5 takeaways from the ThinkingData Summit: AI, drowning in data, and Habby's secret sauce

Taking place during GDC week today was the ThinkingData Summit, bringing together companies like Blizzard, Mavis Games, Habby and others together to discuss how data is being utilised in mobile gaming.

Speakers shared insights on data and AI through to game design and industry trends. We also got the chance to interview Habby director of business development David Pann on stage during the show. 

Below we’ve put together a few key takeaways from the show.

1. How to survive an era of rapid change

The Thinking Data Summit opened with a talk from the firm’s co-founder Chris Han.

Setting the scene for the event, he posed the question he said every company is asking: how can we survive in this era where everything changes so fast? UA algorithms can change rapidly, while competitors can launch new features in live games that can leave your title behind. “The window to impact has never been shorter,” he said.

Han said when it comes to utilising data for your business, the real value is behind the data. “We have the rich data, but not the actionable steps, the actionable insights.”

He added teams like live ops and UA can typically be silo’d - which means data may also be silo’d, too.

Han also stated the opportunity for AI, which it’s now integrating into its products. It aims to launch its first ever ThinkingEngine AI agent next month, while it's plotting a multi-agent ecosystem by May to June. It even has plans to rebrand the company to reflect its AI ambitions.

2. “Every team is now drowning in data”

Blizzard senior director of product management Chase Shi discussed merging qualitative and quantitative data to shape the product roadmap.

He noted that every team is now drowning in data. It’s not about how much data a team has, but how teams can interpret it.

He used an example of where the product, community and data teams might interpret player data differently, asking how do you decide who to listen to? He showed the following chart to explain how overtime the data volume goes up, but the decision clarity goes down:

Shi said Blizzard puts players into different motivational segments - akin to the Bartle taxonomy of player types, but uniquely built around their own games. Developers can then map these players’ in-game behaviours.

“Don’t treat your players only as LTV tiers,” he said. “Treat them as motivational groups.”

He added: "You have to understand your players more before you read the data.”

3. Agentic live ops to challenge assumptions

Former AWS for Games head of live service games Tim Hong discussed agentic live ops and how developers could utilise AI tech. 

One insight Hong offered is how AI agents can be used to challenge assumptions made in the earlier lifespan of the game that continue to be considered as true. For example, an early insight could have been not to show more than three ads or over five at one time.

Hong said the problem with those rules is that they are probably directionally correct, but probably wrong in their implementation years later.

“The reality is that at most studios, nobody looks back, nobody audits that shit to say, oh, that’s the wrong approach. You can implement an agent whose job it is to basically check those things.”

4. Does Habby care about hybridcasual?

We interviewed Habby’s director of business development David Pann on stage at the summit and discussed how the company builds and runs successful hybridcasual games like Archero, Surivor.io, Capybara Go and others.

Asked how Habby defines hybridcasual - and if it even does, despite being a leader in the space - Pann said the team doesn’t see itself as a hybricasual developer or publisher, but instead focusing on midcore.

“We’re also branching out into more genres, as with a lot of companies. But the reason for that is, hybridcasual has a lot of implications on long-term retention. Meanwhile, for midcore we understand that always has longtail retention.

“Now, the type of midcore game that Habby does is not so much like 4X games, where you're talking about super insane mechanics that get people in the game, and those that churn, they're gone, and those that stay, they say forever. But the point is, at Habby, we're interested in building games that keep players well beyond day 30, well into day 100, and beyond.”

Pann also revealed that the original Archero had day one retention above 75%.

We’ll publish the full interview soon.

5. How to come up with new game ideas and verify them

Mavis Games CEO Saygin Topatan took to the stage to discuss the creative aspects of new game design, rather than just focusing on the data.

Opening on the topic of copycats, Topatan said from a business perspective, it can make sense. It’s easy for the team to know what to do, a developer can reverse engineer an existing game, and there are less decisions to make. From that, they could get good metrics.

“But we know most of these aren’t successful because they aren’t first on the market,” he warned.

However, Topatan claimed that it’s okay to copy some parts. For example, with a match-3 or a merge game, teams can adopt some UX patterns and meta habits. But there should be something to give the game a core identity, “one weird rule”, as he put it.

Giving just a test example to explain his theory, he described taking a match-3 game like Candy Crush Saga, and making matches in a T or L shape. While this may not actually work on the market, it’s an example of how one change could completely alter the experience for players.

Topatan also offered insights on how to find creativity without business-led decisions and KPIs getting in the way during the ideation phase. He said sometimes teams can look up the same five, 10 or 15 games that are successful in the market, which could lead to decisions like combining elements of successful games and trying to merge them, which might not work.

Topatan advised that the idea generation phase should just be for talking about ideas, without discussions around marketability and engagement. Then teams can move on to the editing phase and whether they need to add or remove something. After that, they can get to the building stage, where teams can talk about the pipeline of how to effectively build a prototype with the resources they have.

If you talk about the building stage in the first phase, it can obstruct all the creativity, he said.