There’s a lot of noise when it comes to AI search, visibility, how it’s measured and how much it really matters. Janaina Barreto-Romero, Senior Technical SEO at Oncrawl, had some great insight into how to quiet the noise and what we should really be looking at during her recent talk at BrightonSEO. But the conversation didn’t stop there.
On June 18, Janaina hosted a roundtable on the subject with four search and content experts: Rejoice Ojiaku (Content Senior Specialist at Wise), Sara Taher (Senior SEO Analyst at CBC), Andreas Voniatis (Founder of Artios), and Patryk Wawok (Founder & Head of Technical SEO at Organic Hackers).
The premise was simple. Nobody can see all the cards in AI search yet, so what can you actually do with the hand you’ve been dealt? Here is what stood out.
Traffic is no longer the whole story
Revenue has long been the real measure of business success, so it was where the panel started. Clicks may be down, but is revenue?
The speakers kept returning to one idea in particular: falling traffic is not the same as falling business.
Patryk shared a case where his team removed about 92% of a site’s pages, many of them out-of-stock products indexed for years. Traffic dropped roughly 30%, but SEO revenue rose around 50% because the remaining pages ranked for more valuable terms. As he put it, traffic has quietly “become a vanity metric.”
Rejoice framed the harder part as a people problem, not a data problem. The job now is reframing years of stakeholder conditioning around ranking, clicks, and sessions and segmenting traffic by intent so a drop in low-value informational visits is not mistaken for a business decline.
The technical fundamentals matter more, not less
There was strong agreement around the idea that technical SEO has not been reinvented, but the stakes have risen. LLM crawlers handle JavaScript far worse than Google does, so Sara and Patryk both pointed to rendering, crawlability, and trimming low-value pages as foundational for AI visibility.
Andreas added a newer layer: entity optimization through schema and emerging formats like llms.txt and Markdown. His closing takeaway was clear: get serious about entities now.
Brand presence: AI visibility is everyone’s job
When someone asks an AI system about your brand, the answer is built from the whole web, not just your site. Rejoice argued that this exposes an old problem, with SEO, brand, and PR working toward the same goal while measuring it differently.
Her recommendation was to treat AI visibility as a shared brand presence problem rather than an SEO channel to own alone.
The panel also got practical on how to fix bad answers in AI search. Patryk traced inaccurate AI responses back to neglected sources like a forgotten Trustpilot profile full of unanswered reviews.
Andreas noted a recent German court ruling against Google over an AI Overview hallucination, a sign that accountability is shifting.
Sara’s rule of thumb tied it together: “The best strategy is just being right.”
The one-line version
There was a lot said in one hour, but Rejoice summed it up well:
“AI hasn’t broken search. It’s changed the customer journey.”
The fundamentals still hold; what’s evolving is how we measure success.
These are just a few highlights from the conversation. Watch the full replay for a deeper dive into what these experts are seeing and how they’re adjusting.
