SORN.AI is an AI SEO agency running content, technical SEO, link building and GEO across SaaS, e-commerce, cybersecurity, hospitality and legal. Their client PagePilot.ai builds AI-powered Shopify landing and product pages, and it arrived with solid rankings, a real content library and almost none of it surfacing in AI answers. Three months later pagepilot.ai was the No. 5 most-cited domain in its category. Strategist Stanislava Smiljanic and cofounder Eri Georgiev walked us through what changed.
Key Results
- A 3% citation share and the No. 5 position among the most-cited domains in PagePilot’s category across tracked prompts.
- AI chat referral sessions up 31.1% month over month in June 2026, from a baseline of roughly 0.7% of organic sessions.
- Average time on site from AI referrals nearly tripled, up 215.9%.
- Traditional search rose with it: organic traffic increased by 27% and average position improved from 9.9 to 6.3.
The problem: strong SEO that AI search could not read
“They had all ducks in a row, but nothing was happening,” Stanislava says. “Both on Ahrefs and in Search Console it was a plateau, a flat line for a really long time. The website traffic was stable, but stable is not always good.”
Three things held it back: AI was not resolving PagePilot cleanly as a brand, the content had no FAQ sections, and the headless CMS supplied no structured data.
“There were these huge blocks of text, and it was ruining the entire experience,” she says. “It wasn’t parsing the content the way we expected.”
Shopify merchants ask ChatGPT or Perplexity for tool recommendations before they open a search engine, and AI platforms were already sending PagePilot roughly 0.7% of organic sessions, close to 9 in 10 (89%) of it from ChatGPT.
What SORN.AI changed
Existing high-performing pages were rewritten first, with net-new comparison and pricing pages treated as a three to nine month play. On the pages: answer the prompt in the intro rather than building up to it, add comparison tables, break text into bullets, name weaknesses openly, and turn tracked prompts into H2 headings.
“What we noticed is that we put these tables, and that is something that is helping LLMs understand what we are talking about,” Stanislava says. “We injected some of the prompts we generated in OtterlyAI into the FAQ segment, and that creates a huge difference.”
Because the CMS supplied no schema, SORN.AI built FAQ structured data custom.
“Structured data is basically translating your entire website to LLMs,” she says. “We added FAQ structured data, and the first leading indicator we saw was appearing in AI Overviews. Then LLMs started citing and following.”
She is blunt about what stopped working. “Articles that are fluffy and 5,000 words long are no longer working. That was back in the dark ages when keyword stuffing was a thing,” she says. “We tried our best to do everything reasonably and not praise PagePilot too much. We wanted to do things objectively.”
Off-page ran on the same claims: community placements and listicles backing what the pages say.
PagePilot’s results after three months
Search quality moved with the volume. Click-through rate rose 2.2 points to 19.4% and average Google position improved from 9.9 to 6.3 in the same month. On Ahrefs, PagePilot’s Google AI Overviews response index reached 58, a gain of 48, with AI Mode responses at 41, a gain of 14.
Against its four tracked competitors, PagePilot now holds the Leaders quadrant of the OtterlyAI Brand Visibility Index on 3% of category citations.
The win Stanislava names first is not in the charts above. “That all-time high on Search Console. We come from an SEO background, and now when everyone is seeing a drop, we are seeing a peak,” she says. “It also coincided with sign-ups doubling.”
Five more accounts, five different blockers
PagePilot started from strength. Most accounts do not, and the blocker is rarely content quality on its own.
Two of the five needed an entity fix first. “A brand name is just a string of characters. The entity holds the context,” Stanislava says. The mortgage case is the cheapest lesson: the crawler block was found in a GEO Audit before any content was commissioned.
“Every client is a story for itself, but we have three pillars that we believe in,” she says. “Solid technical foundations, content tied to your ICP, and finding your missing links.”
How OtterlyAI fits into SORN.AI’s delivery
“It would be very manual, with a ton of print screens and manual research of prompts with VPN,” Stanislava says of the alternative.
- Kickoff. The AI Prompt Research Tool builds the prompt universe from ICP inputs, with bottom-of-funnel prompts added by hand.
- Delivery. Crawlability Checker before budget is committed, Content Checker during rewrites.
- Off-page. The Citations report shows which sources AI engines already trust.
- Reporting. Top cited URLs and the Brand Visibility Index paired with Search Console, shared weekly because citations move faster than rankings did.
“Keyword research tells you what people search. Prompt research tells you what conversation you need to participate in,” she says. “And there is always something about clients seeing a graph of how they are doing better than their competitors. That is what they all want to see.”
It also runs before the contract is signed. “We use OtterlyAI to show the client where their site is and where the gaps are,” Eri says. “It’s a tool that we use to help close clients.”
For Stanislava the value is credibility in a discipline clients cannot see. “People are often suspicious about everything SEO because they don’t see it or understand it, and AI SEO is not spared from this,” she says. “Our strategy and the data from Otterly combined with GA4 is great for that moment: this is working.”
Asked what she would tell a peer agency weighing the platform up, she keeps it to four words: “It is Otterly amazing.”
