Democratise first. Agentify second. The answer is the easy bit.
Jem Lloyd-Williams President of Mindshare & Chief Strategy Officer, WPP Media
Interviewed by Justin Cooke
Published
Jem Lloyd-Williams is President of Mindshare and Chief Strategy Officer of WPP Media. He has spent his career at the intersection of strategy, innovation, and media leadership through multiple waves of disruption, and now leads one of the UK's best-known media agencies through the AI moment. In this conversation he sets out the speed and complexity change in media that the industry has kept pace with; the just because you can doesn't mean you should strategic discipline; the operational complexity has thrown the emphasis back on strategic insight observation; the mindset over demographics targeting principle; Unilever as a culture-first, then broadcast case study; the structural advantage of working as one WPP; the democratise AI across the business to demystify it discipline he's been running for two and a half to three years; the Brand Asset Valuator example of agentification turning weeks of strategic discovery into hours; the measurement and analytics is the next frontier prediction; the AI hasn't produced a Cannes-winning idea yet line on creative; the empathy as a leadership superpower principle; the energy, inquisitiveness, comfort with data recipe for the next generation; and the if we get rid of the less-experienced generation, who learns to validate the machine? warning that closes the conversation.
Speed against complexity, and the strategic discipline that follows
On the change.
Speed of delivery for clients is the biggest change. Compound that with the number of channels, fragmentation of audiences, and complexity of the media ecosystem, and we've kept pace. We're now probably outrunning the complexity.
Operationally it's more difficult, which has slung the emphasis back to strategic insights and very specific recommendations for clients. Being across a thousand different channels is not a strategic response on its own.
The principle.
Just because you can doesn't mean you should. Really understand what the brand is trying to do with its communications. What equity are we trying to drive? How does it differentiate? Where are the best opportunities? Then double down.
Find the audience: easier in components, still complex in equation
On where the audience lives now.
If you're clear what you want to achieve with the audience, pinpointing where they show up and where they'll be most receptive has never been easier. A vast array of digital feedback signals, a clear understanding of data, more precise ability to analyse and identify where to show up. Easier means more components in what's still a very complicated equation. The skill: use AI to marshal, analyse, and bring together the multiple data sources quickly. The human element is still incredibly important; transferring information into knowledge is still a human pursuit.
On mindset over demographics.
Demographics is still an input. Mindset, or passion, or however you want to frame it is a more useful guide to where you need to show up. Marketing is the business of persuasion. Find people primed to receive the message because it resonates with their passions. Higher return for the brand-equity money.
Democratise AI first, then agentify the work for clients
On the discipline.
First, AI is available and democratised to everyone in the business. The negative press around AI talks about robots and job loss, so it's crucial everyone sees AI as an enabler rather than a replacement or a job killer. We made sure everyone can safely and securely access AI on multiple levels across every capability. They can hack their own workflows, improve processes, save time on internal admin. Democratising it takes away the anxiety because everyone is comfortable with the kit.
We've been doing that for two and a half to three years. Everyone in the business can also build their own agents (small or scalable). Second, agentification of key parts of our own process for client benefit.
On the Brand Asset Valuator worked example.
Brand Asset Valuator is WPP's oldest and most preeminent brand survey, around 33,000 brands. Before agentic technology it was a hand-crank thing. You went, found the data, pulled it out, hunted down the golden nuggets. A brilliant tool, but slow.
We put the data into a safe, secure language model and trained an AI to replicate the questions our strategists ask and find the answers manually. Now those questions can be derived in hours rather than days or weeks. The human is still driving the machine and validating the output. A different conversation.
Measurement and analytics is the next frontier
On where AI matters most.
The really interesting area is how quickly we can derive insights about how effective marketing is being, where it's being effective, and how to move budget from channel to channel strategically. Beyond ROAS or ROI: how to enable clients to make on-the-fly decisions with confidence about the outcome of moving budget.
On creative.
I've yet to see evidence that an AI can produce a Cannes-winning idea. The creatives who see media as a signals and data lab that enriches their thinking around what resonates with different audiences (and how to iterate an idea or evolve a campaign across touchpoints) are the ones who'll succeed.
Empathy as the leadership superpower, and the warning about junior talent
On the leadership principle.
Empathy is essential. Easy to see people's anxiety; harder to put yourself in their shoes and help them work through it. Sometimes change is seismic and they're right; sometimes it feels seismic but isn't. Empathy helps people understand what the change really means for them.
Leaders struggle to understand what transformation means at team level. The big-picture transformation can look small from the inside: can I still sit with my mates? Will I be able to use the kit? What does this mean for my career path? Empathy helps you work out what to say at macro level and at micro level.
On the warning.
AI capability requires experience and a really good grasp of heuristics: understanding what looks right before you put it together. If we get rid of the less-experienced generation because AI replaces them, who are going to be the next generation of experts? Who's going to understand whether the output from an AI or an agent is even vaguely valid? If we don't nurture that group, we're voting for our own downfall.
The question for the board
If the discipline now is understanding the problem better than anyone else, what share of our work goes into problem discovery versus executing fast answers?
