data science

2019 is the year that data science, machine learning and artificial intelligence for business will become ubiquitous. Most organizations large and small, across all industries, have recognized the benefits and competitive advantage that these capabilities bring to bear. If you have not already begun the journey, chances are this will be the year you begin to develop this competency. Whether you’re about to take your first step, you’re a team of one looking to scale, or even a more mature organization that is always seeking self-improvement, consider the following traits to maximize your chances of success with data science.

1. They Have a Strategy

Effective data science teams have taken a little time up front to develop a plan for how they intend to generate value from the organization’s data. They can easily articulate what the program KPIs are, how projects are prioritized, and what their roadmap looks like down to specific use cases and models. They’ve been mindful enough to align their strategy with the business strategy so that priorities and objectives are reflective of the broader business goals and growth initiatives. They also know well enough not to over-plan and be too detailed with their roadmap given the highly empirical and iterative nature of data science.

2. They Apply Lean Principles & Design Thinking

When it comes time to execute, highly effective data science teams know that being able to identify a good data science or machine learning problem is the most critical step in the process. They are quick to apply lean principles to reduce risk, waste, and stay hyper-focused on generating business value. This means at first they may not even have a full time data scientist, fancy tools, a data lake, or many of the other buzzwords often thought to be necessary for success. They know what the KPIs are for their use case and know how they will most effectively change them for the better.

data science design thinking

They also know that the process of creating value with data doesn’t end with a trained model. Before even touching any data, they often use a heavy helping of design thinking to envision how machine learning predictions, classifications and recommendations will change the way the business operates. They compile journey maps and process flows, sketch out wireframes, and build mockups. In deployment, they re-engineer the business process and build rapid prototypes to demonstrate how the process will work before the use case is industrialized into the business. 

3. They Have Governance

High performing data science teams can see the future for their corporate discipline, and they anticipate where things will break down without some well thought out process. They establish a steering committee for guiding the program and making decisions about investments and activities. They formalize a disciplined full-lifecycle process for doing data science that extends to managing and monitoring deployed models over their lifetime. The team has established channels for knowledge exchange and curation of intellectual property and data science assets between teams. They strive for better data governance and quality to enable new use cases and improve model performance. They also have a code of ethics around the usage of data and readily use the core values of the organization as a framework for decision making about how information is handled.

4. They Collaborate & Empower Others

The most effective data science teams view their role as one of mentorship, evangelism and education. They acknowledge that they cannot just be individually productive, but that to properly scale and develop organizational maturity, they must enable data science thinking and skills across the entire company.

data science

In contrast, bad data science teams seek to build walls, create bottlenecks, and continue to perpetuate stereotypes that theirs is a mysterious and inaccessible dark art. Bad data science teams look at modern self-service tools as a threat or danger because they can be wielded by non-statisticians. Great data science teams stick to their lean principles and see opportunity in these new modes, where with proper checks and balances, they can generate more use cases and ideas without having to field an army of data scientists. They also see this as a way to more effectively engage the business subject matter experts that are essential to designing great use cases for machine learning and AI.

5. They Use Smart Technology for Speed & Scale

Highly effective data science teams know that you can have the best laid plans, process, and approach, but working with data is where the rubber meets the road. Smart teams are not afraid to experiment with new technology that could make their jobs easier. They use elastic or cloud computing because data storage costs are lower and it allows them to adhere to lean principles and only pay for compute when necessary. They use technology like automated machine learning because it lets them move faster with fewer resources and it supports their mission of empowerment by letting non-data scientists contribute more directly. They implement data wrangling and model development tools for both programmer and non-programmer types, so that no matter what your technical background you can leverage data science skills and tools.

In addition, the IT organization they partner with understands how their creative process is highly iterative and experimental, and that when the data science team says they “need all the data,” they aren’t being lazy or facetious, but they really don’t yet know what will be important, and anything less would be success-limiting. Together, an effective data science team strives to create a data science habitat that allows the team to move very quickly in an unencumbered fashion.

 

AI Jump-Start: Powerful Data Science Accelerator

If you’re just starting your organization’s journey towards AI competency, Ironside recommends our AI Jump-Start: a powerful data science accelerator that brings together strategy, execution, process and governance, mentorship and education, and best-in-class smart technology in a single managed platform.

Drawing upon over a decade of applying data science to business problems, we’ve combined our consulting expertise with class-leading technology partners such as DataRobot, Alteryx and Amazon Web Services to generate rapid business value, and onboard data science as an organizational capability.

 

 

About Ironside

Ironside was founded in 1999 as an enterprise data and analytics solution provider and system integrator. Our clients hire us to acquire, enrich and measure their data so they can make smarter, better decisions about their business. No matter your industry or specific business challenges, Ironside has the experience, perspective and agility to help transform your analytic environment.