Creating a Successful Data Science Team in Retail

Data Science Team in Retail
Picture of Linda Whitaker

Linda Whitaker

VP of Science Delivery

Data Science has the power to positively impact retail companies due to its fact-based data-driven insights. Though successful Data Science teams are often associated with technology companies, many successful retail companies are utilizing Data Science daily to further drive their success and stay ahead of their competitors (Home Depot, Kroger, and Target ring a bell?)

While it may seem simple to hire Data Scientists, it is complicated to create a long-lasting successful team without first understanding:

  1. Goals/Objective for your Data Science team
  2. Composition of your Data Science Team
  3. Where your Data Science Team should be located within an organization

Linda Whitaker, VP of Science Delivery at Cognira, takes us through the journey of how to strategically place a Data Science Team into your retail business.

What is Data Science?

Artificial intelligence and data science

Before we get started, let’s define what we at Cognira mean by Data Science.

Data Science can be confused with AI because the latter is getting a lot of publicity. But Data Science is a multidisciplinary activity that combines domain expertise, specialized programming, mathematics, and statistics to extract meaningful and actionable insights from data. Analytical thinking and problem-solving skills are a top requirement of data scientists. 

So, with that, let’s discover how to create a successful Data Science team in retail!

Create Objective/Goals for your Data Science Team

The first, and ultimately, most important step to building out a Data Science team is to understand your objective and goals. 

Take a moment to explore if your organization is ready to embrace science and analytics on a full-time basis, or if you only have the capacity for projects that can be completed as one-off consulting assignments.

Objective Goals

In essence, you’re creating an internal consulting group, which can take on projects. It is then up to the company to decide what you do with those. If most or all of an organization is not ready to embrace science and analytics on a full-time basis and make it part of the company’s culture and the company’s DNA, then starting with self-contained analysis projects is a perfect starting point.

However, if your company is ready to take the plunge and invest in the integration of Data Science into your business, you’ve come to the right place. Read on for more helpful tips on how to create a Data Science group that has lasting and permanent effects. (Don’t worry, we’ll focus on the consulting option as well)

Understand the Composition of your Data Science Team

A Data Science Team in Retail

Regardless of how deeply you want to embed Data Science into your organization, it’s very important that your Data Science team is not made up of only data scientists.

Too often, data scientists are isolated, working alone on projects, and unable to get an audience for their work. Perhaps they can’t communicate in business terms, or the business doesn’t buy into what they’re doing (the results may go against conventional wisdom). Almost certainly, they won’t be able to undertake cross-departmental projects. 

To succeed, it is exceedingly important that you set up your Data Science team to be within a more comprehensive strategic group.

We’ve laid out the roles required for successful Data Science projects, and you will notice they are not very different than those required for any project. It is possible for more than 1 person to play each role. Each role may not be needed at each phase.

Full-Time team members

Leader (Solution Architect)

This person is a data-driven Solution Architect. The leader of this group must be someone that understands retail data, systems, and processes. They must be able to communicate with the business, IT, and analysts/scientists. 

Data Scientists

There are multiple levels of data scientists. It’s good to know what you want/need so that you do not hire someone without the desire/skills to do the work you want them to do. 

  • Research Data Scientists: Pure researchers who typically work for large corporations (Google, Facebook, etc.) or academic research centers. They develop new algorithms and new approaches in artificial intelligence and advanced analytics. Their research is typically published in conferences and technical journals. Their work leads to AI/ML and other libraries included in various cloud offerings and general-purpose analytical software platforms and tools. 

You don’t need or want these types of scientists.

  • Retail Data Scientists: These Scientists have profiles very similar to Research Data Scientists, only, their goal is to leverage and augment what the Research Data Scientists do to produce algorithms and approaches that are specifically designed to address Retail’s business challenges. In general, their work is not published and it is what constitutes the ‘secret sauce’ of software companies such as Cognira. 

You may want to have a couple of Retail Data Scientists 

  • Applied Retail Data Scientists: These Scientists leverage what the Retail Data Scientists produce and apply it to a specific retailer’s challenge.

These are people who will spend a lot of time working with the specific retailer to understand their business challenges, their data, and their processes. They can audit and configure advanced software and algorithms — produced by Retail Data Scientists — to solve the specific challenge. They spend a lot of effort on data engineering (understanding, cleansing, slicing, and dicing the retailer’s data) and output measurement (impact, accuracy, ROI). Their outputs range from well-tuned analytical software solutions to process recommendations to ROI reports. Each will deliver value to the retailer. 

Most of your data scientists should fit this description.

Analysts

If you have hired new data scientists without a lot of retail experience, we recommend you have an analyst(s) that has a good feel for retail data, your processes and systems, and can also help ensure the results are described in a way that is understandable by the company. Your Solution Architect may know this, but they are too valuable, expensive, and scarce to use in this manner. They can help in all the phases, formatting reports, QA’ing the results, going through details with the business, formatting reports, and data, etc. 

Rest of Roles

The rest of the team can be people that are assigned to work on projects with the team members above, but who report to other areas. Depending upon the scope and the phase of the project this could range from part-time to multiple full-time resources. 

Project Manager:

Needed when projects reach a certain size.

Business and IT Sponsor:

As with other types of projects, you shouldn’t go into a Data Science project without a business and an IT sponsor, right up front.

Business Owner:

The person who will ultimately own this solution (this could change of course, and could start as the sponsor).

Business SMEs:

Business Experts, and they later become evangelists

IT Solution Architect:

The person responsible for the technical architecture, design, and/or technical implementation of new systems. This is an incredibly difficult job:

  • The technology to execute Data Science applications moves fast and skills are hard to find
  • Science projects (like all projects) will not solve all problems to begin with. They have to be phased and extensible

If you do not have these skills internally, I recommend that you consider hiring or contracting some support for your first solutions.

IT Data / System Engineers/QA:

Development and testing of systems.

Know Where to Strategically Place Your Data Science Team

Now that you know the objectives/goals, compositions, and roles for a Data Science Team, you need to place them strategically in your business. 

If it is your goal to start with a consulting group that can do one-off projects, then it doesn’t matter where the group resides. It could reside anywhere, and you could even have small groups set within different business units.

Structure of A Data Science Team in Retail

However, if your goal is to embed Data Science across the company, in processes and systems, it matters where the team resides. My recommendation is to start with a single group that can address problems across the company. This could be either in IT or as a separate group reporting to an Executive. The success will still depend upon the company culture and political landscape. 

Regardless of the official reporting structure, I believe that the Data Science Team needs to be seen as a separate strategy team that works together with IT and the Business. All projects have to be blessed by both. 

If I had to give one general recommendation, it would be to start with a cross-functional team dedicated and labeled as a strategic team. The team would be headed by a VP of analytics, or a VP of strategic projects, reporting up to a C-level executive.

Takeaways

Data Science in retail has the power to positively impact your organization. Embracing these advancing technologies can not only enhance your marketing strategies, operations, and financial performance but also give you the advantage that your competitors aren’t utilizing. 

Remember, before hiring your Data Science Team, ensure you have considered the 3 important factors: Objective, Composition, and Location. 

Have additional questions, or not quite sure where to get started? Get in touch with Cognira – the leading artificial intelligence solutions providers for retailers.

Cognira Logo White

About Cognira

About Cognira

Cognira is the leading artificial intelligence solutions provider for retailers. Cognira is passionate about helping retailers unlock valuable, transformative business insights from their data.

We know retail. We love data.

To learn more, check out our website at cognira.com or contact us today to get started. 

Subscribe To Our Newsletter

Get the latest updates about retail technology

Share This Post

+ More To Explore

No more posts to show