CMOs Are Chasing AI Search Scores. Is It Paying Off?

Brands are investing heavily to appear in AI search results. The harder question — whether that visibility is driving purchases — remains unanswered.

Any day now, consumers will be handing over decisions about a range of purchases of — everything from groceries to airline tickets— to agentic chatbots. At least that’s what it seems, judging by the headlines, ad tech industry conferences, and reams of marketer surveys.

But consumers’ love/hate/hope/hellscape views of artificial intelligence have certainly complicated that picture, as Pew Research’s ongoing polling of Americans’ uncertain views on AI shows. 

Still, marketers’ remain buoyant about AI’s role as a shopping assistant, much less, a decision-maker. It is expected that more than one-in-10 online shoppers will routinely use AI agents to purchase products on their behalf by 2030, according to a September Mastercard report, A Short History of the Future of Shopping and Payments.

But most advertisers aren’t waiting four years. Seventy-three percent of marketers have already invested in tools to monitor their AI search visibility, according to a recent industry survey cited by Digiday. 

The tools — Semrush, Profound, and Scrunch — can tell a brand how often it appears in a given LLM’s responses and what sources that model is drawing from. What they cannot do is connect that presence to a sale. “There is no one tool out there that can paint you a picture of the universe,” John Barham, managing partner at performance media agency Roast, told Digiday.

That blind spot occupies the center of one of marketing’s most consequential debates. Generative engine optimization (GEO, although some in the industry use “AIO” for AI optimization)  has become a budget line item at brands across categories, a spend driven by the fear that a company invisible to AI search engines will be invisible to the consumers those engines increasingly influence. Whether that investment is producing commercial returns, or producing citation scores that look good in a deck and mean little in a P&L, is a question most CMOs cannot yet answer cleanly.

‍

Visibility Is Real. Attribution Isn’t.

Joshua Nu’u-Steele, co-founder and chief revenue officer at market research firm Ideally, has been eyeing the new discovery race among marketers. “Every brand wants to be the one AI recommends,” Nu’u-Steele says. “There’s a lot of budget and priority on chasing AI visibility right now with little grounding in how people are actually using AI to shop. You can win the recommendation and still lose the sale if you don’t know where your customer is going next.”

Liz Murnin, associate partner and head of growth at VSA Partners, has watched that dynamic play out in her own business. Traffic from AI search is arriving and the companies sending it match VSA’s ideal client profile. The number of potential clients VSA identified  using this method tripled last quarter, she says. 

Prospects on new-business calls tell her that an AI tool pointed them to VSA. What she cannot yet show is a clean line from that first visit to a signed contract. “For a consultancy like ours, or any business with a long sales cycle, AI visibility tells us who’s paying attention,” Murnin says. “It’s an early signal, and I don’t attribute it to pipeline until it turns into a real conversation.”

Still, Murin does see moments when that signal is louder. Like when Expedia Group’s CMO, Jochen Koedijk, spoke at Ad Age’s Business of Brands event and described the travel booking agent’s ballooning AI search traffic. It grew from nearly nothing to a meaningful visit source in roughly two years, Koedijk told the audience, adding that conversions are climbing faster than traffic, suggesting the company’s finding its way in the emerging  “zero-click economy” era. Expedia has a short, owned path to purchase. Most brands don’t, which makes the Expedia case instructive, although still not directly transferable.

Kristina Coughlin, general manager at Trevant, a Fluent Company, describes what AI search is doing in most purchase journeys. “We’re still seeing that most consumers aren’t converting inside the LLM,” Coughlin says. “They’re using it to narrow the field, then doing their own research on the brand’s site and social.” In her framing, AI opens the door. Owned channels and trust signals close the sale.

Mark Wahl, VP of technology and information at integrated creative agency Worst Kept Secret, adds a useful data point on just how early-stage AI-referred traffic remains. Google Analytics only added a dedicated AI Assistant channel in May, and across the sites his agency manages, that traffic typically runs well under 1% of users. 

