How To Win in Search Through AI Discovery

In this week’s episode of Voices of Search, former host Benjamin Shapiro returned to the host’s chair to speak with Jordan Koene, co-founder and CEO of Previsible. 

Ben kicked the conversation off with a specific number: 29%. Search volume across more than one million high-volume keywords fell 29% last year, and the queries behind that drop didn’t vanish. Instead, they moved to resolving customer intent without a single click.

This conversation covered how SEO leaders defend their work when traffic drops, which numbers should replace clicks on the board slide, and why Jordan puts “AI discovery” on his business card.

Key Takeaways From This Episode:

• Visibility is a useful impression count, but it’s a synthetic proxy because AI responses differ for every person
• Falling Google traffic with steady revenue points to an attribution problem rather than lost search value
• ChatGPT visitors arrive pre-qualified, because the model has already educated them using content SEOs built over decades
• New reporting should center on revenue, cited sources, coverage against competitors, and review freshness and volume
• Outdated brand assets such as help content and pricing pages feed models inaccurate information
• None of Previsible’s clients earn more from AI than from Google, so teams have to optimize for two fronts at once

Visibility Is a Proxy, Not a Scoreboard

Ben’s first question was whether visibility is the problem for a team watching organic traffic fall 30%. Jordan’s answer was blunt.

“Visibility is not your problem. Visibility is also not your friend,” he said.

He doesn’t dismiss it. Visibility works like an impression count, a body count of how often a brand shows up in LLM responses, and people who stopped clicking are still searching and finding the brand. What it can’t do is show where search is headed, which is what rankings and search volume once did.

The way it’s built explains why. Visibility tools:

• Generate a set of prompts that stand in for real questions
• Record where brands appear in the mentions those prompts produce
• Combine variables such as citations and the grounding queries a model uses to build its answer

Every one of those prompts is synthetic. Each person uses AI models in their own way, tied to their own data, and gets a different response, so visibility shows a general direction and not what a specific buyer saw. Nothing equals the old certainty of ranking at a known position for a query.

Model changes add another variable. Jordan said his team can track a drop in clicks across clients after a recent ChatGPT release, but fewer clicks doesn’t mean a brand disappeared from the answers.

Falling Traffic Is an Attribution Problem

Ben then put the question the way a CFO would. Traffic is down 30%, so why is SEO still funded? Jordan called it the crisis facing the industry, a battle to justify the value of organic.

With SEO under fire, Ben argued for patience. Revenue is steady, people are using LLMs and can’t be tracked, and direct traffic is climbing, which is the one indicator an SEO can point to. Jordan agreed with the logic and added a caution. Direct traffic is as much a mystery as visibility, so it can’t carry the case alone.

“What you can focus on is where are customers going, what are customers doing, and are customers converting, and why are they converting?” he said.

The why matters most, since it points to the citations and sources models use to justify showing a brand.

ChatGPT is also becoming a real traffic driver. In the scenario Ben laid out, Google is down about 30% while ChatGPT grows a little, Claude barely registers, and other LLMs are negligible. Ben asked whether he should request more budget.

Jordan said he should, with a condition. “The work is different. For Google, teams wrote content, created pages, and kept optimizing, refreshing, and linking them. LLMs don’t work that way,” and he described the moment as a brand renaissance, where teams invest in the elements that help a brand show up inside the answers.

ChatGPT Visitors Arrive Pre-Qualified

On why ChatGPT converts better than other channels, Jordan pointed to what happens before the click.

“The user is already pre-qualified,” he said. “All the work and energy has been done by the model to help inform, educate, refine, shortlist.”

The model gets that information by grounding, going into search engines to retrieve context for its answers. That means it draws on decades of content SEOs built, structured, and attributed. Ben, still in character, took the credit: the conversion rate comes from a strong SEO team.

Jordan added that Google’s own documentation says the best thing to do for AI is optimize for SEO. Quality content, good assets, and good brand mentions that perform in search are also the ones models use to shortlist brands.

The Dashboard Needs New Numbers

Ben pressed the obvious follow-up: what number goes on the board slide when clicks disappear? 

