In this week’s episode of Voices of Search, we spoke with Eli Schwartz, growth adviser, strategic consultant, and author of Product-Led SEO, about why the industry’s obsession with AI visibility tracking is missing the point. With companies spending heavily on dashboards that count ChatGPT mentions and citations, Eli challenged the premise that visibility metrics matter if they don’t drive qualified traffic, pipeline, and revenue. 

Having helped companies like Tinder, Coinbase, LinkedIn, and WordPress generate millions in organic revenue, Eli brought a refreshingly contrarian perspective to the AI search debate. He explained why SEO teams need more swagger when reporting their impact, why good products trump marketing manipulation in an AI world, and why the fundamentals of user-focused optimization matter more than ever as search interfaces evolve.

Key Takeaways From this Episode:

• AI visibility tracking resembles old rank tracking schemes that charged per keyword, measuring prompts instead of actual customer behavior and business outcomes.
• SEO teams need to claim credit for brand traffic and first-click attribution the same way paid teams use last-click, because two-thirds of conversions often have an SEO touch in their journey.
• Most AI prompts are deeply personal and won’t appear in traditional tracking because users search directly for brands after getting AI recommendations, creating dark traffic that’s impossible to measure.
• Quality beats manipulation in AI search because LLMs understand intent rather than matching strings, meaning bad products with good marketing will be filtered out, while good products with basic SEO will be found.
• The future of search isn’t about new acronyms like GEO or AEO but about the same SEO teams adapting to new interfaces, including wearables, cars, and home assistants, where search requests will originate.

The Great Visibility Spending Spree

An entire industry has emerged selling AI visibility dashboards, promising to track your brand’s mentions in ChatGPT, citations in Google AI Overviews, and appearances across various LLM platforms. Companies are spending significant budgets on these tools, desperate to understand their AI search presence. But Eli sees a fundamental problem with this approach.

“I would push back against the way the industry is approaching visibility right now. I think they’re getting too granular on what visibility means, and it reeks of rank tracking from many, many years ago,” Eli said. “We’re back to that place where now we’re measuring prompts instead of measuring customers.”

The comparison to legacy rank tracking is damning. Years ago, SEO tools charged clients based on the number of keywords tracked. If you wanted to track “survey,” you’d be charged again for “surveys” and again for “online surveys.” The business model incentivized tracking more keywords regardless of whether they drove meaningful results.

Today’s AI visibility tools operate similarly. They charge based on prompts monitored, creating incentives to track more variations without proving those prompts represent real user behavior or drive actual conversions. Eli pointed out several critical flaws in this approach.

“You don’t know if any of those prompts are real. They might not even be real. You don’t know where in the funnel those users are,” Eli explained. “In many cases, I’ve had access to these tools. I can’t replicate the prompt experience that the tools are claiming that I had. Like, I don’t see the websites that the tools say were there.”

This lack of verification creates a dangerous situation where teams optimize for metrics that may not exist, targeting visibility in AI responses that either aren’t happening or don’t drive business value.

The Integrity Problem: Mystery Meat Input Factors

One of the most significant challenges with AI visibility tracking is the complete lack of transparency around input factors. Unlike traditional SEO, where Google provided keyword data through Search Console and advertisers got detailed query information through paid campaigns, LLM platforms share virtually nothing.

“Today, there isn’t a really clear way to get data on what’s happening in an LLM. Today, ChatGPT isn’t kindly sharing with us what all the prompts our consumers are putting in,” Jordan noted during the conversation. “Toda,y there isn’t some URL identifier that tells us what the prompt could have been or the intent of the prompt could have been before they landed on our page. None of that exists.”

Eli’s response was blunt: “They’re not.”

This creates a fundamental attribution problem. The tracking tools claim to show how your brand appears in AI responses, but without access to actual user prompts or verification from the LLM platforms themselves, these metrics are essentially unverifiable. You’re being asked to optimize for and report on metrics you can’t independently confirm.

The contrast with traditional SEO is stark. Google sells keyword data through its advertising platform. Search Console provides real query information tied to actual clicks. These data sources have verifiable connections to real user behavior and business outcomes. AI visibility tracking has none of these verification mechanisms.

Why Most AI Prompts Are Invisible and Personal

Even if visibility tracking tools had perfect data access, Eli argues they’d still miss the majority of meaningful AI search behavior because of how people actually use these tools.

