Over the last few years, enterprises have recognized that their future and success are going to depend vastly on their data and AI capabilities. Leaders have started to explore different opportunities in their journey to create companywide, data-driven business transformation and thus build a proper strategy.

As early as 2017, a Harvard Business Review article by Davenport advised executives that they need not one but two distinct data strategies.

On one hand, there is the defensive data strategy that focuses on safeguarding the company’s data. It recognizes that under the pressures of heightened cybersecurity threats and an ever-increasing regulatory oversight, organizations must develop strong data security and be able to respond to compliance requirements. However, in 2022 only 39.7% of large organizations manage data as an enterprise asset, according to a study performed by NewVenture Partners. At any rate, a heavily regulated industry such as financial or medical services will be inclined to introduce many elements of the defensive strategy.

On the other end of the spectrum stands the offensive data strategy which focuses on flexibility and aims to unlock value and drive innovation through data. This approach is usually much heavier on analytics, modeling, and visualization with data that is “good enough”. The data modelers and the executives take the lead, while the data engineers and the compliance managers merely support them. Usually, businesses leverage the benefits of an analytic project through the offensive data strategy, which is why less regulated industries such as retail focus heavily on this approach.

In reality, enterprises need both of these strategies. Depending on the competitive environment, level of regulation, consumer expectations, and internal capabilities and culture, they lean more towards one.

Building up Data Science

Steps for executing a successful data strategy

No matter which data strategy they choose to embrace, companies usually go through several typical steps, as suggested by a TDWI report.

Step 1: Identify key business drivers for data science

The question here is simple: Do you need data science? Most businesses have already developed, to varying degrees, business intelligence, analytics, and data warehousing. How would data science come into play?

To answer this question, business leaders need to examine two key areas:

  1. What organizational challenges are stopping us from being more competitive, effective, and proactive?
  2. How accurate is our understanding of the factors that affect client behavior or the impact of key trends?

An honest analysis here often reveals knowledge gaps that cannot be filled with traditional business intelligence and data warehousing systems. 

They can, however, be filled by a strong data science strategy in very tangible ways: personalization and computational efficiency in marketing and advertising; establishing a dynamic pricing strategy across multiple channels; autonomic analysis of important documents or images such as call center logs or checks. 

The appeal of these benefits is great but data science is not a magic wand. The entire process starts with an honest inward look at your organization and developing a data strategy that takes into account both strengths and weaknesses.

Step 2: Build a great team for achieving data science goals

Successful data science cannot exist without the rich expertise of the people behind it. It's not just about technical knowledge and having a naturally analytical mind. It's about business acumen and knowing how to deliver tangible value. It's also about creativity and curiosity, a spirit of experimentation. A data scientist is like an archetypal scientist: excited to explore the crossing point between theory and practice. 

How can you hire people who meet these criteria? How do you ensure your data science teams are effective and sustainable?

Of course, investing in the recruitment and training of top talent is key. But so is focusing on building diversified teams. While some individuals might be more technical and others more business-oriented, when they join their forces they become a stronger whole. 

The gap analysis you would have done in the first step will help you identify where your existing teams can step up. Do you have people who are great at understanding client needs? Do you have good engineers and business analysts? They could be an invaluable addition to a data science team. 

Once you know what your internal strengths are, you will know exactly what to look for in hiring external talent.

Step 3: Communicate transparently to increase data science value

The very foundation of data science is transparency. You need to see reality first before you try to change it. So a crucial step in your data science journey is embracing transparent communication as a central organizational value.

Obtaining important analytics insights is great. But in order to put them into practice and make a real impact, then you must explain what they mean. Why are they important? How do they connect to the company's goals or challenges? What old problems will they solve and what new problems have they identified?

Transparent communication is the bridge between knowing and doing.

This doesn't apply just to the top decision-makers in the company. Statisticians, business analysts, data analysts, developers, and business stakeholders should all communicate freely and proactively. If data science teams can cooperate across divisions or silos to acquire a more global picture, they can gain important new perspectives, brainstorm ideas, and clarify their understanding of the data. 

Since the purpose of data science is to optimize processes by designing automated decision-making algorithms, data science teams need to have a very clear understanding of how optimization may affect certain operations, such as data collection and analysis. 

Transparency and open communication are a must for a successful data science strategy.

