Experian is a long-standing Jellyfish client, and this project builds on a partnership that has evolved over several years. Together, we explored how AI-generated creative could be embedded directly into performance media to unlock scale, speed, and sustained impact. Focusing on Performance Max, the work reimagined how creative is produced and refreshed in automated environments. The project was recognised with Gold at the BIMA Awards 2025 for innovative use of digital media.

Brief

With Performance Max becoming increasingly reliant on creative variety, Experian needed a way to scale high-quality assets without increasing production costs. As an established partner, Jellyfish was trusted to go beyond incremental optimisation and help rethink the creative model itself. The brief was to increase mid-funnel reach and engagement while building a more sustainable, future-proof approach to performance creative.

Challenge

As Performance Max matured, the scale and speed required from creative naturally increased. While the partnership was already strong, existing production rhythms weren’t designed for the pace of automated media, making it harder to refresh assets as frequently as the platform demanded. There was an opportunity to better connect creative development with live performance signals, so ideas could evolve more fluidly alongside the media rather than through slower, manual cycles.

Solution

Using Pencil, we integrated AI-generated creative directly into Performance Max and treated creative as a living system rather than a static output. Assets were continuously generated, tested, and refreshed based on real performance signals, allowing campaigns to adapt at scale. The success of this approach led Experian to become the first client to extend their Pencil contract for an additional year, reinforcing confidence in the model and its long-term value.

Result

My thinking

From a design perspective, the most rewarding part of this project was seeing a long-term client embrace a new creative mindset. Instead of designing individual assets, we designed a framework that could grow and learn over time. AI handled variation and speed, while design focused on structure, tone, and visual consistency. That balance made it possible to move faster without losing intent - and it reinforced the idea that good performance design isn’t static, it’s something you build to evolve.