$2.5 million for a moat

In March, Howatson and Company announced they’d built their own AI platform. It’s called Unicorn. Their digital division, Plus Also Studio, spent 18 months and $2.5 million on it. It takes a master creative asset and adapts it into hundreds of formats for digital and out of home. They now sell access to it at $5,000 a campaign.

Mumbrella’s headline quoted Chris Howatson describing it as “a moat around our core work.”

I read that and felt two things at once, which is usually a sign there’s something worth writing about.

The first was respect. That is a real business decision made with real money by someone who runs a real agency. Eighteen months of engineering is not a press release. They saw the adaptation work getting commoditised, they got in front of it, and they built a revenue line out of the thing that was about to stop being billable. Most agency owners talked about doing that in 2025. Howatson actually did it.

The second thing I felt was less comfortable. Because I don’t think that’s a moat. I think it’s a very expensive bridge.

A moat has to be hard for someone else to build. That’s the whole idea. Not valuable, not clever. Hard.

Four months after the Unicorn announcement, Vinne Schifferstein Vidal wrote a piece for Mumbrella arguing that agencies selling their AI platform as the differentiator are selling the wrong thing. Her line was this:

Read that again, because it’s doing something careful. She’s not saying the platforms don’t work. She’s saying they work for everyone, at roughly the same time, at roughly the same level. WPP has Open. Publicis has CoreAI. Omnicom has Omni. Monks has Flow. They are all built on the same frontier models that you and I can log into this afternoon.

So if the technology is evenly distributed and the work is still wildly uneven, the unevenness is coming from somewhere else. Which means whatever is actually creating the difference is not the platform.

That’s not a criticism of Unicorn. It’s a description of what a moat is, and a platform built on rented intelligence isn’t one. It’s a head start. Head starts are worth having. They’re just worth having for a shorter time than $2.5 million wants them to be worth.

Here’s what got under my skin about this, and it took me a few days to work out.

I wrote a while back about clients getting rational. A client sent me a Canva deck she’d made herself, and three years earlier she’d have paid us four grand for it. The mistake agencies make with those clients is thinking the deliverable was the product. It never was. The deliverable was the receipt.

Building your own AI platform and calling it a moat is the same mistake wearing a better suit.

Clients think the deliverable is the product. Agencies now think the tool is the product. Both are looking at the artefact instead of the judgement that produced it, and both are going to get the same answer from the market, which is: fine, but cheaper next year.

In June, Rocket Agency announced a partnership with Canva. James Lawrence, one of the founders, said Canva “has built an incredible platform that has fundamentally changed how teams create, collaborate and bring ideas to life.” The point of the partnership is to help in-house marketing teams make their own content.

Sit with that for a second. An Australian agency, partnering with an Australian tech company, to get better at helping clients not need an agency for that part.

I don’t think that’s surrender. I think it’s the most honest read of the situation anyone has published this year. Lawrence has looked at where the production work is going, agreed with it, and moved to the part of the business that survives it. That takes more nerve than building a platform does.

We’re seven people. $2.5 million isn’t a number I get to have an opinion about. So take this as arithmetic rather than principle.

But if I had it, I wouldn’t spend it on adaptation software, because adaptation is exactly the kind of work that gets cheaper every six months whether I invest in it or not. I’d spend it on the thing that got more expensive.

Judgement. Not the ability to make a hundred versions of an asset. The ability to know which one to run, and to be right often enough that a client stops asking for the other ninety-nine.

Being in the room. Every time we’ve won something we shouldn’t have, it’s because somebody on our team noticed a thing that wasn’t in the brief. A tone in a meeting. A hesitation in the way a founder describes their own company. You cannot buy a platform that attends the meeting.

The relationship that survives a bad quarter. The client who came back to us six months after leaving didn’t come back because our output improved. They came back because the work they got elsewhere was competent and did nothing.

None of those three things scale. That’s the part everyone hates. You can’t licence them at $5,000 a campaign, you can’t put them in a deck, and you can’t build them in 18 months with a development budget.

You build them by being the same people, doing careful work, for long enough that it compounds. Which is a terrible business model right up until the moment everyone else’s moat fills in.

I might be wrong about this in a specific way, and it’s worth naming.

If Unicorn earns its $2.5 million back at $5,000 a campaign, then Howatson didn’t build a moat. He built a product. And a product is a genuinely good answer to commoditisation, maybe a better one than mine. He’ll have turned the part of the business that was dying into the part that pays for the part that isn’t.

That’s not the same as defending the core work. But it might be smarter than defending it.

What I’m certain of is narrower. The tool is not the difference. It can’t be, because your competitor is buying the same one this quarter, and their client is buying it next quarter, and the price only goes one way.

The tool is the floor. The floor is now free. Everybody is standing on it.

What you build on top of it is the only thing anyone is actually paying for.

Have you built something like this, or watched someone spend real money on one? I want to know whether it earned it. Hit reply, I read every one.

Sammy