“I think that most of the things that I do specifically in Gemini or even AI mode, they’re not search queries and they’re very, very personal to me,” Eli shared. “I use Gemini to find a paint color of a wall that I’m going to buy the paint for. Gemini told me what paint color is, and I’m going to buy the paint, but that website is going to have no idea that I started with Gemini.”

This example illustrates a critical gap in AI search attribution. The user asks Gemini for paint color matching advice, gets a recommendation, searches for that brand directly on Google, and makes a purchase. From the paint company’s perspective, this looks like branded search traffic. There’s no indication it originated from an AI interaction.

“If I think about that from a query standpoint, the companies that I’m going to buy that paint from, there’s nothing they could have done to show up and have their paint color match the paint of my wall if they didn’t deserve to be there,” Eli explained. “Everyone’s trying to put, ‘Oh, we want to show up,’ like I don’t know, paint color matching. Well, you don’t sell the right paint. Don’t show up. I don’t want an LLM that gives me the wrong paint.”

This creates massive dark traffic that will never appear in any tracking tool. Users get AI recommendations, then conduct direct branded searches or navigate directly to websites. The AI interaction remains invisible to traditional analytics, and the ultimate conversion appears to come from branded search or direct traffic.

“This stuff is personal, and it’s going to lead to so much direct search, dark searches that no one’s going to be able to track,” Eli predicted.

The Citation Manipulation Trap

As Reddit citations dominated early AI responses, marketers rushed to game the platform. When certain sources appeared frequently in ChatGPT outputs, agencies quickly pivoted to offer “Reddit marketing” or “Quora optimization” services. Eli sees this as a repeat of the backlink manipulation that plagued traditional SEO.

“The way backlinks and citations should have always worked is if you’re legitimate, you earn the right backlinks, and that’s what Google wanted,” Eli explained. “When Google measured the worldwide web, before marketers went and bought backlinks and poisoned the entire ecosystem by manipulating backlinks. Now we’re getting back to a world where that actually matters again.”

The problem with focusing on specific citation sources is that AI platforms will adapt as manipulation increases. “You get this data set that says, ‘Oh, Reddit shows up very often in an LLM citation.’ So then CMOs and agencies rush into, like, let’s go and game Reddit. But then I think the LLMs are learning and they’ll say well, Reddit is not really a reliable source anymore because it’s been gamed.”

This creates a perpetual cat-and-mouse game where marketers chase whatever source currently appears most frequently in AI citations, only to have that source devalued as manipulation increases. Eli advocates for a completely different approach.

“Actually, what I advise clients is to think about this as quantity over quality. The more citations you get, the more likely some citation will matter,” Eli said. “If you throw all your eggs into the Reddit basket and then Reddit doesn’t matter anymore, you have no citations. But if you think about massive amounts of citations, which again is regular marketing—if you do good stuff and people talk about you, you’ll have quantity.”

This returns to fundamental marketing principles: create genuinely valuable products and experiences, and citations will accumulate naturally across many sources. Trying to manipulate specific platforms for AI visibility is both harder to sustain and more likely to backfire as platforms adjust their algorithms.

The Share of Voice Alternative

Rather than obsessing over granular prompt tracking, Eli recommends focusing on broader share of voice metrics that indicate competitive positioning.

“I want to think about broader visibility, which is a share of voice and kind of harder to measure. But I like what Semrush has done. I like what Ahrefs has done,” Eli said. “Here are the general words you have. Here’s some sort of share of voice where we’re normalizing versus all your competitors.”

This approach looks at whether you appear in the same contexts as your competitors rather than trying to optimize for every possible prompt variation. If you’re a payroll company like Gusto, the question isn’t whether you appear for “California payroll software platform.” It’s whether you appear wherever your competitors, like ADP and Paychex, appear.

“Take example queries and say share of voice—are there words where ADP is showing up, and Gusto is not? Are there words where Paychex is showing u,p, and Gusto is not? That would be a negative share of voice,” Eli explained. “If Gusto is showing up all the places the competitors are showing up, that’s what you want.”

This competitive context matters more than absolute prompt counts because it indicates whether you’re in the consideration set for relevant queries. Share of voice metrics also benefit from being relative rather than absolute—they normalize for overall search volume changes and AI adoption rates.

“Instead of getting too granular where we get into the rank tracking space, what are the general prompts and general words that you see competitors showing up for and make sure that you’re there,” Eli advised.

SEO Teams Need More Swagger

One of Eli’s most passionate arguments is that SEO teams systematically undersell their contribution to business outcomes, especially compared to paid marketing teams.

