Found in AI Ep 6: Measuring GEO: 5 Metrics That Matter for AI Search Success
Google Is Moving From AI Search to AI Agents – Found in AI: AI Search Visibility, SEO, & GEO
Welcome to Found in AI, the podcast about getting your business found in AI search.
In this episode, I sit down with Kristina Frunze of WebView SEO to break down the metrics that actually matter for Generative Engine Optimization (GEO).
You’ll learn:
• Why AI traffic is still just 1% of overall website visits and converts at the same rate as organic
• Why AI share of voice is more important than traffic alone, and how to measure it against competitors
• How to think about AI citations as the new backlinks and why brand mentions matter more than links
• Which types of bottom-of-funnel content drive AI traffic leads (competitor comparisons, alternatives, integrations)
• What AI sentiment means, why it’s tricky to control, and the steps brands can take to influence it
Transcript: Measuring GEO: 5 Metrics that Matter for AI Success
Cassie (Host):
Only one percent. Seriously—one percent. That’s how much overall website traffic is currently coming from AI tools, according to a recent BrightEdge study.
And yet every marketer I know, myself included, is trying to figure out how to measure success in AI search.
Hey, if we haven’t met yet, I’m Cassie Clark. I’m a fractional content strategist and the host of Found in AI, the podcast about getting your business found in AI search.
In today’s episode, I’m joined again by Kristina Frunze of WebView SEO. Last time we talked about whether SEO was dead—spoiler: it’s not.
This time, we’re digging into the metrics that matter for GEO: AI traffic, citations, share of voice, leads, and sentiment. Here’s part of that conversation.
Cassie:
Since last time you and I talked, you pushed back on the “SEO is dead” theory. Has anything shifted since then?
Kristina:
I don’t think so. In fact, I think SEOs who declared SEO dead have been proven wrong. I saw a study recently from BrightEdge—maybe Amazon, I’d have to double-check—that said AI referral traffic is still just about 1% of overall website traffic.
That means organic traffic is still driving the vast majority of both traffic and leads.
What’s interesting is that before, I assumed AI traffic would convert better than organic. It felt logical: people using AI search had already vetted you before clicking through, so they’d arrive more prepared to convert. But the data doesn’t support that. A study from Omze found AI traffic converts at about the same rate as organic—roughly 4.5%.
So SEO is still strong. It’s still the foundation.
Cassie:
That’s interesting—and honestly, it’s why I started this podcast. Everything seems to change with the wind. I’d heard—and even thought myself—that people coming through ChatGPT or other AI search engines would be more qualified and ready to buy. But that’s not necessarily true.
Kristina:
Exactly. Maybe that will change as AI traffic volume grows, but right now, conversions are basically the same.
Cassie:
So what do marketers do? Do we keep trucking ahead with our current strategy, assuming AI searchers are more qualified? Or go back to pre-AI tactics?
Kristina:
It’s a mixed bag. A lot of people are confused: “Do I do SEO, or do I do AI SEO?” The answer is both. Any modern SEO strategy should consider the AI search landscape. You can’t exist in a vacuum.
Cassie:
You posted on LinkedIn this morning about GEO metrics. Of the five you listed, which one do you think will emerge as the number-one metric most companies should track?
Kristina:
The easiest to track right now is AI traffic. It feels familiar because we already know how to measure organic traffic. But the problem is not all LLMs pass referral parameters properly, so the picture is incomplete.
So yes, AI traffic will be the most popular—but not the most important.
For me, AI share of voice is the most important. It shows how often your brand appears versus competitors in AI responses. If you only look at traffic and citations, you miss the bigger picture.
Cassie:
What’s your favorite tool so far?
Kristina:
It’s hard to say because some of the best tools I’ve tested aren’t live yet. But I like ziptidev. It shows your AI share of voice and breaks it down by citations (when an LLM names your brand) versus domain mentions (when they link to your website). That’s been really useful in client audits.
Cassie:
And how do you explain “share of voice” to non-marketing stakeholders?
Kristina:
I remind them how the landscape has changed. In Google, you used to get ten blue links. Now people are seeing AI Overviews or going straight to their favorite AI tool.
So share of voice is about how often your brand comes up compared to competitors, and how much “real estate” you own in LLMs.
The tricky part is that AI answers aren’t static. If you run the same prompt in GPT three times, you might get three different answers. Some tools now test prompts multiple times and average the results to give a more accurate picture.
Cassie:
The third metric you shared was AI citations. Do you treat paraphrase mentions without links as wins, or only direct brand mentions?
Kristina:
For now, I’d say direct brand mentions are the best way to track citations. Tools are evolving and not perfectly accurate, but brand mentions—whether linked or not—are the closest equivalent to backlinks in AI SEO.
Cassie:
Chris Pantelli told me backlinks still matter as a way to earn those citations. Do you agree?
Kristina:
Yes—but in AI SEO, the mention itself is often more important than the link. You want your brand discussed in third-party sites, communities, directories—anywhere LLMs pull from. That ties into the fifth metric: sentiment.
Cassie:
So let’s talk about sentiment. How much control do brands realistically have over how AI describes them?
Kristina:
Not a lot. AI hallucinates features, gets facts wrong, sometimes even adds negative information. There are tools to track sentiment, but influence is limited.
What I recommend is clarity: make sure your product pages are detailed; your information is syndicated consistently across directories; and you address negative reviews. Google Business Profile (GBP) information is already being pulled into AI Overviews, so it matters.
And while Reddit is one of the most cited domains on ChatGPT, it’s also a brutal community. If negative threads pop up, the best option is to engage directly, make it right with the user, and treat it like managing a Yelp or Google review.
Cassie:
Okay, metric four: AI traffic leads. Are you seeing patterns in what types of content actually convert from AI-driven traffic?
Kristina:
Yes—bottom-of-funnel content works best. Competitor comparisons, alternatives, integration pages. People running those queries are already evaluating options, so if your content positions you factually but favorably, it can drive conversions.
Cassie:
Of the five metrics, do you think they’ll standardize like SEO metrics did, or stay fragmented?
Kristina:
I don’t think SEO metrics were ever truly standardized. And GEO has even more complexity with different LLMs—some do web search, some don’t. Tracking will get better, but I think the key will be narrowing focus.
Personally, I’d choose three: AI share of voice, AI traffic, and AI leads.
Cassie:
That’s it for today’s episode of Found in AI. Huge thanks to Kristina Frunze for breaking down the five GEO metrics. If you’re a marketer wondering how to measure success in AI search, I hope this gave you a clearer picture and a few ideas to test.
Quick promo before we wrap: we’re officially in beta with ThoughtTree, the workspace my team and I are building to organize your prompts, reports, and templates into repeatable workflows. If you’d like to test it out and help shape the platform, you can sign up for early access at thoughttree.io.
And here’s this week’s experiment: pick one of your priority keywords and run it through three AI tools — Google AI Overviews, ChatGPT, and Perplexity. Note how often your brand shows up, and repeat the search a few times to see how the answers shift.
If you want a structured way to track results, I’ve built a free AI Visibility Tracker on my blog at cassieclarkmarketing.com. I’ll link it in the show notes.
And if your team is ready to go deeper — building the kind of content engine that gets you found in AI search and drives pipeline — that’s exactly what I do as a fractional content strategist. You can learn more and get in touch at cassieclarkmarketing.com.
I’ll be back next week with Romana Kuts of SaaStorm, who’s been helping clients add FAQ schema to their blog posts and seeing great results. She’s sharing exactly how and why it works, so don’t miss it. See you then.
subscribe to my newsletter to get insights like this straight in your inbox