Jordan started by retiring the old ones. Crawl, indexation, and the count of indexed pages were sold to CMOs for years as the path to traffic and revenue, and they no longer explain why a model chooses a brand.

The replacement starts with lagging indicators like where sales happen, which products drive them, and which categories matter most. From there, Jordan wants these on the slide:

• Revenue and its attribution across the site, by product and category
• The cited sources that shape how models describe the brand and its products
• Coverage against competitors in the sources models draw on
• Review freshness and review volume

Reviews matter for a simple reason. A brand whose last review is six months old doesn’t look relevant, and one with less than a third of its closest competitor’s review volume doesn’t look competitive to an LLM.

Jordan said most SEOs spend too long mining for inputs elsewhere when the best clues are already in the company’s own data.

“The inputs are sitting in our own revenue tables, our own sales tables, our own signup tables,” he said.

Your Own Brand Assets Now Carry the Weight

Teams can’t control how many reviews land on a site like G2, but they can partner with other teams, educate them, and control what sits on their own site. In what Jordan calls a brand renaissance, the assets that matter most are the ones that describe the company:

• About pages
• Pricing pages
• Core company information pages
• Support and help content

Reinvesting there can shift sentiment and expectations, and it feeds the material models use to build a response. It also matters because LLMs lack Google’s integrity when it comes to accuracy. Jordan cited a report from an LLM tracking vendor, 50,000 pages long, that Google removed from its index within two weeks. Six months later, ChatGPT still cites it as a source even though it’s false.

Old B2B help content makes the problem worse. He described help centers with 15- to 20-year-old articles carrying outdated pricing and features, and a client that renamed one product four times in five years, with help content spread across all four names.

“What do you expect an LLM to do if that’s the underlying training it has absorbed?” he said. “It’s going to give an inaccurate, incomplete answer.”

The old SEO hygiene work still applies, he said. It now points at more brand-centered assets like help and support content.

Mobile Showed SEOs How to Reset the Scoreboard

Ben asked where SEO has been through a shift like this. Jordan pointed to their time together at eBay, where from about 2009 to 2014 users moved from desktops to mobile devices.

The entire measurement framework changed. Input metrics had to account for mobile speeds and different viewports, and SEOs retrained themselves to think about those KPIs and sell them to engineering VPs and CMOs. 

“Users will move in droves to AI experiences the same way,” Jordan said, “and the models themselves are what matter. Baselining today what a brand has, what sits on its site, and what gets cited will pay off.”

Ben noted what’s different this time. Back then, SEOs could at least track their rankings. Now it feels like throwing darts at a board without knowing where in the room it is.

Follow the Money, Then Fight on Two Fronts

Jordan’s advice to a CMO digesting all this is to follow the money. Previsible has nearly a hundred clients, and none of them make more money from AI than from Google, which remains the largest source of referral traffic and sales.

He also points to scale. Google has 14 products with more than a billion monthly active users, and ChatGPT passed that mark only a few months ago. If he were hitching his wagon, he said, it would be to YouTube, Gmail, Google Search, and Gemini.

“People are still on Google, but more of them are moving to LLMs, so teams have to do both. Optimizing for Google also optimizes for Gemini, and the SEO team remains the bread-and-butter channel. It’s a two-front war right now,” he said.

Revenue and Sources Are the New Scoreboard

Ben closed by asking what Jordan puts on his business card, with GEO, SEO, AEO, and LLMO all in contention.

“We don’t know what these three letters are, and we are still transforming what this industry looks like,” Jordan said. “So until we really know what it is, it’s AI discovery.”

But everything discussed can be traced back to just one shift—clicks no longer prove value, so SEO leaders have to defend their work through revenue, the sources shaping how models describe a brand, and brand assets that stay accurate. Visibility alone can’t carry that case, because it measures a synthetic picture that changes for each person.

None of it means abandoning what already works. Google still pays the bills, ChatGPT visitors convert well because of the content SEOs built, and the work now splits across both. 

Teams that follow the money, keep their own content clean, and report on sources instead of clicks will keep their seat at the table, whatever the category ends up being called.

Voices of Search is a daily SEO and content marketing podcast hosted by Jordan Koene 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.