“I don’t think that SEO teams have enough swagger. SEO teams speak a different language. They come in, and they start talking about black boxes and algorithms and LLMs and all sorts of dumb ranking things that SEO teams talk about,” Eli said. “Then you have SEM teams, and they talk about dollars and ROI and LTV, and that makes a lot of sense. CFOs love it.”

This communication gap creates budget allocation problems. When SEO teams can’t articulate their value in business terms, they lose out to channels that speak the CFO’s language, even if those channels deliver less actual value.

Eli shared a powerful example from his time at SurveyMonkey: “I noticed that the SEM team had a very easy time getting budget, and the SEO team did not have an easy time getting budget. The SEM team was responsible mostly for brand bidding and brand traffic. They were using the last click. It’s very difficult to do blended attribution.”

His solution was to model SEO attribution using first-click rather than last-click: “I had our BI team model out SEO first click, and when I had them model out first click it turned out that I was responsible from a first-click standpoint for $200 million in revenue. Two-thirds of revenue had an SEO touch as the first part of their journey.”

Armed with this data, Eli could start budget conversations from a position of strength: “SEO is responsible for two-thirds of our conversions. Now I’d like to talk about another head, and another head was going to cost me $200,000 or $300,000 all in. A lot easier to have that conversation when I’m talking about $200 million to start.”

The lesson applies broadly: SEO teams should claim credit for their full contribution to customer journeys, including brand traffic and first-touch attribution, rather than limiting themselves to non-brand last-click conversions that understate their impact.

The Brand Traffic Double Standard

Eli called out a systematic double standard in how different marketing teams report their results. Paid search teams routinely take credit for brand traffic despite the fact that users would have found the company anyway. SEO teams, meanwhile, often exclude brand traffic from their reporting.

“I’ve never ever met an SEM team that doesn’t take credit for brand traffic. They never walk into a meeting and say, ‘We spend $4 million a month. $3.9 million of that is our own brand name where we’re being extorted by Google,'” Eli said. “They’re just saying we’re a $4 million team and look at our LTV from our $4 million.”

SEO teams do the opposite, focusing exclusively on non-brand keywords: “When it comes to SEO, the SEO team — they don’t report on brand. They’re just like, ‘Here’s all our non-brand keywords.'”

This creates a skewed perception of channel value. The paid team gets credit for the full $4 million in spend and associated revenue, even though the vast majority represents brand bidding. The SEO team only claims credit for a small slice of non-brand traffic, even though their efforts contribute to brand awareness that drives both paid and organic brand searches.

Eli’s advice is simple: adopt the same reporting standards as other channels. “SEO is part of this holistic marketing effort. SEO is benefiting from brand. Brand is benefiting from SEO. Paid is benefiting from all of it. Both of those teams, SEM and SEO, are benefiting from the efforts that the brand team does.”

Report on total SEO traffic, including brand, then break down the brand versus non-brand components deeper in your analysis. This gives executive teams a complete picture of SEO’s contribution while still maintaining transparency about traffic sources.

Product Quality Beats Marketing Manipulation

One of the most optimistic aspects of Eli’s perspective on AI search is his belief that it will reward genuinely good products over clever marketing manipulation.

“We’re in a world where you kind of have to be the same. You can’t have bad products and have good marketing. You can’t have bad marketing on good products. It comes together,” Eli explained. “AI is helping that out. One of my favorite things about AI from a user standpoint is you can write the stupidest prompts ever, but then the LLM figures out what you want from an intent standpoint.”

This intent understanding changes the game. In traditional SEO, you could rank for keywords through technical optimization and backlink building, regardless of whether your product actually deserved to rank. AI search makes this harder.

“If you have a great product but you’ve done very bad marketing, AI will still help you find it. If you have a really bad product but you’ve done really good marketing, AI will squash it, and you won’t be found,” Eli said.

The implication is profound: focus on building genuinely great products and customer experiences first, then ensure basic SEO hygiene so AI can find and recommend those products. Trying to manipulate your way into AI visibility without the underlying product quality is a losing strategy.

“The secret is to have a good product and have good experiences with your customers and then make sure that there is awareness baked into this good product experience,” Eli advised.

The Fundamentals Still Matter

Despite all the changes in search interfaces and AI adoption, Eli’s core advice remains surprisingly traditional: do good SEO fundamentals.

“Make sure that you follow best practices around SEO. Have a website that’s crawlable. Have content that talks about your products,” Eli said. He shared a concrete example from his work with Tinder that illustrates this point.

