Omnicom’s Analytics Chief: For AI, Guardrails Will Set Business Outcomes Free

Why It Matters

By tying AI to explicit business goals and secure, cross‑platform measurement, marketers can convert data overload into measurable revenue growth, giving agencies a competitive edge in an increasingly fragmented digital ecosystem.

Key Takeaways

• Data abundance outpaces actionable insight and long‑term decision making.
• AI must be anchored to specific business outcomes from the start.
• Guardrails are essential to balance efficiency with revenue‑focused metrics.
• Unified identifiers enable cross‑platform measurement and secure clean‑room analytics.
• Continuous experimentation creates a durable, adaptable measurement framework.

Summary

Omnicom’s analytics chief highlighted a growing paradox: agencies now have unprecedented data volumes, yet the ability to translate that data into long‑term, revenue‑driving decisions remains limited. Traditional dashboards deliver short‑term vanity metrics, leaving a gap between insight and actionable strategy for clients.

The speaker argued that artificial intelligence can only unlock true business value if it is tethered to clearly defined outcomes—revenue growth, client retention, or frequency metrics—right from the outset. Guardrails must be programmed to keep AI focused on efficiency without sacrificing effectiveness, ensuring the technology respects the boundaries of brand goals and regulatory constraints.

Examples included Acxiom’s Real ID, a unified customer identifier that stitches together cross‑channel signals within a secure clean‑room environment. This unified view enables marketers to assess the collective impact of multiple touchpoints rather than isolated channel performance. An "always‑on" experimentation layer was also cited as essential for a durable measurement framework that evolves with shifting consumer behavior.

For agencies and advertisers, embedding outcome‑driven AI guardrails and a unified identity layer promises more reliable attribution, faster optimization, and a resilient analytics foundation that can adapt to the fragmented data landscape of 2026 and beyond.

Original Description

The data has never been more abundant. The problem is what to do with it.

Three-quarters of U.S. buy-side leaders now believe traditional ad measurement approaches are underperforming, according to a February 2026 eMarketer report, a figure that suggests the dashboard era may have run its course. The answer, said Holly Yonosko, chief analytics officer at Omnicom Media Group, is not more data – it is smarter constraints on how AI uses the data already available.

“We have access right now to the most data we’ve ever had access to,” Yonosko said. “But then where the issue comes in is the measurement and the actionability of that data.”

The fix, she argued, requires moving well past the vanity metrics that populate most client dashboards and toward frameworks built around durable, longer-term business outcomes.

The guardrails problem

Faster optimization sounds like an unambiguous good, but Yonosko cautioned that speed without direction is its own kind of risk — particularly when AI agents are left to chase efficiency metrics in isolation.

“It’s really important first to anchor your AI, if you’re building an AI agent for your client, in those business outcomes,” she said. “Ensure you’re defining those up front, whether it’s revenue, whether it’s retention of your clients, whether it’s growing clients.” Without that anchoring, she argued, the system will optimize for whatever it can measure most easily, which is rarely the same thing as what the client actually needs.

“You need to make sure it understands what are some of those efficiency metrics that it should be aware of as well as those business outcomes of revenue or frequency or some of our attention curves,” Yonosko said. “So that it has more context. So it’s not just getting out of control.”

The unified identity layer

Cross-platform measurement has long been the industry’s white whale, and fragmented identity signals have only made the hunt harder. Yonosko pointed to Acxiom’s Real ID – Acxiom being an Omnicom-owned data and technology company – as the connective tissue enabling a more coherent view of consumers across channels.

“We have Real ID, which is a unified identifier,” she said. “We can connect cross channels to identify unique individuals, to then also create a clean room and do measurement within a secure closed loop environment so that we can provide real optimization metrics and measurement for our clients.” The clean room model has become a standard-bearer for privacy-safe measurement, allowing brands to analyze audience behavior without exposing raw personal data to third parties.

IABs survey have found a growing number of ad buyers planning to increase their focus on cross-platform measurement. “We’re not really looking to understand how does each individual channel perform,” Yonosko said, “but what is the outcome of having these channels work together on each consumer.”