Insights from PGC London

By Eve, Artstash Creative · Published 23 January 2026 · Last updated August 2026

Quick answer: Two themes dominated PGC London. First, creative data labeling and taxonomy is becoming the real constraint on performance analytics — you cannot learn from creative testing at scale if your assets are not consistently labeled, so the taxonomy becomes the analysis. Second, playables continue to take share, with AppLovin still the default host, and AI playables moving from experiment toward production. Both point the same way: the bottleneck has shifted from making assets to organizing and understanding them.

Earlier this week the Artstash Creative team attended Pocket Gamer Connect in London. It was an opportunity to learn more about the current landscape of game publishing and marketing, and we’ve gathered some key insights to share from our time at the event.

Why Is Creative Data Labeling the Next Frontier?

When it comes to performance analytics and creative concept planning, Creative Data Labeling is the next frontier. Taking existing ads then (using AI or manually) labeling relevant aspects with tags – such as game mechanics, characters, motivations – allows you to see which tags perform best across your creatives. Then it’s a simple matter of scale, generating new creative with these tags in mind.

Why Are Playables Taking Over UA?

When teams think playables, Applovin is likely the first name that comes to mind. It’s been the platform of choice to host playables for years, but at PGC we’ve learned that playables are starting to become more important on non-Applovin platforms. Google is trying to step up in that space, and while the tech giant has offered playable hosting for some time, they’re now making a big push to capture more market.

Some quick numbers from the event:

• Top games are putting out over 60 playables per month
• Games produce an average of 2,743 creatives per month, with that number trending upwards
• Non gaming ads only produce 2,365

The conversation then quickly pivoted towards AI, as is often the case when scale is mentioned. AI Playables have been in a precarious position due to the difficulties around implementation and quality, but some major breakthroughs are beginning to appear – playablemaker.com was mentioned, a service that provides AI-generated playables. We will definitely be keeping our eyes on them in the coming weeks and follow their development.

Frequently asked questions

What is creative data labeling?

A consistent taxonomy applied to creative assets so their attributes — hook type, character, mechanic, format — can be analyzed against performance. Without it, creative testing produces results you cannot generalise from.

Why does creative taxonomy matter for performance analytics?

Because insight depends on comparability. If two assets are not labeled against the same attribute scheme, you can see which won but not why, which makes the next round of concepting guesswork.

Are AI playables production-ready?

They moved from pure experiment toward practical use during 2025 and were a major topic at PGC London. The realistic position is that they are viable for volume and iteration while still needing human oversight on quality.

Why are playables such a strong UA format?

They let a user experience the core loop before installing, which qualifies intent in a way video cannot. That tends to produce higher-quality installs, which is why platforms like AppLovin built their offering around hosting them.