by 
Three new studies are making the rounds as evidence that AI search visibility may not be worth investing in. AI assistants account for far fewer website visits and conversions than Google organic. Search Engine Roundtable‘s Barry Schwartz suggested that spending time and money on AI visibility might not be worthwhile.
The traffic numbers are compelling. The conclusion isn’t.
For a B2B company, the important question isn’t just how many people click a link in ChatGPT. It’s whether your company comes up when a potential customer is deciding which suppliers or partners deserve a closer look. That decision can happen weeks before anyone visits your website, fills out a form or speaks to sales.
Low measurable referral volume is not the same as low potential influence on buyer consideration.
The shortlist comes before the click
This isn’t a niche behavior. In Forrester’s most recent survey of business buyers, generative AI tools were the most frequently cited meaningful source buyers used to research purchases. Buyers also told Forrester they check what AI tells them against peers, experts and vendors. That’s the part referral reports miss: the research can start in an AI tool, while the visit arrives later through a different channel.
Here’s a hypothetical. A growing beverage brand needs a new manufacturing partner. Someone in operations asks an AI assistant which Florida companies can handle shelf-stable beverage filling and packaging. The answer names several manufacturers, with a quick explanation of their capabilities.
Those names go into a spreadsheet. The team checks certifications, production capacity and relevant experience, eliminates poor fits and sends a few candidates to procurement. A month later, a buyer searches one of the remaining companies by name and requests a quote. Google Analytics credits Google organic. The CRM records an inbound lead.
AI didn’t choose the supplier. But it may have helped determine which suppliers got considered. If that company ultimately wins a substantial contract, there may never have been a single ChatGPT referral session in the entire buying journey.
That’s why judging AI’s value by referral clicks alone is a problem, particularly when a B2B buying process involves multiple people, several rounds of research and a handful of high-value opportunities each year.
Our manufacturing research shows how narrow the field can get
At Evoke, we tested 70 unbranded buyer-discovery questions across ChatGPT, Claude, Gemini and Perplexity for our 2026 Florida Manufacturing AI Visibility Report. Across 280 responses, we counted 877 mentions of 54 companies and organizations.
Just 10 companies and organizations accounted for 57.7% of those mentions. Twelve established manufacturers we reviewed didn’t appear at all, despite an estimated $8 billion in combined revenue.
Share of 877 counted mentions across 54 companies and organizations
Those are recommendation patterns, not sales results. But they reveal something a referral report can’t: an AI assistant can repeatedly surface the same small group of names while leaving other qualified companies out of the conversation.
A manufacturer may rank well when someone searches its name, have a credible website and a strong sales operation, yet still be absent when a buyer asks the broader question: who can actually do this work? That gap deserves attention.
The studies don’t settle the business question
Marie Haynes analyzed more than 145 million sessions across 100 sites and found ChatGPT sent only 0.62% as much traffic as Google organic. The SEO Works found organic delivered 39 times more visits, and AI referrals generated just 3.2 conversions for every 100 organic conversions. Google remains the far larger measurable channel. No argument there.
But Haynes’ own conversion data complicates the takeaway. Among the 55 sites tracking conversions, ChatGPT visitors converted at 13.8%, compared with 3.02% for Google organic. That’s more than four times the rate, even if the total number of conversions was much smaller. The SEO Works found weaker AI conversion rates for B2B product companies, so that advantage isn’t universal. Still, counting sessions alone misses differences in the value of those visits.
The SEO Works’ own limitations section makes an important distinction. Google AI Overviews aren’t included in its AI referral numbers because those clicks arrive as ordinary organic sessions. The study also can’t track someone who reads an AI answer and later searches for a company by name. Its authors say the measured AI contribution "should be read as a floor." AI Mode is also part of Google Search, not a standalone assistant referral. Those are significant omissions from a simple AI-versus-Google traffic comparison.
Kevin Indig went further than referral counts, testing whether AI visibility led to increases in branded search or direct traffic. He couldn’t establish a reliable relationship. The SEO Works reached a similar conclusion on direct traffic: once overall site growth was controlled for, more AI referral traffic didn’t reliably translate into more direct visits. Indig recommends combining visibility benchmarks with what customers say about how they discovered a company and what happened commercially.
That’s the distinction. These studies tell us a great deal about measurable traffic. They don’t tell us how often an AI recommendation puts a B2B supplier on a buyer’s initial list.
AI search is part of a bigger picture
This isn’t a new challenge. An industry story might send only a handful of visits, yet help a buyer recognize and trust a company months later. PR has worked that way for decades. That doesn’t make every placement valuable, but it does mean referral traffic alone is a poor way to judge one.
That’s the thinking behind Evoke’s approach: Build Authority. Increase Visibility. Create Demand. Measure Impact. Authority gives buyers a reason to trust a company. Visibility puts it where they’re researching, whether that’s industry media, Google or an AI-generated shortlist. The point is to turn that attention into consideration and opportunities, then find out whether it’s contributing to results.
That’s also how we approach AI search optimization: as part of that broader effort, not a separate program that replaces SEO. Clear capabilities, useful expertise, accurate company information and credible third-party coverage are still the foundation.
What B2B companies should measure
Test the actual questions buyers ask before they know your name. Track who gets recommended, how often you appear and whether the information is accurate. Repeat the tests over time, rather than relying on a favorable screenshot.
Then ask new prospects how they built their vendor shortlist. Don’t stop at "How did you find our website?" Ask which sources, colleagues and tools helped them decide whom to consider. Capture that in the CRM and connect it to qualified opportunities and closed business.
That won’t produce a perfect attribution model. It will give you a better basis for deciding whether AI discovery is contributing to the pipeline than a report showing a handful of ChatGPT clicks.
The wrong conclusion from the right numbers
The studies are a useful reminder not to confuse AI search hype with measurable business performance. But concluding that AI visibility isn’t worth attention because referrals are small gets the question backward.
For companies with long sales cycles and high-value contracts, being considered often matters more than getting an immediate click. If prospective customers are asking AI which firms belong on their shortlist, do you know whether yours is on it?
Want to see who AI recommends when buyers search for companies like yours? Start with Evoke’s free AI Visibility Check.
