9 Pieces of a Data Science Strategic Plan

A data science strategic plan isn’t just about data. Rather, it is a comprehensive plan that defines how you build and maintain an ecosystem that delivers sustainable value from your data science investments.

Developing this plan is not easy. So why bother?

Well, without one, you could be steering down a path without knowing where it is going. Technical and business risks increase, priorities might be unfocused, teamwork won’t be coordinated, and odds are your projects will fail.

Rather, an effective plan paves the path toward an organizational competency around data. This path is fraught with challenges, and an effective data science strategic plan won’t guarantee success. But it can help get everyone rowing in the right direction so that you can focus on what matters most.

The Data Science Strategic Plan

There are many possible components to your data science strategic plan. Let’s dive into 9 key ones.

1. Vision

A compelling vision inspires a greater purpose, motivates change, and aligns everyone involved to a common direction. Without one, you’ll likely get bogged down into the minutia of the day-to-day challenges and overly focus on short-term opportunities.

Therefore, build an effective vision. The best ones align with the organization’s mission, focus on successful outcomes, and avoid technical or business jargon. Supporting artifacts like mission statements, and value statements/philosophies can augment the vision and provide further definition to the why, how, and what of your organizational data science strategy.

Whereas the following key components can logically fall into a different order, the vision should always remain first. Indeed Start with Why and from there, you can layer other components onto a solid vision-based foundation. In other words, this vision serves as a guide to define the rest of the data science strategic plan.

2. Culture

According to a NewVantage Partners Big Data and AI 2021 Survey, “The greatest challenge for leading companies in their efforts to becoming data-driven continues to be due to cultural barriers – 92.2%”

Consequently, an effective data science strategic plan looks into industry-wide and organizational-specific cultural challenges. Common challenges that the survey identifies are:

  • organizational alignment
  • business processes
  • change management
  • communication
  • people skillsets
  • resistance or lack of understanding to enable change

On the more positive side, an effective strategic plan also assesses the cultural tailwinds that can facilitate data-driven adoption. Look at your organization’s values and mission as well as your co-workers’ individual motivators and assess how data can support them.

Specific tactics to help support the organization’s culture could include a data science club, lunch and learns, dev discussions, hands-on labs, recruitment events, dedicated chat/support rooms, and support from internal communications.

3. Team

The data science strategic plan also defines the team composition, organizational structure, and training plans to upskill the broader organization. Common questions that the plan should answer include:

  • Who is ultimately responsible for the vision and execution? Larger organizations might lean on a dedicated Chief Data Officer while smaller organizations might look to a Data Science Team Manager or to a more generalized Chief Information Officer.
  • How is the data science function structured? Would a centralized or decentralized structure be more effective? Regardless, ensure you align the communication strategy among all the key players.
  • Who should be on the data science teams? It’s a team sport. So be sure to think beyond just the data scientist and include the broader set of team roles.

4. Data

Move over oil!

A May 2017 cover of the Economist declared data as the world’s most valuable resource. Fast forward today, 5 of the 6 companies worth over $1 trillion are built on top of strong data strategies.

CompanyMarket CapStrong Data Strategy
Microsoft$2.59 Tyes
Apple$2.48 Tyes
Saudi Aramco$2.01 Toil (not sure)
Alphabet (Google)$1.97 Tyes
Amazon$1.71 Tyes
Tesla$1.12 Tyes
World’s most valuable companies as of Oct 31, 2021 per companiesmarketcap.com/

This trend will continue as companies that understand data as a key organizational asset will continue to thrive. Thus build a data strategy to support the capture, storage, and retrieval of data.

An effective data strategy will define comprehensive data governance issues such as ethical data use and how to make data FAIR (see go-fair.org/):

  • Findable – Make data and metadata easy to find by both humans and machines.
  • Accessible – Authorized and authenticated users can access the appropriate data.
  • Interoperable – Data sets should be easily combined with other data sets and applications.
  • Reusable – Don’t reinvent the wheel by creating redundant data sets representing multiple different sources of truth. Rather, data sets should be reused.

Don’t Miss Out on the Latest

Sign up for the Data Science Project Manager’s Tips to learn 4 differentiating factors to better manage data science projects. Plus, you’ll get monthly updates on the latest articles, research, and offers.


5. Technology

The tech stack for any technology-driven field is constantly evolving. This is especially true in an emerging field such as data science. Don’t be left behind. Rather, set a plan to frequently discover, vet, purchase, and maintain an ever-evolving data science tech stack.

This ecosystem should provide data professionals with the proper machine learning tools and computational resources to allow them to get the job done. It should also provide end-users with the means to access the results of their work.

An effective data science technology strategy will outline how to:

  • Vet and manage vendors and their platforms.
  • Standardize toolsets when possible (key for scalability and maintenance).
  • Drive cloud-native adoption when possible.
  • Efficiently handle security risks and software updates.
  • Train end-users.
  • Define common protocols (such as defining data pipelines or API deployments).
  • Evolve from the current technology tools into future tools.

6. Product Management

There are nearly endless opportunities that your data talent can explore. However, their capacity is limited. How do you know what are the most important opportunities to focus on? And which ones to defer?

Enter the emerging world of data science product management.

To support this function, a data science strategy needs to outline the type of research and products that the data science teams will deliver. More broadly, it will also help define a discovery and prioritization process that guides data science investments.

Your organizational product strategy may differ, but key tenants I evangelize are in this Data Science Product Manifesto.

7. Program Management

Data science projects are unique from software projects. As such, an organization should define a strategy behind how it manages data science projects. Six tips include:

  • Define a data science life cycle to outline the steps necessary to define and deliver a  project.
  • Define a collaboration framework to guide how the team should communicate and coordinate its tasks.
  • Build a comprehensive process that combines the life cycle and collaboration framework.
  • Adopt agile principles so that the team can quickly deliver and solicit value on incremental value.
  • Make your processes repeatable.
  • Don’t make your processes overly burdensome or otherwise constraining to the data science life cycle.

Read Jeff’s post on Defining a Data Science Process for greater detail.

8. Machine Learning Operations

machine learning operations venn diagram

Just because you got a model working, doesn’t mean your work is done. Rather, you need to figure out how to maintain, sustain, and even accelerate value after production.

Why not just use your organization’s software operations plan?

Well. machine learning models require a heavier focus on maintaining data and the models. Some of these concepts are foreign in software (e.g. you don’t need to retrain software). Consequently, build your machine learning operations around how you will maintain machine learning systems and consider:

  • Operational strategy and processes
  • Cloud systems management
  • Data management
  • Model management

9. Strategic Roadmap

You can’t execute the entire data science strategic plan at once. Rather, prioritize specific elements of your plan and use these as building blocks to the overall implementation. Lay these into a timeline so that you have a roadmap to communicate priorities and timing expectations. 

All of the strategic components are dynamic and should update to reflect current realities and re-shuffled priorities. This roadmap component is the most dynamic. Re-visit it on at least a monthly basis

Bringing it all together…

But what about …?

Yeah, there are a lot of additional components that your data science strategic plan should include. But these nine major items serve as a strong starting point for you to build a plan that meets the needs of your enterprise.

Good luck! And reach out if Jeff or I can help.

Handy External Links to Learn More…


Master Data Science Project Management

Apply the latest data science process research with practical tips from the field.

With over 6 hours of on-demand content and 2 hours of personalized coaching, the Data Science Team Lead course provides the leading agile project management certification focused on data science projects.

Deliver data science outcomes. Differentiate yourself. Get certified.

Share this Post: