The uncomfortable truth of AI search: marketing teams are still working in silos

AI search is forcing a reset across marketing, communications and digital teams, but most organisations are still structured in ways that make it harder to compete.

New research from Sefiani and Clarity Global found 84% of Australian marketing and communications leaders disagree on who owns AI visibility, even as brands overhaul strategies to show up in tools such as ChatGPT and Google’s AI Overviews.

That lack of ownership is creating a structural problem at the exact moment discovery is shifting.

Tom Telford, chief digital officer at Clarity Global, said the industry is facing what he describes as the “uncomfortable truth of integration”.

SEE MORE: Brands face visibility battle as AI reshapes the buying journey

Right now, SEO, comms, content and social teams operate with separate metrics, budgets and leadership. In an AI-driven environment, that fragmentation is a liability.

“Customers are looking at AI before they speak to your sales teams,” Telford said. “If you are in shambles at one end, it will be shambles at the other end.”

The shift is already underway. According to the research, 91% of leaders are reworking their strategies to influence AI-driven discovery, with budgets moving toward what is now being described as a “citations race”, competing to be referenced in AI-generated answers rather than simply ranking in search results.

Nearly half of organisations have already allocated 5 to 10% of marketing and communications budgets to AI visibility, often pulled from paid and brand spend.

“Reputation used to be managed channel by channel, but AI search has changed the rules,” Telford said.

"Because these systems read across everything - earned coverage, on-site content, social signals, and search authority - siloed marketing and communications are quietly muting your AI visibility.”

The implication is clear: visibility is no longer channel-based. It is cumulative.

AI models pull from across earned media, owned content, social signals and search authority. When those channels are misaligned, the output is inconsistent, or worse, incorrect.

One in four leaders said inaccurate or outdated brand information has already appeared in AI-generated answers, while 77% reported siloed structures caused real issues in the past year.

That risk is amplified by how AI systems behave.

Where traditional search could bury negative or outdated content over time, large language models resurface patterns across years of data, presenting them as current truth.

For Mandy Galmes, managing partner at Sefiani, the shift places earned media and credible third-party sources at the centre of discovery.

“When LLMs answer a question in your category, they’re drawing overwhelmingly on non-paid, third party sources,” she said.

“If your spokespeople, experts, case studies and proof points aren’t in those sources, you’re invisible at a key moment in the buyer journey.”

That is changing how teams need to operate. Telford argues the next phase of marketing requires a unified approach, with shared goals, shared metrics and shared accountability across traditionally separate functions.

“It’s not a GEO strategy, it’s a marcomms strategy,” he said.

The concept of generative engine optimisation, or GEO, is emerging as a new layer of planning, one that cuts across the entire organisation rather than sitting within a single team.

“If you have SEO, comms, content and social all lined up, the fifth channel - AI - will be free,” Telford said.

That level of integration is still rare. Only 16% of organisations surveyed said they are taking a fully coordinated approach to AI visibility.

In response, consultancies are beginning to build new tools and frameworks to help brands understand how they appear in AI-generated answers.

Sefiani’s new proprietary tool Surfacd tracks how brands surface across systems such as ChatGPT, Gemini and Perplexity. But the bigger challenge is not measurement. It is structure.

“Surfacd is unlocking the black box of AI search for Australian organisations,” Galmes said.

“It provides the data that marketing and communications teams need to own the answers, and ensure they are consistently included when AI tools discuss their category.”