Search is fracturing right before our eyes. When someone types a complex prompt into an AI today, the engine breaks down the intent and hunts across multiple pockets of the web to piece together a synthesized answer. Honestly? That’s exactly how we as humans handle things. We don't look for everything in one place. We instinctively bounce between platforms depending on what we need, heading to Reddit for unfiltered opinions, TikTok for trends, or official sites for hard metrics. People are still just hunting for authentic information; the only difference now is that AI has joined that multi-platform journey to take the heavy lifting off our hands. This means Generative Engine Optimization (GEO) isn't an isolated trick or a marketing silo; it is just one piece of a much larger puzzle. To build a successful campaign today, you have to look at the big picture. You can't just optimize a single website; you have to ensure your brand's narrative is alive across the entire ecosystem that your audience, and the AI engines scraping them, actually trust.
.jpg)
The message is loud and clear: if you aren't putting your facts out there in a clean and discoverable way across this broad landscape, you aren't just losing rankings, to a huge chunk of your audience, you won't even exist.
How It All Started
In October 2025, we decided to step back and audit exactly how visible and accurate Outfit7 was across the major LLMs. We weren't just checking if the basic facts on our Wikipedia page were correct; we wanted to know if these engines would actually recommend us when users asked the kind of everyday, open-ended questions we assumed we’d dominate.
But right out of the gate, we hit a massive paradox: how do you audit your visibility against user prompts when AI companies keep their prompt data locked away in a black box? For SEO, most search engines provide Webmaster tools that help you understand your search performance and spot opportunities for further optimizations to expand your presence. There is no such thing for GEO. It is impossible to just download a spreadsheet of what people are asking LLMs... To break through this blind spot, we had to become detectives of consumer intent first. Before we could even test the AI engines, we had to map out the actual questions our audience was asking.
Once we built that list of real-world consumer prompts, we used them as our baseline to stress-test the LLMs. The results were a massive reality check. When we asked an AI point-blank, “Who is Outfit7?”, it easily spit out accurate data pulled directly from our Wikipedia page. It knew our founders, our scale, and our corporate moves.
But when we tested indirect, discovery-driven category prompts - the kinds of queries people use to find something new to play - we were nowhere to be found. Even though Talking Tom & Friends has racked up billions of downloads and essentially defined the modern mobile virtual pet genre, the AI engines completely bypassed us, recommending smaller, niche competitors instead. Having a massive global brand didn't mean a thing if AI crawlers couldn't connect the dots in a conversational prompt. That blind spot is exactly where this playbook began.
1. SEO Is Not Dead: It Is Evolving and Becoming More Critical Than Ever
There is a loud narrative in marketing claiming that "SEO is dead." We completely disagree. SEO is undergoing its most profound evolution yet. It is becoming even more critical to a brand’s survival, but the scorecard is changing. Traditional SEO was about winning the click to a blue link. Modern SEO, augmented by Generative Engine Optimization (GEO), is about winning the citation, the summary mention, and the absolute recommendation within an AI’s final answer.
They are not competing strategies; they are two sides of the same coin - authority. If traditional SEO is about link authority, GEO is about topical authority. It is no longer enough that relevant websites and pages link to your content. Content around the topics important to you has to mention you as well and support your narrative.
Generative engines read, extract, and trust the very infrastructure that robust SEO builds. If your technical SEO is broken, AI engines cannot find your data. But if you don't layer conversational optimization on top, they cannot synthesize or recommend you. GEO represents the next maturity stage. Here is how the two work together as a single search ecosystem:
In this new hybrid ecosystem, a weak technical SEO setup is fatal. AI search engines retrieve information from the highest-authority, most technically sound web structures. This explains a well-known paradox in modern search: industry research on GEO has shown that nearly 90% of pages cited by AI actually rank at position 21 or lower on Google organic search.
Traditional SEO ensures the AI bot can technically crawl and index your site, but GEO ensures the AI chooses to recommend your content because it directly answers the user's prompt.
By aligning our technical hygiene with natural conversational formats, we are future-proofing our footprint for the models that now dominate consumer decision-making.
2. Prompts: Figuring Out the Questions People Actually Ask
Once we realized how AI answers were put together, we had to stop tracking traditional "keywords" and start thinking about "prompts." People don’t type “best virtual pet mobile game download” into LLMs. They treat the AI like a friend, typing long, conversational questions like, “I am bored, what virtual pet mobile games would you recommend?” or “What is the best mobile game where you take care of a pet?”
Since we couldn't just look at a dashboard of LLM queries, we had to piece together our audience's natural language from the outside in:
Digging into Real Conversations: We went to the places where our audience talks naturally. We combed through customer support emails, app store reviews, Reddit forums, and Google Trends. These spaces showed us the exact language, frustrations, and casual phrases players use when they aren't trying to format a traditional search.
Looking for the "Why" and the Specifics: We didn't just look for game genres; we looked at how people framed their specific needs.
Making Smart Guesses: Using these real-world constraints, we connected the dots to build a master list of predicted prompts. We basically simulated what a player, a casual gamer, or a tech fan would ask an LLM if they were looking for something new.