More importantly, Wahl notes, most AI influence never produces a click at all. “Someone asks ChatGPT for recommendations, gets an answer, and later comes directly to the site or buys somewhere else,” Wahl says. “That doesn’t mean AI isn’t driving conversions. It means the measurement isn’t there yet, and I’d be skeptical of anyone claiming a clean line from an AI answer to a purchase.”

‍

A Maturity Problem, Not a Permanence Problem

Heather Salkin, CEO of creative intelligence studio Artists & Robots, puts the current moment in historical context. The industry, she argues, is in roughly the same place digital marketing occupied around 2005. That’s when brands knew they needed a website but couldn’t yet prove what it was worth. 

“Visibility in AI answers is real and measurable,” Salkin says, “But where we stumble is with attribution to a purchase. I see this as a maturity problem, not a permanence problem.”

Her forward-looking read is that AI platforms will move from answering questions to executing tasks — comparing products, checking prices, and completing purchases inside the conversation. When that shift arrives, the line between “we got cited” and “we got chosen” collapses into something direct and measurable. 

“The brands paying attention to GEO now aren’t chasing a metric in search of a business outcome,” Salkin says. “They’re building the presence they’ll need once these platforms stop being ‘answer engines’ and start being transaction engines.”

Daelin Mackey, integrated media director at True Media, reaches a similar conclusion from a planning perspective. Direct commercial impact from LLM visibility isn’t where CMOs would like it to be. But GEO, in her view, should be treated as a future-facing investment. It’s best used to “optimize toward the moment conversions begin to flow,” rather than waiting until they do and then scrambling to catch up.

‍

Making the CFO Case

While  measurement currently remains a challenge, practitioners are nevertheless moving ahead, piecing together answers through a combination of proxy signals and modeled attribution, rather than clean platform data. Digiday reported that B2B SaaS company Rippling triangulates AI visibility using tools like  Profound and AirOps alongside conversion data from paid ChatGPT ads, web traffic signals, and a bespoke media mix model built on Google’s open-source Meridian platform. Roast has used Google’s Causal Impact model — a Bayesian open-source tool — to link search inputs to business outcomes.

Murnin recommends three steps toward making the “chief financial officer case”track which companies are arriving via AI search and whether they match the target customer profile, agree on what success looks like before spending begins, and be explicit about what cannot be measured yet. “CFOs will trust a number you can defend far more than a big one that sounds nice,” she says.

Coughlin maps the CFO conversation onto how upper-funnel media earned its budget years ago: tie GEO investment to consideration and shortlist metrics rather than last-touch revenue, and the case becomes familiar territory for a finance team that already funds brand spending.

Salkin expects the tooling to catch up within 12-to-18 months, producing GEO analytics platforms that tie citation and sentiment data to referral traffic, conversion, and downstream revenue — the same evolution web analytics went through after Google Analytics matured.

‍

If It’s a Brand Play, Fund It Like One

The question, “What changes if GEO never produces a clean bottom-of-funnel signal?”  produces the most revealing answers.

Murnin would treat it as brand building and measure it accordingly. 

Her current investment in GEO runs through content, PR, and executive visibility as a single integrated program rather than a separate line item. “These investments pay off either way,” she says. “If an AI model never sends us a single deal, we still have better content for outbound, stronger material for pitches, and leaders people recognize.”

Salkin views the risk of inaction in terms that go beyond missed citations. 

AI platforms are increasingly drawing from sources brands don’t own — forums, reviews, third-party platforms — while citations from brand-owned websites are shrinking as a share of what gets surfaced. “The next phase of GEO isn’t just about optimizing your own content,” Salkin says. “It’s about actively managing your presence across an ecosystem that you don’t control.” 

The discipline, Salkin argues, belongs to brand and communications leadership rather than an SEO line item — because the risk is being visible and misrepresented in a channel where there’s no opportunity to correct the record after the fact.

Coughlin considers the eyeballs, repositioning GEO as “consideration media” and shifting the content strategy accordingly. Rather than producing new material to chase citations, the focus moves to ensuring that existing reviews, user-generated content, and owned content are actually accessible to AI systems.

 “A review stuck behind JavaScript is invisible to an AI system no matter how good it is,” Coughlin says. “Fix the access problem, and visibility improves. Consideration follows.”