Retail media grows up when media and merchant stop working in silos. The store is becoming media. And agents are about to do the buying.

Townsend knew from university he wanted to work in marketing and advertising, drawn by a field that felt both creative and analytical. He describes himself as business-led rather than a collector of ads: he saw a big opportunity for a job and had a crack at it.

His last business built scaled small and medium firms through technology, working closely with Facebook from its infancy to mass growth over a good 10 years. When he sold it, Meta began talking about a retail and brand product gaining traction, and he saw the same growth connection he had watched in social. He jumped in heavily about five years ago.

The space is still working through omni-channel and measurement. Networks a few years down the track are maturing: most start with top-line growth, new high-margin money, grow it as it grows. A few years in it becomes about margin pressure on top and bottom line. The merchant team and the media team, the two siloed parts of the business, have to connect into a single view for any network trying to grow for the long term.

Media brought in the publisher metrics, ROAS, conversion and click-through rate. But retail adds sell-through velocity, things that move product off the shelf, plus brand and incrementality. It's a transition where media metrics arrive and new metrics come in to make the space tangible, because most brands are spending here to sell product.

Brand opens up a different budget. This space is built to convert, sitting as close to the purchase decision as possible, so performance is critical. But no amount of media moves product people don't want in a given season. So a brand might run brand over winter so its incrementality in summer outpaces competitors. Performance drives incrementality in the right periods, and brand comes in once people understand when to use it inside the retail infrastructure.

Retailers overestimate the revenue they'll make and underestimate how hard it is to sell inventory. Moving from being great at retail into media means packaging and selling inventory so it adds incremental value, which is why programmatic is arriving to bring external demand. It only works when bought in by the chief executive down as a core, multi-year part of the business, not quick, cheap money, which goes stagnant fast. The first 12 to 24 months build teams, technology and networks.

You can't just bring in the media piece and think you're done; the core retail structure has to understand media. The critical hire is a leader who understands the space, then the right tech and data infrastructure as the spine, then sales and delivery teams. A few years in, there's a bigger pool of people who've done it and know the standards, where early on networks just threw agency and media people together.

Europe started in store then built to e-commerce and offsite. The US fast-followed Amazon, went hard on e-commerce, and is now coming into merchant-owned stores, so friction grows over who owns the relationship. APAC tried to connect the dots from the start. Externally it's fragmentation: 280 networks now, and the question of how an agency or brand buys across them. Retailers don't want to conform to be the same as everyone, because their DNA is being different.

A lot of the Cannes conversations are about connecting the media org into the retail org so the whole business grows together. Media can add value alone, 90 cents on the dollar on a sponsored product, 30 cents on a Google campaign. But if a product is on price promotion and you run media against it, you can lose money. So margin has to be read as media margin, product margin and overall bottom-line retail margin together.

The basic questions are how do I see, control and govern all my inventory, and how do I move from managed service into one place and then to self-service. Major publishers grow from the torso and tail, not just top spenders, so a five thousand dollar campaign should be self-service. Zitcha plumbs the retailer's system, serving ads across their websites, in-store screens and audio, and across Meta, Google, Snap and TikTok, then the retailer runs managed service on top or opens it to brands.

Some retailers are building their own ad serving and orchestration. Tier ones have the team, expertise and growth to do it. Tier two, three and four want speed to market without the resourcing to build, so technology moves at the speed of light while retail doesn't, and many are changing internal systems to catch up. AI now adds models that bring years of data to bear and automate decisions, with people in the loop beside a machine.

The breadth of delivery has grown: brands want to buy more and want it connected across store and e-commerce. The RFPs are much broader than four years ago. Beyond retail media sits a bigger question of how AI affects pricing, supply chain and whether shoppers or agents arrive. Five years ago nobody had all the answers in retail media, and we're at that point again with AI. The open question is timing: will retailers adopt it quickly, or the way they adopted digital over 5, 10 or 20 years.

The mainstream retailers don't show up in the answers because the niche ones generate authoritative content. A product feed is usually just product, price and description, but retailers must generate far more content around each SKU to appear in the LLM. Then comes the opportunity to monetise it: if an agent is told to buy a list, the retailer needs to intercept that journey and control it so the agent comes to their shop. Not everyone understands this problem yet, but it will move very fast.

New ways of creating content are coming. The content wrapping products gets fuller and connected back to the retailer, but the bolder ideas stand out: Albertsons running a sitcom in stores you can run a story against, Uber moving beyond its app into physical surfaces like airports and the journey from A to B. Expect deeper connection with the store, more theatre around it, and new monetised surfaces.

If everyone uses AI, sometimes it's the same slop that keeps coming out, and that's why creative will never die. AI will always create content that does the job that needs doing, but the human eye is what stops you mid-scroll. When you're trying to create concepts that capture the audience you want, you'll always need human-to-human idea creation.

For a business setting up a network, keep your target real, set up with the end in mind first, and make sure you have buy-in across the org. Those are the three biggest things.

For someone starting out, lean into the future, because innovation and future thinking have always been the biggest opportunity. Pick a path, but be relaxed if it moves, because the ground changes under your feet, so be agile. And have fun: you spend so much time in this space, full of creative people and great thinkers, and you're here for a short time, not a long time.