3. The Dream Team: How We Assembled and Divided the Project
Realizing we were invisible to conversational prompts, we knew we couldn't treat this like a standard technical patch. We needed a true cross-functional team to rethink how our brand lived across the internet. We sat down and intentionally built a core team out of three distinct pillars at Outfit7:
Content Creators & Copywriters: The people who actually write our app store descriptions, web articles, and video descriptions.
Marketing, Legal & PR: The owners of our brand narrative, responsible for external communications and storytelling.
Technical SEO: The architects behind our web best practices, technical optimizations, metadata, structured rich data, and web vitals.
We divided our entire roadmap into two classic strategic lanes: Owned Media (the channels we completely control) and Earned Media (the exposure we get through third-party platforms and word-of-mouth).
An AI doesn't just read your website, believe you, and move on. Consistency across completely different domains is what turns a corporate claim into an AI-verified truth.
4. Owned Media: Rewriting Content for a Hybrid Human-Machine World
Our first major execution block was taking a hard look at our owned properties, our corporate site, game landing pages, and App Store Optimization (ASO) data. We discovered that writing polished corporate marketing content was actively causing LLM crawlers to ignore us. We fundamentally overhauled our owned content architecture through four strategic steps:
Speaking Simply to the Machine: We stripped out the corporate fluff and started writing text that was direct, educational, and matched natural conversation. Instead of using general pronouns like "our studio's premium products," we used explicit proper names like "Outfit7" and "Talking Tom & Friends" so models could clearly index our entity.
Deploying the Q&A Framework: Industry data shows that structuring text with direct questions followed by immediate, explicit answers significantly increases the likelihood of being extracted and used as a source by conversational engines. We refactored our landing pages to include question-based subheadings (H2 and H3) that perfectly mirrored real conversational prompts, following each with a tight, standalone answer.
Structural Chunking & Technical SEO Cleanup: LLMs prefer clear, digestible data chunks. We organized our content using clear bullet points, short paragraphs, and side-by-side comparison tables. On the technical backend, we enrolled our sites into Bing Webmaster Tools and added structured JSON-LD schemas to give AI crawlers clean organizational facts.
Content Expansion & Regular Updates: We prepared dedicated content for each of our games, diving deeper into gameplay, important milestones, and awards we won. But we haven’t stopped there; instead, we doubled down and continued with regular content updates, releasing various guides and feature deep dives, expanding on the now growing topical coverage of our products.
5. Earned Media: Building Trust Through High-Authority Sources
Overhauling our website wasn't going to fix everything. As our study proved, AI engines routinely prioritize neutral collective wisdom over a brand's own marketing copy. We targeted the external spaces where AIs look for proof, what we internally call the Super Feeders.
Press Releases: We didn't stop doing PR; we structured releases so AI systems could easily extract the key facts, stripping the fluff and adding short, simple summaries right at the top.
PR Relationships: This is where all your past media networking pays off. We found the high-authority sites AI treats as trusted "super feeders," and worked with those outlets through standard editorial pitching to secure earned coverage in places we knew LLMs frequently reference.
Listicles: "Top 5" and "Top 10" lists for recommendations work. We pitched media to get into their roundups, giving AI systems credible third-party validation to draw on when answering those searches.
Results: What Changed and What We’re Watching Next
By changing how we handle both our owned text structures and our earned community footprints, we began to actively shift how AI models index Outfit7. Because LLM search logs are a black box, we measured our success using concrete proxy metrics, specifically running internal tracking audits across major engines using our master list of 50 simulated customer prompts, alongside tracking AI referral traffic.
The data over six months showed a massive shift:
+357% LLM Traffic: Since we started tracking and optimizing in October, the number of users finding our website through AI chatbots/LLMs has more than quadrupled.
90% Prompt Presence: We are now showing up in AI prompts where we used to be completely invisible.
We Have A Better Understanding On How It Works Now: the biggest win isn't even the traffic, it’s the knowledge we’ve gained.
We're Already Using It Everywhere: This isn't just theory. We are already using this know-how to tweak our PR articles, daily communications, and website content so AI models actually pick them up.
The Path Forward
The AI landscape is shifting so rapidly that there is no "set it and forget it" strategy. Generative Engine Optimization (GEO) is not a one-time technical patch; it is a moving target that requires us to remain fast, agile, and ready to adapt.
AI search is not replacing SEO; it is changing what authority looks like. The brands that win this new search equation won't be the ones producing the most content or shouting the loudest. Instead, they will be the ones building the strongest, most consistent digital evidence across the web.
For Outfit7, the project became about building what brands have always needed: authority and consistent presence across the web, communicated clearly enough for both people and AI to trust. The goal was to make accurate, useful information about our games easier to discover, verify, and understand across the platforms people already use-by keeping our facts simple, our data structured, and our brand deeply embedded across the digital ecosystem. Ultimately, AI does not choose the loudest voice; it chooses the brand the internet consistently trusts and recommends.
