ServiceNow, a global leader in digital workflow automation, powers mission-critical operations for most Fortune 500 organizations. The company’s cloud-based platform integrates AI to streamline processes, enhance decision-making, and deliver measurable business value.

In this first installment of a three-part interview series, Vijay Kotu, Chief Analytics Officer at ServiceNow, speaks with Vamsi Duvvuri, TMT AI Leader at EY, in a video interview about the formative career experiences that shaped his philosophy on analytics, trust, and strategic focus.

Kotu also shares five key metrics for AI use case evaluation, turning purpose into value, and embedding governance into AI. 

A lesson in trust

Reflecting on his journey, Kotu recalls two key milestones. One such moment occurred nearly two decades ago at a small San Francisco startup. Kotu was building prediction models and writing database queries when an unexpected endorsement in an executive meeting made him realize the weight of his work.

He explains how his work unexpectedly became a focal point in an executive discussion when someone challenged a business leader’s assumptions, asking where the data came from. The leader said it came from him, and that ended the discussion.

That moment made him realize the deeper role of analytics: “One thing that established that is it’s like trust. We are in a trust business. Particularly, we are providing a single source of truth for the entire company, and that trust that people have in us is critical.”

The second lesson came after joining ServiceNow, when he presented an ambitious 30-page analytics strategy to then-CEO John Donahoe, and he asked Kotu to list five things he could do for the company.

Kotu shares that the moment taught him that the real purpose is to create decision-making capabilities across the company with clarity and focus.

Measuring trust and value in AI: 5 metrics for success

The leap from an AI pilot to a production-ready product is rooted in data-driven rigor, Kotu believes. “The one thing I keep saying is the connection between a pilot and the actual product is going to be math. Trusting in math.”

He outlines five key measures that matter for any AI use case evaluation:

1. Target persona: The persona and the role a use case targets should be clear.
2. Adoption rate: A great use case will be used by all, which makes it necessary to measure how many people are actually using the product.
3. Accuracy and sentiment: “For any AI use case, there are some technical accuracy measures that we can use, and sentiment is how people are using it. What is their thumbs-up, thumbs-down feedback?”
4. Business metric impact: Understanding which operational business metric this use case is changing is critical.
5. Value conversion: According to Kotu, “Anything should be converted into value. For most organizations, the value is going to be improving the top line or bottom line.”

For him, these five metrics allow leaders to see clearly which AI use cases are worth scaling and which are not. This approach ensures that AI projects don’t remain stuck as pilots but evolve into impactful solutions tied directly to measurable business outcomes.

The most common struggle: Turning purpose into value

Moving forward, Kotu acknowledges that while the five measures for AI use case evaluation are all essential, one in particular challenges most leaders: Converting the purpose of a use case into tangible business value. He emphasizes that assumptions are acceptable — provided they are transparent. The key, he says, is translating performance into clear top-line or bottom-line impact.

Embedding governance into AI

Kotu notes that governance, privacy, and security remain top concerns for enterprises. However, in most organizations, only a small group of people think about improving governance daily.

“There are special sets of people. We have a couple in our own team. But for the rest of the teams, we need to make it simple, so that everyone can adhere to the governance principles.”

For embedding governance directly into the AI lifecycle, ServiceNow created AI Control Tower, a tool that:

• Tracks how a use case is created
• Captures the datasets involved
• Ensures compliance with policies from the start
• Triggers workflows for governance, privacy, and security team approvals

Kotu explains, “That’s the only way for you to launch an AI use case inside the company.”

Once approved, the AI Control Tower becomes the single interface for measurement and management. The same five metrics outlined earlier are coded into the tool, enabling enterprise-wide visibility, he concludes.

CDO Magazine appreciates Vijay Kotu for sharing his insights with our global community.