Step 4: Use visualizations and storytelling for better data science impact

Data science thrives in an analytics culture. But how do you make analytics culture thrive? How do you make it appealing, relevant, and interesting to more employees, opinion leaders, and decision-makers in your company?

Decisions are often made based on preconceived ideas, assumptions, and what "worked in the past." Our bias gets in our way. Data analytics offers the antidote: clear facts. When you have a firm grasp on reality, you make informed, clear-headed decisions. 

But conveying facts is often a challenge, especially to multiple stakeholders. Analytics culture always sounds great in theory but when it delivers actual findings, it is usually met with some resistance. So how do you tackle this challenge?

One way is data visualization. Bridging the gap between data and person by making the data accessible, easy to understand, and easy to interact with. It has the potential for storytelling - for example, a fusion between visualization, data analysis, and usually verbal or written discussion, like an infographic, to provide interpretation of data science results and why they are significant. 

Data science teams who use visualization and storytelling to explain what they've discovered instead of merely presenting figures that could be misconstrued are much more successful in delivering their findings to a receptive audience.

Step 5: Give teams access to all data

Data is the raw material that scientists to extract insights. Most conventional BI and data warehouse solutions only provide users with pre-selected data samples, subsets, and pre-aggregated reports that have been scrubbed by data experts.

Most people see the structured, polished result. But a vast amount of diverse data comes from multiple sources and only a select few people are able to examine it and potentially detect inconsistencies and poor quality.

Data science, in its commitment to transparency, examines semi-structured and unstructured data directly, performing quality assessment and gap detection. This is the only way to get a clear, unfiltered picture. 

Companies should create the space and conditions to examine and compile data from new sources such as computer logs, social media, and sensors, while personnel familiar with raw data should be encouraged to join data science teams.

Step 6: Prepare processes for operationalizing analytics

It has been very clear so far that data science requires an honest look at how things are, and there are several steps organizations need to take to get this far. But this is laying the foundation. The next step is to build up.

Namely, to transition from descriptive analytics to prescriptive analytics. Move from explaining how things are to generating insights and actionable recommendations.

In practice, this can take many forms. For example, improve consumer marketing by offering targeted cross-sell, up-sell, and next-best-action offerings. Another example could be large-scale supply chains that could use predictive modeling to forecast product manufacturing, packing, and shipping outcomes.

To take this step, you need to operationalize analytics. On one hand, this means reducing the costs of developing and deploying analytic models by simplifying workflows and processes.  On the other hand, it means expanding the usage of predictive models. 

Most of all, if you want to generate business value you should be prepared to change the organizational culture so that decisions are based increasingly on fact and data and not intuition and personal preference. This is by far the hardest for many companies that start building a data analytics capacity and pursue competitive advantage in this domain. 

The key to success is to be willing to turn insights into actions. 

Step 7: Govern data to avoid mistrust

Cybersecurity and data integrity flaws have become a permanent fixture of the tech world. The ethics around how businesses manage data have become the focus of governmental and public concern. 

Any company that adopts advanced consumer data analysis must also ensure that the ethical side is taken into serious consideration.

This might seem like a delicate balance to strike. But if your company accounts for regulations and requirements to protect sensitive data at the planning state of data science projects, then it will be several steps ahead. To be at the forefront when it comes to data governance, companies must fully embrace the responsibility for protecting the data in their care and proactively find ways to do so - as opposed ot only responding to new regulations. In the spirit of transparent communication, data science teams should be informed about consumer and public reactions to actions based on data findings in order to give feedback to executives on how specific programs might impact the company's public image.

Govern data to avoid mistrust

Conclusion

Data science is about honesty. First of all, inward honesty - looking at your organization and objectively examining its strengths and weaknesses. Then, data science requires honesty in accepting what you find and communicating it with the world - with your peers, with key stakeholders, with customers. Finally, it’s about being honest with the public and guaranteeing that you meet the ethical demand about the data in your care. If your organization is committed to honesty and transparency, then data science will be less of a challenge and more of an adventure that brings exciting and meaningful rewards.

Do you need a partner in navigating through times of change?

At Prime, we specialize in delivering success and will be happy to accompany you through your data science and analytics journey, all the way into the stratosphere. Learn all you need to know about data science or just book a consultation with our team of experts to start your data science journey efficiently, with the right team on your side.

Mihail Yanchev
Data Science Analytics Lead Prime Holding JSC View all posts