The 3 AI operating models — and how to know which works best for you

Venture BeatThis post was originally published by at Venture Beat

“No one should select an AI operating model just to select an AI operating model,” says Jennifer Roubaud-Smith, VP, global head of strategic advisory at Dataiku. “No one should set up centers of excellence just to set up centers of excellence. It should always be about the kind of business challenges you want to solve.”

Whether your objective is to solve some key enterprise-wide business challenges, or whether your objective is to transform entirely the way everyone in your organization works with analytics, your operating model is central and foundational to the success of that initiative.

Integrating analytics and business

In the early days of advanced analytics and AI, teams tended to focus on projects that were very innovative or technically challenging, which called for a center of excellence model. But today, as Robaud-Smith explains, more and more organizations and analytics teams understand that if you want to survive, if you want to be there in two years when the next CEO arrives and looks at your results, your AI initiatives need to be adding business value to the organization.

“Whatever operating model you choose, the most important thing is to ensure that there isn’t a disconnect between the analytics teams and the business, and that the mindset of the analytics team is wired in the right way,” Roubaud-Smith says. “Whether it’s centered within a line of business or centered in the entire organization, you need people who are accountable for the success of driving awareness around analytics and driving usage around analytics-powered products.”

Ideally, those people need to come from the business, know the business well, and act as engagement or business translators between the business and the analytics teams.

The 3 operating models

A center of excellence is one type of operating model to drive your analytics strategy. As the importance of analytics to the future of any organization became undeniable, CoEs grew in prominence, providing an organizational north star for AI and analytics. The AI and analytics talent is unified and located in one centralized department. This department then acts almost like a consulting firm for the rest of the organization on topics related to AI and machine learning.

In the decentralized model, the analytics team sits within the lines of business and works very closely with SMEs, or might even be SMEs themselves. There might be an IT department centrally working with them, or the IT team might also be inside the business unit.

The third model is the hub and spoke model. This is an approach aiming to have the best of both worlds, looking for a way to get the benefits of having a centralized team, while keeping analytics talent embedded within the business. The hub and spoke model has one central team working in coordination with the AI and analytics talent that is scattered across the organization.

Choosing between these requires going back to the specifics of your organization’s business problems, objectives, and goals (which you’ll learn more about in the upcoming webinar).

Facing the biggest challenges

One of the biggest challenges business leaders face in establishing a functional AI operating model is simply that it’s not something an organization is prioritizing at the very top. Instead, it’s something that lives outside the C-suite while others try to make the current organizational setup support AI initiatives.

“The CEO needs to understand that succeeding with AI at scale in a sustainable way requires a big transformation with a capital ‘T,’” Roubaud-Smith says. “There is a whole change management aspect to AI and analytics that will make the organization successful, or less so.”

That means managing the AI strategy as a clear program, setting expectations with leadership, and sharing progress on how the entire program is going. Keeping that communication open and fluid. It requires actively finding advocates within the business to be champions of the program.

Another limiting factor is when the approach is too narrow and fails to establish a model that will ensure smooth collaboration between the business, the analytics, and the IT teams all together.

Finally, an AI strategy is often launched with too small a team. From the outset, it’s important to establish a foundation that will allow your organization to scale without creating a lot of governance challenges as you grow.

Whatever the choices you make, there is one clear understanding to start with, Roubaud-Smith says.

“To a data leader, you can’t succeed without the business, and for a business exec, you can’t succeed without the data leader,” she says. “If we’re talking about sustainable, ambitious business value, then they can’t do without each other.”

Spread the word

This post was originally published by at Venture Beat

Related posts