AI Search Optimization Case Study: Ranked #1 Across ChatGPT, Gemini, Perplexity & Claude

The Challenge: Winning Trust in a Category Buyers Vet Hard

uproas.io operates in a category where the buyer does serious due diligence before spending a cent: premium agency ad accounts for media buyers running on Meta, Google, TikTok and Bing. When the project began, the domain was brand new. No authority, no brand searches, no organic footprint. In a trust-sensitive niche, that is a steep starting line, because the people searching are sophisticated advertisers comparing providers, reading reviews, and weighing reputation before they ever reach out.

Then the search landscape shifted under everyone's feet. By the time the project matured, a growing share of high-intent buyers had stopped opening Google at all. They were asking ChatGPT, Gemini, Perplexity and Claude a direct question: "best agency to buy an ad account." Whatever name those tools returned first won the consideration set before a single classic search result loaded.

So the brief had two layers stacked on top of each other. Rank in classic organic search for the research and commercial keywords. And become the entity the AI engines name when a buyer asks for the best provider. Most agencies were still optimizing only the first layer and leaving the second entirely to chance. uproas.io needed to own both.

The Strategy: Optimize for the Answer, Not Just the Ranking

Classic SEO earns a position on a results page. Answer Engine Optimization (AEO) earns a mention inside the answer itself. The mechanics are different. An LLM does not hand back ten blue links; it synthesizes one response and leans on a small set of sources to do it. To be one of those sources, a brand needs three things working at once: it has to be present in the data these models retrieve from, it has to be described consistently as an entity across the web, and it has to publish information useful and well-structured enough that a model reaching for an authoritative reference reaches for yours.

The plan ran on two tracks that fed each other. Track one was a classic organic foundation: rank uproas.io for the commercial keywords (the service pages) and the research keywords buyers hit earlier in the journey (the statistics and benchmark content). Track two was answer-engine visibility: turn that same content into the most quotable, most-cited reference in the niche, so that when an assistant assembles an answer about ad accounts, uproas.io is the source it pulls from.

The insight tying the two together is the part most teams miss: the assets that win classic organic (linkable, data-rich, frequently referenced pages) are the exact same assets LLMs prefer to cite. Build the right thing once and you win twice.

The Build: Two Years of Compounding Assets

We have worked on uproas.io's SEO since the project launched, roughly two years of compounding work rather than a 30-day sprint. AEO is not a switch you flip. It is the slow accumulation of signals that eventually tips a model toward naming you, and then keeps you there.

The content engine. The center of gravity was a set of data-rich reference hubs: Facebook ads statistics, Google ads statistics, Facebook ad benchmarks, and the core service pages for agency ad accounts across each platform. These were never thin keyword pages. Each one consolidates numbers a media buyer actually needs, formatted so that both a human skimming and a model parsing can lift a clean answer out of it. That structure is doing real work: assistants preferentially quote sources that state facts plainly and attribute them clearly.

Entity consistency. Across directories, profiles, mentions and third-party write-ups, uproas.io is described the same way every time: what it is, what it does, which platforms it covers. A model builds its internal picture of an entity from many scattered references. The more consistent those references, the more confidently it can name and recommend that entity.

Off-site citations and digital PR. The retrieval layer of modern AI search reads the live web, so the work extended well beyond the domain. We earned mentions on the pages and communities these systems crawl, so that when a model gathers context for "best agency to buy an ad account," uproas.io shows up in the supporting material, not only on its own site.

Indexing and structure. Clean markup, structured data and a crawl-friendly architecture made every asset easy to ingest. The running goal across all of it was simple: remove every reason a model or a search engine might have to skip the page.

AI Search Results: The #1 Recommendation on Every Major LLM

Ask the major assistants who the best provider is, and uproas.io comes back first. Not buried in a list of options: as the lead recommendation, by name.

ChatGPT returns uproas.io as the overall best option, citing strong reputation, multi-platform support and enterprise-grade accounts. Gemini describes it as the gold standard for established brands and heavy media buyers. Perplexity names it the strongest option in its sources, pointing to the balance of fast approvals, trust and scaling. And Claude shows up in the referral data as a citing source too.

That is the same result across four independent systems, each with its own training data and its own retrieval stack. When the recommendation converges like that, it reflects genuine entity authority rather than a one-off quirk of a single model.

The traffic followed the recommendations. AI assistants now send a measurable, growing stream of referral sessions: 1,639 in the period, up 111% over the prior six months, climbing from effectively zero in early 2025. ChatGPT leads with 936 sessions, followed by Gemini (342), Perplexity (189) and Claude (173). The pages those tools send people to tell the story on their own: the homepage, the statistics hubs (Facebook ads statistics up 675%, Google ads statistics up 3,850%), the benchmark content, and the agency ad account service page. The reference assets are doing precisely what they were built to do.

And this traffic converts. AI-sourced visits drove 91 contact conversions (Telegram and WhatsApp clicks), up 98% in the period. For a high-consideration purchase, having the AI layer pre-sell the recommendation before the buyer even lands on the site shortens the path to a real conversation dramatically. The visitor often arrives already convinced.

Organic Search Results: 3.3x Clicks, 11x Impressions

The AI win sits on top of a classic organic foundation that grew just as hard. The same reference assets that LLMs quote are ranking in Google for the queries buyers type before they buy.

Over the last six months versus the prior six, Search Console clicks rose from 7,920 to 26,200, roughly 3.3x. Impressions went from 376,000 to 4.15 million, more than 11x, as the site began surfacing for a far wider set of queries. Average position improved from 23.7 to 9.5, moving the site off page three and onto the first page for its core terms.

In Google Analytics, organic users grew 140% year on year (6,981 to 16,752). Organic search drove the overwhelming majority (15,523 users, up 141%), with organic social and organic video adding the rest. Classic organic sessions reached 38,749 for the period, up 139%.

Why It Worked

Three principles did the heavy lifting, and they transfer to any brand fighting for AI visibility.

The Honest Caveats

We report what the data actually shows, including the parts a polished case study usually rounds away:

Disclosure: uproas.io is a client SEOSkit has worked with since the project launched. Every analytics figure here is pulled directly from the client's Google Search Console and GA4.

Key Takeaways