Agencies Set New Tests for AI Search Measurement Tools

By Priya Raman

Michael Sondak, Omnicom’s North America SVP and head of search, said: “This is not a lift-and-shift from the success that we’ve had for the last 20+ years in running both paid and organic search programs.”

Sondak added: “It’s wildly different: Different orchestrators, different technology, different dependents, different teams, [and] different talent.”

Across interviews with Marketing Brew, Omnicom and VML executives said leading purchase criteria for answer engine optimization (AEO) platforms include the ability to measure brand visibility across several large language models (LLMs) and the method used to collect data. They also examine refresh frequency, treating it as a strategic choice rather than a simple feature comparison.

His comparison points to a larger operating-model gap than a new channel label suggests. Agencies need dedicated orchestration, technology and talent rather than simply extending existing paid and organic search processes, Sondak said.

VML avoids a single-platform commitment

VML has not committed to one AEO platform. Chief discoverability officer Heather Physioc said the agency considers each client’s marketing mix and needs to help clients choose the right tool.

Two platforms come up more often in client conversations, Physioc said:

• Profound, in part because of its heavy marketing efforts.
• Adobe Brand Visibility, which can plug into some companies’ existing technology stacks through platforms such as Adobe Analytics.

Physioc named two key screening features:

• The ability to measure brand visibility across several LLMs.
• The ability to measure trends in the data over time.

Some platforms also offer less common capabilities, such as tracking brand sentiment within LLM responses, Physioc said.

She also set a limit on what any single metric can establish: “None of these, in my mind, are the true North Star, definitive metrics that we will always consider the extent of measuring AI search,” she said. “I think there are more metrics to come.”

Collection methods determine score credibility

At Omnicom, data-collection methodology is a central screening factor, according to Dan Toplitt, head of AI platforms, North America. Teams need to know whether a platform gathers LLM responses through an “API or a headless browser,” he said.

Reliability is equally critical, Toplitt said, because LLM responses can vary with different user prompts. The agency also asks how each platform normalizes its data set.

He wants to know “how platforms approach the normalizing of that data set to give us a confidence level in how representative of a large-scale category [this data is],” he said.

Refresh speed needs a decision window

Refresh cadence determines how agencies can use AEO data. Vendors make different claims:

• Bluefish AI claims it tracks daily citations across more than 10 LLMs.
• Ahrefs Brand Radar claims it updates visibility data for Google AI Overviews and AI Mode continuously.

Toplitt said he may review client data “week to week” when a marketing leader wants weekly executive updates. He avoids making major strategy changes based on minor variances, he said.

“ If I’m looking at data over a period of three days and I’m trying to make decisions about a roadmap for the next quarter or year’s worth of activation work, that three-day period is not going to be sufficient to give you the level of insight you need to be smart about the recommendations you’re making,” he said.

Physioc warned of a separate risk: “some risk to overchecking” constantly refreshed data and making “knee-jerk decisions when there’s not yet a pattern.”

Proprietary scores lack a shared benchmark

Physioc expects the market to move toward standardized measurement. “Every one of those platforms measures some kind of AI brand score with a completely unique proprietary metric,” she said. “Eventually, we may see some kind of standardized expectation for what that translates to.”

What does this change for agency teams?

It changes the buying checklist. Coverage and collection methods establish whether a platform’s score can represent a market; refresh behavior determines whether an agency can distinguish a sustained trend from noise. Agencies should assess those factors before treating any proprietary AI brand score as comparable.

Physioc’s expectation of a standardized translation gives buyers a concrete benchmark to watch. Until it arrives, agencies need platform-by-platform evidence on LLM coverage, sampling, normalization and refresh cadence before using AEO scores to guide strategy.