Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

Some marketers scale GEO faster than they can measure it

Here's why that's actually fine.

[Photo: Getty Images]

Profile photo of Andrea Lechner-Becker

Chief Strategy Officer · GNW Consulting

If you read my previous column, you came away with the argument that the brands establishing themselves in AI search right now are setting citation patterns that will be expensive to dislodge later.

That argument has a follow-on problem. The AI search measurement infrastructure hasn’t caught up to the urgency. Most marketing leaders are making investment decisions in GEO, or the work of getting your brand named and cited inside AI-generated answers.

Right now, they’re doing so on data they don’t fully trust, with tools that overpromise and underdeliver on precision. The temptation is to either wait for perfect measurement or overtrust imperfect data.

Both are mistakes. Both are wrong. The right move is to develop the discipline to distinguish what you can measure from what you can only pattern-detect, and to make decisions accordingly. That discipline is the new maturity test for marketing leaders operating in this category.

The trust gap is real

In the State of GEO in B2B Marketing study I coauthored with Demand Metric, 73% of the 225 B2B marketing and revenue leaders surveyed said they measure AI referral traffic, but only 34% highly trust those numbers.

That gap points to a discipline that is scaling faster than its standards. Most GEO measurement tools provide a visibility score, but understanding what it means requires learning an entirely new vocabulary.

These are the building blocks of GEO visibility scores:

  • Prompts: questions submitted to AI models
  • Answers: the responses those models generate
  • Mentions: instances where your brand appears
  • Citations: URLs the model references

GEO tools run prompts across multiple AI models, then analyze the answers and citations to calculate your visibility score. A visibility score is useful for spotting patterns and directional changes but not yet reliable enough to bet your quarterly forecast on.

Unfortunately, the 34% trust number is probably too high, for structural reasons. Various LLMs are at different stages of driving trackable referrals to websites.

Take, for example, ChatGPT, which began including links to brand websites when recommending them in May 2026. LLMs are uniquely secretive about these kinds of updates, and they only surface when companies find and write about them, as Profound did.

Claude presents a different challenge. Unlike ChatGPT, it typically doesn’t include clickable links to company websites in its standard interface. When Claude recommends your brand, there’s often no referral for your Google Analytics to capture. Instead, users get their short list and then Google the brand.

That visit appears in your analytics as branded organic search rather than AI referral traffic. The influence is still there, but the attribution isn’t. AI helped the buyer discover your company, yet your analytics credit Google instead. As Claude adoption continues to grow, an increasing share of AI-driven discovery will be literally invisible to the tracking infrastructure most marketers have today. So 34% trust is probably the high-water mark, not the floor.

Pattern detection vs. performance measurement

The most useful mental model I’ve found for navigating this is to clearly separate two categories of data. Marketing teams are mistakenly reporting both categories with the same level of confidence, but this overstates the precision of one and undersells the reliability of the other.

Pattern detection

The first category is AI-answer pattern detection, including:

  • GEO visibility scores: Do you appear at all, and how often?
  • Citation footprint trackers: Which specific sources get cited, and are you among them?
  • Share-of-voice metrics: How often you appear versus competitors?

These show whether visibility is changing and in which direction. Treat them accordingly as directional indicators, not forecasting tools.

Performance measurement

When someone clicks through from an AI answer, you can follow them like any other visitor:

  • AI-referred traffic volume: Visitors who arrive by clicking a link inside an AI answer (if you’ve separated it out in your analytics).
  • Traffic conversion: Your conversion rate, average deal size, and time to close—each compared with your baseline.

Your chief financial officer will fund against these numbers, because they connect cleanly to revenue. They still miss a meaningful portion of AI-driven influence (the Claude problem above), but what they capture, they capture reliably.

What to actually trust right now

If you’re a marketing leader trying to operate in this gap, the practical framing is to track everything, trust selectively, and report transparently about which is which.

Track everything because visibility tools are valuable early-warning systems. If your share of voice in ChatGPT responses for “best [your category] vendor” is rising or falling, that’s worth knowing even if the exact number isn’t precise.

Trust selectively because the harder business signals (including referral traffic conversion rate, deal size from AI-referred prospects, time to close, and especially branded search volume trends) are the data points you can actually defend in a budget conversation.

If your branded queries in Google Search Console are climbing without a corresponding lift from your paid acquisition or PR, the most likely explanation is AI-driven discovery you can’t see directly. That’s a measurable proxy for a nonmeasurable cause.

Report transparently because the credibility of your GEO program internally—with your chief financial officer, your CEO, your board—depends on whether you’re seen as the leader who knows the difference between a pattern and a measurement, or the one who got sold on dashboards that turned out to be less reliable than they looked.

The maturity test

Smart leaders in GEO will be the ones who develop the discipline to act decisively on imperfect data, while clearly labeling for themselves and their teams which parts of that data are reliable and which are directional.

That discipline is harder than it sounds because it requires resisting two equally tempting mistakes: waiting for certainty that isn’t coming or accepting confidence that isn’t earned. The leaders who learn to operate in the middle—with eyes open about what they know and what they don’t—will enjoy compounding advantages that have very much to do with their judgment, not their budget.

Andrea Lechner-Becker is the chief strategy officer at GNW Consulting.