Talking Tom Had Billions of Downloads but Was Missing from AI Recommendations. Here's How Outfit7 Changed That

Talking Tom & Friends has attracted billions of downloads. Yet when its developer, Outfit7, asked AI assistants to recommend virtual-pet games, its own globally recognised games were missing from the answers. The assistants knew Outfit7 when asked about the company directly, but recommended smaller alternatives when asked what to play.

Instead of guessing why, Outfit7 decided to investigate. Its team developed 50 realistic player questions, improved information about its games across websites and independent sources, and measured what changed. Six months later, Outfit7 reported appearing in approximately 90% of its tracked AI prompts and a 357% increase in website traffic from AI assistants.

Who should read this article? Owners, marketers and decision-makers at established B2B and professional-services companies that already have real expertise and a strong reputation, but want to know whether AI assistants recommend them when potential clients ask whom to hire.

Key takeaways

1. A famous brand can still be missing from category recommendations. Outfit7 found that AI could identify the company but failed to recommend its games for relevant player questions.
2. Outfit7 investigated the problem itself. Its team built 50 realistic prompts using customer-support messages, app-store reviews, Reddit discussions and Google Trends.
3. The changes went beyond the company's website. The team improved its own content and technical structure while strengthening accurate independent coverage.
4. Outfit7 reported measurable progress. After six months, its games appeared in approximately 90% of tracked prompts, while website traffic from AI assistants rose 357% from its starting level.
5. GO!GEO applies this research logic to B2B supplier selection. We study the questions buyers ask, record which suppliers AI recommends without prompting a brand name, connect real services to credible evidence and repeat the tests.

Outfit7's results are company-reported, not outcomes independently measured by GO!GEO.

Why would an AI assistant know Talking Tom but recommend other games?

Outfit7 described the problem in its case study published on 1 September 2026.

When it asked AI assistants about Outfit7 by name, they could explain who the company was and which games it made. But when the question changed to which virtual-pet games someone should try, Talking Tom was absent from the recommendations.

For a person who already knows the brand, that may sound strange. For somebody discovering a product for the first time, the distinction is decisive: they cannot choose an option they never see.

The same challenge can arise when someone asks for an EU funding consultant, an engineering specialist or a corporate tax adviser. A business may have decades of experience, but does the AI assistant connect that experience with the particular service the buyer needs?

The purchasing process for a game and for professional services is not the same. What they share is the difference between an assistant knowing a company's name and offering it as a relevant choice.

Outfit7 didn't know what people asked AI, so it built 50 realistic questions to test.

Companies can examine many of the search queries that bring people to their websites. They generally cannot see every private question somebody asks an AI assistant about their category.

Outfit7 therefore looked at the language players used in places it could study: customer-support emails, app-store reviews, Reddit discussions and Google Trends. From these sources it developed 50 simulated player prompts.

They were not 50 recorded conversations with AI users. They were realistic questions, designed to reflect how people might search for games. The team then tested those questions and recorded whether its products appeared in the responses.

This gave the company a baseline it could repeat after making improvements.

At GO!GEO, we use the same underlying logic for B2B companies, but the questions are different. We first establish which services, markets and types of clients matter commercially. We then ask AI assistants realistic buyer questions without naming the company. Only after recording those recommendations do we ask about the company directly.

A positive answer to “Is this a good consultancy?” is useful, but it does not mean the assistant would have suggested the consultancy to somebody who had never heard of it.

Outfit7 worked on its own content and on information published elsewhere.

Outfit7 brought together colleagues responsible for content, marketing, PR, legal matters and technical SEO. According to its account, the work included improving game pages, headings, answers to common questions, structured information and clear links between the company and its products.

The team also studied external articles, press coverage and game recommendations. It worked to make relevant, accurate information more available through sources beyond its own website.

That is an important distinction. A company's website is the place where it can directly explain its work. External coverage and independent records can help people check whether the claims are credible. Neither is a substitute for genuine product or professional quality.

For a B2B company, I find it useful to consider three parts of the information buyers may encounter.

Where should a B2B company make its expertise visible and verifiable?

These sources should tell a consistent, checkable story about the same real expertise.

Imagine an EU funding consultancy. Its website describes proposal-development services and the programmes it knows. Named specialists explain their roles in successful applications. A case study describes a real project and outcome. Where appropriate, an independent programme or consortium record confirms the relevant facts.

That is more useful than simply publishing more articles that repeat general advice about EU grants. The goal is to make real experience easy for a buyer to follow and assess.

What changed after six months of work at Outfit7?

Outfit7 reported two results from its monitoring after six months:

Approximately 90% presence in the set of 50 AI prompts it tracked. That means presence in its measured question set, not 90% of every possible recommendation a person might receive.

A 357% increase in website traffic referred by AI assistants. That is a traffic measure, not a reported number of paying users, qualified B2B leads or signed contracts.

The underlying records have not been published in sufficient detail for GO!GEO to reproduce the complete measurement independently. The case also does not prove which individual website or PR change caused the improvement.

What it demonstrates is a practical sequence: investigate the actual recommendation problem, agree what can be improved, implement changes and measure again.

At GO!GEO, we apply this approach to established B2B service providers.

Outfit7 had the internal specialists to investigate its own AI Search Visibility. Many established professional-services companies have strong expertise but no dedicated team to research how AI assistants recommend suppliers.

That is where GO!GEO works. We start by identifying which services and markets offer a credible business opportunity. We research the questions those buyers ask and record recommendations from ChatGPT, Claude, Gemini, Perplexity and Google AI Mode. We then examine whether the company's own website, professional profiles and independent evidence make its qualifications clear.

If a service is described vaguely, a relevant expert is disconnected from the service page, or a successful project cannot be verified easily, we turn those findings into a Generative Engine Optimization (GEO) strategy. We implement or supervise the agreed work and check the same buyer questions again.

For example, our research for Tiko Pro, an established EU-funding consultancy, identified a marked difference between recommendations in its home market of Slovenia - where Outfit7 was also founded - and recommendations in other European buyer situations. This pointed to concrete opportunities to connect existing services, specialist expertise, client projects and independent evidence more clearly.

The detailed findings belong in their own B2B research article. Our 16-page original research report and recorded AI answers are available as supporting material.

Our goal is not simply more mentions of a company name. We want credible AI recommendations for the services a business actually provides - and a path from that visibility into a new source of qualified leads. Recommendation inclusion, website visits, enquiries and signed clients remain different outcomes, and we don't treat them as interchangeable.

You can begin by asking AI two questions about your own company.

Choose one important service and one market where you want to attract clients.

First, start a fresh conversation and ask a realistic question without naming your company. For example: “We're a manufacturing company looking for an experienced consultancy to help prepare a European funding proposal. Which providers would you recommend?”

Record which providers appear in the initial answer and what sources, if any, are linked.

Then ask: “What can you tell me about [company name], and would you recommend it for this work?”

If the company is described positively in the second answer but absent from the first, you've found a useful gap to investigate. That alone doesn't prove what caused it. Repeat the test with several relevant questions, record the systems and dates and compare the same questions after verified changes.

The result may be different for each service, market and buyer situation. That's exactly why real testing is more valuable than making assumptions about what AI knows.

What should established B2B companies and professional service providers learn from Talking Tom?

Outfit7 had a globally successful game franchise, yet that success did not automatically make its games appear in relevant AI recommendations. The company investigated the problem instead of guessing, improved the information around its products and reported progress after six months.

The lesson for a professional-services business isn't to copy the same game-marketing tactics. It's to use a disciplined method: ask the questions buyers really ask, see which providers AI recommends, make actual expertise easier to understand and verify, then test again.

If your company already has a hard-earned reputation, capable people and successful projects, you have something real to work with. The challenge is to connect those strengths with the specific service decisions potential clients are making.

Want to know whether AI recommends your company when someone asks who to hire?

Start with a Free AI Visibility Audit from GO!GEO.

---

Sources and further reading

• Outfit7 - If an AI Can't Find You, Do You Exist? Outfit7's Deep Dive into the New Search Equation. Published 1 September 2026; source for Outfit7's approach and company-reported results.
• GO!GEO - Full original Tiko Pro EU Expansion Research. Research dated 22 October 2026.
• GO!GEO - Original AI Visibility Baseline. Recorded answers from 17 September and 21 October 2026.
• Google Search Central - Creating Helpful, Reliable, People-First Content.