“Tinder didn’t have any pages about their subscription products. When someone looked for Tinder Platinum, they found websites that ranked all dating apps, and they talked about Tinder’s product that Tinder didn’t have a page on. When we created these pages for Tinde,r magically we ranked number one, and we converted a crazy amount from these pages.”

This same principle applies to AI search. If you don’t have pages describing your products, AI engines can’t recommend them properly. Users will find competitor sites or aggregators talking about your products instead of going directly to you.

“Make sure your product pages exist. Make sure that you follow SEO and search best practices to make sure they’re visible. You don’t have to follow technical best practices like you used to years ago because the engines will make up for it, but that’s still what I would recommend,” Eli said.

The good news is that AI engines are becoming more forgiving of technical imperfections. Their ability to understand content despite technical issues means you don’t need perfect schema markup or flawless site architecture to be found. But you do need the basic content to exist and be accessible.

The Annual Planning Problem

Eli called out a practice he sees repeatedly in SEO teams that wastes enormous resources: creating elaborate annual plans that no one ever checks.

“I would meet companies, and they would say we need you to come in and help us plan for our SEO year,” Eli recounted. “I asked them questions. Who’s the audience? Who are you presenting this annual plan to? What happens if you hit your annual plan? What happens if you miss your annual plan?”

The answers revealed the problem: “The overwhelming answer to that question was that the audience is the executives. They sit there in the meeting on their phones and totally ignore the presentation. And then if I hit the annual plan, I don’t know, nothing happens because no one ever checked. And if I miss the annual plan, same. It’s just kind of a plan.”

SEO teams spend a quarter of their year creating plans that become shelf-ware. Meanwhile, product teams can’t afford this luxury because they’re allocating engineering resources that would sit idle without clear direction.

“Product teams don’t do that because product teams are allocating resources and they do stuff and they say they’re launching something and the engineers are sitting around doing nothing if they don’t actually do that,” Eli explained.

His advice is to create annual plans only if they’ll actually drive resource allocation and accountability: “Annual plans should be something where you’re actually going to present something, and you’re going to stick to i,t and this is what you want to do and how you’re going to do it, and then you can be successful.”

Otherwise, skip the elaborate planning document and focus on executing tactical improvements that demonstrably move business metrics.

Why New Search Interfaces Won’t Kill SEO

Despite predictions of SEO’s death, Eli sees search evolving rather than disappearing. The key insight is that search represents a fundamental human behavior: pulling information when you need it, rather than having it pushed at you.

“I don’t think search goes away. I think that there will always be two aspects to search, which is you’re pushing something at someone, and organic, which is someone’s pulling. We are always going to live in a world where you pull things where I make requests,” Eli explained.

What’s changing is the interface through which those requests happen. Today it’s mostly computers and phones. Tomorrow, it will include cars, wearables, smart home devices, and interfaces we haven’t imagined yet.

“Very, very soon it’s going to be cars, it’s going to be wearables, it’s going to be other—we’re already getting into this—Gemini is now baked into all the Google products that Google sold like the home assistants,” Eli noted.

These new interfaces create interesting attribution challenges. “Now I’m going to go and talk to my thermostat and say, ‘I need a new electric company because my thermostat is making me pay too much money.’ And it’s going to tell you the name of a company. And then you’re going to pull out your phone, and you’re going to Google the name of that company. That brand is never going to know it started with an LLM search.”

But the fundamental work remains the same: ensuring your brand and products are findable when people look for solutions to their problems, regardless of which interface they use to conduct that search.

The Acronym Debate: SEO vs. GEO vs. AEO

There’s been a proliferation of new acronyms attempting to rebrand search optimization for the AI era: GEO (Generative Engine Optimization), AEO (Answer Engine Optimization), AIO (AI Optimization). Eli argues this misses the point.

“Should it be called SEO? I wrote a LinkedIn post about how it shouldn’t be called SEO because it’s putting the onus and the focus on the search engine rather than the focus on the user. But that’s the name we have and the team we have, the agencies we have, and the consultants we have, they’re going to be responsible for this new way of search,” Eli explained.

The parallel he draws is to mobile search. “Many years ago, like 15 years ago, we didn’t create a new acronym for mobile engine optimization. We just stuck with SEO. We’re going to keep the acronym for LLMs. We’re going to keep this acronym for wearables.”

