AI is changing discoverability: How Talking Tom boosted AI referral traffic by 357%

• One year ago, the Talking Tom IP was barely being suggested to users of AI.
• Outfit7 set out to increase discoverability for its major IP.

As more of the population turns toward AI to answer questions and find recommendations, there’s inevitably a cohort of mobile gamers asking LLMs too. Instead of browsing an app store or going to Google to research their next potential game, some may go to ChatGPT, Claude or one of the many other tools now available.

While this may be a convenience for users, this isn’t necessarily an advantage for developers and publishers. Discoverability is splintering. Search engine optimisation (SEO), generative engine optimisation (GEO), recommendations from AI and rocketing tech advancements are all factors to keep track of. But, if properly optimised, there may be ways to take advantage of the change and boost appearances in AI results, at least for now.

Outfit7 began auditing itself across the major LLMs in October 2025 and found that Talking Tom & Friends had "near-zero visibility" through indirect prompts at the time - such as asking for a virtual pet mobile game to play.

Setting out to change this, the company formed a cross-functional core team with staff across marketing, legal, PR, copywriting, content creation and technical SEO. Now, Outfit7 claims to have seen a 357% increase in AI referral traffic over six months.

"For a game to be discovered, it needs to be present where the users are, whether it’s app stores, social media, traditional search or AI," Outfit7 senior director of engineering Primož Bevk tells PocketGamer.biz.

"So, I think we need to follow the shift if we want to be considered a choice for gamers. In the end, AI is just one of the surfaces through which users interact with the digital world and we should leverage the benefits it brings.

"For us, the goal was to be present where we could reach new potential fans and players, while also building awareness of both the Outfit7 corporate brand and Talking Tom & Friends. It’s usually easy to pass ‘who are you’ but much harder to pass open discovery questions."

Optimising at Outfit7

The company looked at dual strategies: owned media and earned media. While optimising a website may help when the AI comes to pull results, the tech typically draws from multiple sources. As a result, Outfit7 set out to identify sites the AI would deem high authority, then worked with those outlets to secure coverage, increasing its potential for discovery.

Despite amassing billions of downloads, the Talking Tom IP wasn't showing up in indirect AI searches a year ago. Now, Outfit7 claims to have increased Talking Tom’s presence across 50 real-world player discovery prompts to 90%.

To monitor traffic increase, the company used its website as a proxy since traffic couldn’t be directly measured from LLMs to the app stores. For prompt presence, the team decided on 50 prompts where the brand wasn’t originally showing, but where they thought it should.

Establishing what those prompts should be meant studying organic search behaviour, such as defining "high-intent keywords" and determining what users were searching for. Data was gathered from customer support emails and app store reviews to give a picture of how Talking Tom players speak and Outfit7 began to formulate prompts its target audience might ask AI - and how to word them.

Bevk notes that studying the language of customer reviews has also been useful in understanding the existing playerbase. While these users already know the Talking Tom brand, they might only play a handful of the 20-plus Talking Tom & Friends games. This means they might discover a new title through AI, too.

"When we first started, we hit this wall right away. How can we be sure that the strategy we execute even works if we cannot easily measure KPIs that we take for granted with SEO?" Bevk recalls.

"So, we had to get creative and set some proxy indicators, like measuring AI traffic on our brand website and employing a similar GEO strategy for both app store storefronts and our website."

With the lack of SEO-equivalent support with LLMs and no way to simply download a spreadsheet of what people are asking AI, it’s becoming more challenging to understand search performance. Webmaster tools have been a go-to for SEO, but GEO can be more of a blind spot.

Bevk reveals that organic web search has actually never been a major user acquisition channel for Talking Tom, so the company hadn’t put much emphasis on SEO optimisation anyway. That’s meant the team had "a lot to catch up" on with GEO.

"There is a lot of misinformation and misconception about GEO and there is no quick way to make it work - it requires persistence," he adds.

With this in mind, we ask how Outfit7 can be certain its recent efforts are what led to improved results for the IP, when evolutions in tech and improvements to the LLMs themselves are also occurring concurrently.

"The short answer is we can’t. LLMs are advancing at a staggering rate and there is no Google Search Console for AI which could reassure us on the effectiveness of our changes," admits Bevk.

"That being said, looking at our data and comparing it to the trend from before we started implementing our GEO strategy, we can say with confidence that if the trend was on a slow increase before, now it’s decidedly more positive. And we believe that cannot be just from improvements in the LLMs themselves."

Top tips beyond Talking Tom

We ask Bevk what advice he would give to other studios looking to maximise their discoverability via AI. He says a company should focus on what its goals are, on blind spots and on understanding what a specific audience is looking for within that category of games.

"Put yourself in the shoes of your audience. Understand what they are looking for. Why is your game the one they would want to play? You need to find out what questions they are asking," he expands.

He also recommends looking to Reddit and Google Trends, then writing helpful content that answers relevant questions being asked online. Especially for smaller studios, making the most of AI results may prove to be a key avenue going forward.

"I think for indies it’s even more important to have good SEO fundamentals and consider AI discoverability as an important avenue, especially if they are in less saturated genres and categories, where it’s easier to stand out."

Outfit7’s own research concluded that there is no "set it and forget it" model because of how quickly the AI landscape changes. GEO is like a moving target, so studios must be ready to adapt quickly even if they believe they’ve mastered optimisation today.

Bevk concludes that, going forward, Outfit7 will continuously expand the number of prompts it tracks. At the same time, the team must focus on what is useful and relevant for players across the Talking Tom catalogue.