Marketing leaders are building their own measurement stacks to attribute AI search traffic because perfect Generative Engine Optimization (GEO) attribution does not exist. The teams waiting for clean data are systematically underestimating AI's contribution to their pipeline, while those moving with imperfect proxy systems are already capturing a conversion advantage.
Building a proxy system instead of waiting for perfect data
AI assistants do not pass referral data the same way traditional search does. Natalia Bandach, Senior Director of Growth Marketing at Cloudinary, stopped looking for a single metric and built a system of three proxies. "Whoever says they have it perfectly set up is probably trying to sell something," she said. "It doesn't really exist right now."
Her first proxy is an AI visibility score, tracked through Profound and Google Search Console. Cloudinary targets a 75% overall share of voice on tracked keyword prompt groups. The company currently holds a 91% AI visibility score on the question "What's a reliable image API for automatic pre-processing?" - a number Bandach attributes to product-led authority built through years of developer adoption, not a GEO campaign.
The second proxy is self-reported attribution. Bandach shuffles the response options in "How did you hear about us?" surveys periodically to avoid anchoring bias, occasionally making the AI referral the first option. She also split-tests questions between two audience cohorts and looks for consistencies. The third proxy is referral traffic from AI chatbots. In GA4, she tracks ChatGPT, Perplexity, and Claude as identifiable sources, now also surfaced under "AI Assistant." This captures only users who clicked from an AI interface, not those influenced by an AI answer who arrived through a different path.
A vertical software CMO takes a different approach, reading anomalies already present in their analytics. When an "unknown" source spiked one quarter, their team pulled it out of the unknown, attributed it to AI, and tracked accordingly. The logic: if someone lands on a deep, specific product page that previously had no traction, there is only one way they found it. "Would I say it's a 100% science? Absolutely not. But that's the nature of the beast right now," they said.
Why less traffic can signal better business outcomes
Non-branded search is down across the board. Paula Ximena Mejia, VP of Marketing at Wix, noticed non-branded traffic fell while average conversion per user grew double digits year over year. Another team saw conversion to pipeline from GEO-sourced leads run at triple their channel average. "I've never in my career seen a channel increase pipeline 3x," that CMO said.
The explanation is structural. GEO users arrive after completing research through AI. They have evaluated options and formed a preference before visiting a site. The funnel has narrowed at the top and become more efficient in the middle. AI-driven traffic is smaller, but the buyers it sends are better qualified.
Mejia warns against teams that reach for paid to prop volume numbers back up. "You can end up artificially propping traffic for the sake of not losing traffic," she said. "Traffic isn't the goal-conversions are." The more useful question for CMOs right now is whether incremental paid investment is buying real conversion or just covering a number on a dashboard.
Last-click attribution was always a convention
Last-click attribution was never an accurate reflection of how decisions were made, just the most measurable proxy available. GEO makes the fiction harder to maintain. "Decisions now start to happen before the click," Mejia said. "Educational content that used to drive traffic may still be shaping decisions-we just can't see it anymore."
A more resilient indicator, Mejia argues, is branded search performance. People searching a brand name signals that demand is being created upstream through AI recommendations, editorial coverage, or community mentions, even when the exact source cannot be traced. It is the footprint of influence that does not require a click to exist.
Mejia now questions whether all content is worth creating. "Saying yes to something comes at the cost of saying no to something else," she said. The goal is to address what some call a brand's "language layer" gap: optimizing content to align with brand ethos and ensuring third-party language reflects the same messaging.
GEO demands company-wide buy-in
A single team can own SEO because they control most of the inputs. GEO cannot be contained that way. It spans content, PR, community, social, and employee-generated content. At Wix, this meant establishing "semantic triples" - a common language across every team that creates assets. Whether it is an SEO page, a community post, or a journalist briefing, every piece of content describes Wix's products using the same specific language.
The biggest unlock came through PR. At Wix, PR sits outside marketing, but journalists are now briefed using the same vocabulary before every major story. "If it gets published, it helps," Mejia said. "It's become almost an operational requirement for GEO to be effective."
At Cloudinary, Bandach structures GEO as a flywheel divided between branded and non-branded, with distinct owners at each stage of the funnel. Their newest motion is what she calls AX, or agentic experience: designing content and product surfaces for AI agents making decisions on behalf of users. A CMO at a vertical AI company opens every marketing meeting with ten minutes of "who's done something new with AI?" The company's last offsite was dedicated entirely to GEO. The directive to take GEO seriously, they said, has to come from the CEO down.
Start measuring imperfectly now
The teams gaining an advantage have already started measuring despite incomplete data. The vertical AI CMO adds a tax to self-reported survey data when response rates are low enough that raw numbers undercount meaningfully. Their team also runs a weekly review cycle on visibility score, sentiment and truth score, and recommendation position. "Things move too fast now for quarterly reviews," the CMO said.
At Cloudinary, Bandach runs 120-150 experiments per quarter and is scaling up for GEO. One experiment involves recency keywords - time-sensitive terms like "best XYZ in 2026" - that trigger a live search for the latest information. According to Bandach, content optimized with recency keywords can show up in 2-4 weeks. This is a fast win when trying to gain visibility on specific queries within a week or two.
The common thread is a willingness to act on directionally accurate data rather than wait for precise data. Build the proxy stack, review weekly, and treat every measurement call as an opportunity to iterate.
Why this matters for marketing professionals
Marketing leaders who wait for perfect GEO attribution before committing resources are systematically undervaluing a channel that, for some teams, is converting at triple the channel average. The practical path forward is to build a measurement stack using the six KPIs these practitioners are already using: AI visibility score, AI referral traffic, behavioral anomalies, self-reported attribution, branded search volume, and pipeline conversion rate. For CMOs and senior marketers looking to build these systems, structured learning paths like AI Learning Path for CMOs can accelerate the shift from waiting to measuring. The teams treating every measurement decision as an experiment to refine, rather than a methodology to perfect, are the ones capturing the advantage right now.
Your membership also unlocks:
