is the non-AI startups, which again, include both the early stage and late-stage businesses.

The approaches to introduce Data Science into their business models will be distinct for each of the above categories.

Image used under license from

Leave a Reply