The SEO team, whatever they’re called, will continue to be “the team that architects and optimizes visibility on whatever these new platforms are.” Creating new acronyms and hiring new specialists is unnecessary and often counterproductive.

“Anybody who has the best user empathy and understands the way these platform engines work should be able to adapt to whatever the new future is. So you don’t need to invest in a replacement,” Eli said. “If you can’t trust the SEO team, you have to do the AEO, AIO, GEO, whatever it is you have, then don’t trust them to do SEO.”

The Million-Dollar Question: What Should Companies Do?

After dismantling the case for AI visibility tracking and citation manipulation, the obvious question remains: what should companies actually invest in?

Eli’s answer returns to foundational principles: “Have a good product and have good experiences with your customers and then make sure that there is awareness baked into this good product experience.”

This breaks down into specific actions:

First, ensure basic SEO hygiene. “If you have a really good product but you don’t have a website, then it’s going to be hard for you to be visible in search. If you have a really good product but you don’t go and do proactive outreach to make sure people talk about your good product, you may have word of mouth, but you may be missing some of this word of mouth that needs to end up on search.”

Second, create content that describes your products. Many companies, even large ones, lack basic product pages that would help both traditional search engines and AI platforms understand their offerings.

Third, focus on customer experience that generates organic citations. “Create great experiences for customers that people want to talk about, and that’s citations. Real brands don’t need to go out and seek citations.”

Finally, maintain perspective about what’s changing versus what’s staying the same. “The engines will make up for technical imperfections. You don’t have to follow technical best practices like you used to years ago. But you do need to follow SEO and search best practices to make sure they’re visible.”

Staying True to Your Beliefs

In a year Eli described as his most challenging as a consultant, he faced constant pressure to pivot toward AI visibility services that he doesn’t believe deliver value.

“I talk to potential companies, and they say, ‘Well, we don’t want to do SEO, but if you’re able to help us with LLM visibility, that’s all we want,'” Eli shared. “My answer shouldn’t be, ‘Well, do SEO and you’ll be visible in LLMs because then I lose that prospect.'”

But losing prospects by maintaining integrity beats winning projects by selling services you can’t deliver. “If you go and sell something that you can’t actually accomplish, you’re on the hook for it in a few months when it goes away.”

This advice extends beyond consultants to in-house SEO teams facing pressure to rebrand as GEO or AEO specialists. “Stay true to your ideals. If you don’t believe in it, don’t bend and start calling yourself an AIO consultant and recommending things that you don’t believe in.”

The industry will continue evolving, new interfaces will emerge, and search behavior will adapt. But the fundamental principles of understanding user intent, creating valuable content, and building genuinely good products will outlast any specific platform or acronym.

The Bottom Line

The rush to track AI visibility represents a familiar pattern in digital marketing: measurement for measurement’s sake, divorced from business outcomes. Companies are spending heavily on dashboards showing ChatGPT mentions and citations without establishing whether those metrics correlate with traffic, leads, or revenue.

Eli’s perspective cuts through the hype with a simple premise: if AI visibility doesn’t drive qualified traffic and conversions, it doesn’t matter. And given the deeply personal nature of most AI interactions, the prevalence of dark traffic after AI recommendations, and the lack of verifiable input data, most AI visibility metrics are measuring the wrong things.

The alternative isn’t to ignore AI search. It’s to focus on fundamentals that work regardless of interface: build great products, create helpful content, ensure basic SEO hygiene, and let citations accumulate naturally from genuine customer satisfaction. These principles worked before AI search, they work today, and they’ll continue working as search interfaces evolve.

As Eli put it: “Search will always exist. Search engines will always exist. The UI will change. This is the team that architects and optimizes visibility on whatever these new platforms are.”

The future of search isn’t about new acronyms or expensive visibility dashboards. It’s about the same teams doing the same foundational work, adapted for whatever interfaces users adopt next.

Voices of Search is a daily SEO and content marketing podcast hosted by Jordan Keone and Tyson Stockton. The show delivers actionable strategies and data-driven insights to help marketers navigate the ever-evolving world of search engine optimization and content marketing. New episodes air weekly, covering everything from technical SEO to AI discovery, featuring industry leaders and practitioners sharing real-world frameworks and proven tactics.

Subscribe to Voices of Search on Apple Podcasts, Spotify, or your favorite podcast platform. Follow Previsible on LinkedIn for updates and subscribe to the VOS YouTube channel for video episodes and clips. You can also visit the official VOS site to explore the full episode archive and submit your SEO questions for future episodes.