Best Practices for Business Units to Collaborate With a CoE

Dataiku Product, Featured Catie Grasso

Effective collaboration between business units and Centers of Excellence (CoEs) is essential for unlocking the full potential of AI. This article outlines four strategic approaches to enhance this collaboration: establishing strategic partnerships, clarifying responsibilities, actively engaging in projects, and promoting AI advancements. These strategies aim to align CoE expertise with business goals and amplify the impact of your AI initiatives.

Ultimately, it is in the best interest of business units that CoEs scale. Spinning up AI efforts within business units alone can be extremely costly, on top of other potentially insurmountable challenges, like hiring or upskilling the right staff and choosing the right projects. From an organizational perspective, CoEs provide a more unified approach to AI projects (think preventing repeated efforts across the company) and can also bring a more innovative mindset.

For those on the business side, know that the best outcomes come from working with CoEs toward a shared view of success — it takes effort on both sides in order for CoEs to function efficiently. Here are four best practices for collaborating with CoEs to ensure mutual benefit.

1. Partner

View a CoE as a partner, not as a service provider. That means instead of throwing AI projects or initiatives over the fence and expecting the CoE to deliver a result, the teams should commit to working together. In looking at a simple AI project lifecycle (Figure 2), business is obviously involved at a bare minimum in the scoping portion. 

However, subject matter experts on the business side should also be heavily involved in design and even production phases. For example, only someone who knows the use case can decide if batch or real-time scoring is an appropriate business solution, and no one knows the data being used better than analysts from the line of business itself.

Ultimately, whether the CoE is providing a platform, enablement measures, support, operationalization of projects, or the final products for the business to use — or some combination of some or all of these — there should be strong collaboration not only in terms of sharing business objectives and challenges, but in the full creation from start to finish.


 Figure 2: An example AI project lifecycle

Working with CoEs from start to finish will garner better results that are more closely aligned with business goals. And bonus: Working with CoEs also helps upskill members of the business team, who will be more familiar with the process for the next data project and are more empowered to work with data on their own. This close collaboration is what will allow AI projects to succeed and CoEs to scale widely across the enterprise.

2. Establish Clear Responsibilities

Not all AI operating models are created equal — as mentioned in the previous section, some CoEs take ownership of project operationalization, enablement initiatives, training, etc., and for others, this falls on the business. What’s more, AI initiatives are new for most organizations, which means operating models and processes are likely to shift over time.

Therefore, it’s important to understand from the business side which roles and responsibilities fall with the CoE (and which don’t) to devote sufficient time and resources to ensuring AI project success. This helps both sides budget resource needs upfront and also avoid conflict during the course of AI projects — not having clear owners for tasks can result in either projects that never make it to completion or, by contrast, that suffer from having too many cooks in the kitchen, so to speak.

3. Take an Active Role in Your Organization’s AI Transformation

CoEs can’t make up their own use cases without support from the business, so to survive, they need demand. While part of the onus lies with CoEs themselves, the business also has a role to play. For example, it can:

  • Establish and support AI champions that explore and vet use cases, ensuring that CoEs have a steady stream of quality business problems to address.
  • Provide executive support, painting AI initiatives as a priority and thus building awareness and excitement for everyone.
  • Think beyond the dashboard, searching for more creative and cutting-edge use cases that the CoE will be excited to work on. Note that to generate demand for the CoE, the business doesn’t need to come up with a fully baked solution — even bringing simple business problems and then, as in the previous section, partnering with the CoE to solve them can reap quality results.

4. Be an Evangelist

Evangelism is about two things: first, communicating the business value and results the team has seen by partnering with the CoE on specific use cases, and second, about being a cheerleader for AI initiatives at the company at a larger scale.

Regarding the first point, keep in mind that while the CoE and technical experts can communicate on technical matters (like model accuracy, for example), they likely don’t have the proper resources to flesh out business value without your help. Only business teams can evaluate and quantify the value — both implicit and explicit — that they’re seeing in ways that will resonate with other teams in the business unit as well as other business units around the company.

When it comes to the second point, one might ask: Why does proliferation of AI matter? Well, when AI is widespread, there can be reuse and capitalization. While tackling larger, high-priority use cases, the organization or even other teams in the business line can also take on lots of other smaller use cases by reusing bits and pieces, eliminating the need to reinvent the wheel with data cleaning and prep, operationalization, monitoring, and more (Figure 3).

This is the crux of being an AI innovator. It’s not about just one successful use case, but seizing the AI wave to create lots of successful, business-impacting use cases throughout the company. However, this longer-term goal starts with AI evangelists talking about their triumphs — what use cases worked, how they were executed, and what the results ultimately were for the business.

Visuals for Modern AI Platform

Figure 3: AI overall becomes less costly when more groups are involved and pieces can be reused among them.

Bonus: One Best Practice for Everyone

Whether part of an AI CoE or a business unit leveraging a CoE for AI initiatives, both sides must treat the initiative with an open, change-management mindset. Becoming a mature AI organization means democratizing the use of AI throughout the company, and realizing this goal means fundamentally changing the way people work. 

Of course, some teams and people will be more impacted than others, but the ultimate vision is to have a company where everyone’s daily job is intertwined with and enhanced by data, analytics, or AI in some way.  

When taking on an initiative to bring AI to an organization, one must be prepared with the right mindset. There will be resistance at every level, and transformation will not be quick. Those that expect rapid success and change can become easily discouraged, quitting before the company has a chance to succeed.

You May Also Like

Accelerating Biopharma R&D With Dataiku Solutions

Read More

Advancing Healthcare With Dataiku and NVIDIA

Read More

Tips & Tricks to Mastering Data Quality in Dataiku

Read More

How Can CIOs Bridge the Gap Between Modern Analytics Aspirations and Reality?

Read More