In 2019, then-GroupM chief executive Mark Lollback told AdNews there were "only about five advertisers in Australia with the scale and deep pockets" who could truly in-house significant marketing, media and programmatic operations.

Five.

I was at GroupM at the time. I didn't disagree with him. He was 100% right at the time.

He isn't right anymore, and the reason has almost nothing to do with the size of the brand.

Who’s Actually Too Small To In-House?

Every article about in-housing opens with the same statistic. The ANA found 82% of its members now run an in-house agency, up from 78% in 2018 and 42% in 2008. If in-housing is so popular with large advertisers then surely it would be for all?

ANA members have the headcount, the procurement function and enough spend to warrant driving efficiencies and attracting the kind of talent required in-house. So the headlines get written up as a big-brand story but for a marketing manager at a $20m DTC brand who reads it, looks at their team of three and quietly thinks it’s not something they would take on. Or is it?

Nobody Asked A Single Question

I spent a chunk of my career on a team that helped brands in-house their technology stack. Not a side project. Actual, diligent, unglamorous work - meetings, migrations, contracts, seat structures, taxonomy, training...the work that doesn’t make the headlines. Most of the digital team in our own agency had no idea it was happening, let alone across the wider Holding Group. It was largely hidden, it just wasn’t something the agency wanted to highlight.

Within the clients own company: it was the single biggest project happening not only in marketing but across the wider business. The nature of the clients who worked on in-housing was completely different to marketing teams I had worked with before. They rolled their sleeves up, asked questions and treated us like members of their team. Mutual respect on all levels and you genuinely felt part of something more than just spots and dots. Fundamentally we all cared about what we were doing.

We went to a conference to present one of those transformations with the client. Told the story properly, the good, the bad and the ugly. When the session ended and the Q&A opened up, I remember the room was deadly silent. Not one question. You could feel the room deciding whether they’d be the tall poppy or not but Australia being Australia no one publicly dived in.

Then when it finished a handful of the room remained. Some of the biggest brand marketing teams in this country queued up at the side of the stage to ask the things they wouldn't in public. How long did it take? What did it actually cost? Who did you have to fire? What didn’t work?

That's the thing we keep coming back to at Cake.Shop . The demand was never absent, it was just silent. Asking the question in public may have meant admitting to your agency, your peers and possibly your company that you're considering it. So people don't ask, they wait and now the waiting may have real weight to actually doing it.

Buying Media Is Not In-Housing

Here's what I'd have said if anyone had put their hand up and what I said to those who asked since.

The industry uses one word for two completely different things. Bringing media buying in-house - someone on your team pushing budgets around a schedule - is a fundamentally different exercise to in-housing a technology stack. Different skillset, different timeline, different cost base. Confusing the two is how in-housing projects die.

The tech side is contracts, data architecture, identity, taxonomy, billing reconciliation and a hundred decisions nobody puts on a slide. It can be slow and unglamorous and it needs someone who has done it before to tell you which of those hundred decisions you'll regret. The buying side is comparatively simple, and it's the part everyone fixates on because it's the part they can picture.

The numbers back that up. The IAB and Accenture found 69% of brands had moved programmatic partially or completely in-house, and in the same study 16% had tried it and gone back to a partner. One in six. These stories don’t tend to get airtime but they exist.

If you want somebody else to be accountable when a campaign underperforms, don't in-house. That's a legitimate thing to want and it's worth paying for. Just be clear that's what you're buying.

The People Who Left Are Now Available

I wrote an article a few months ago about the talent crisis hollowing out Australian agencies. MFA's 2024 Census put regrettable loss at 26%. IAB's Talent Review confirmed the juniorisation, with more than three-quarters of open roles targeting one to five years' experience. The senior people went first, because the senior people cost the most.

I said then that they hadn't vanished. They'd gone in-house, into consulting, or out of the industry entirely.

If the experienced people left the agencies, they're now somewhere. A decent number of them are available in ways they simply weren't in 2019 - fractionally, on contract, two days a week or on a six-month build. The people who did those migrations with me are not inside a holding group anymore. They're in technology companies, independent agencies and in-house themselves.

The talent crisis and the in-housing opportunity are the same story. They are just being told in silo when really they should be told side by side.

The Decision Is The Only Part Worth Owning

When people say in-housing they picture what the big brands built. A trading desk. Six people. A workflow. Somebody who understands why campaigns didn’t pace on a Tuesday and can fix it before anyone notices. That's a labour-heavy operation and it genuinely does need scale. On that definition, Lollback's five is probably still right.

But media buying was not the part worth owning. The part worth owning is the decision layer. Who holds the seat. Who owns the first-party data and the audience segments built off the back of it. Who can see what the media actually cost versus what everything wrapped around it cost. Who decides what gets bought and why. Lastly, who can walk away in thirty days without losing four years of learning on the way out.

You can own every one of those without hiring anybody. We talk to independent agencies about this all the time - when we in-house technology from intermediaries, the light bulb moment for owners is the fact they see their agency move from transaction to critical thinking. When you move this in-house at brands you create marketers who not only know but fundamentally understand why agencies and staff are good at what they do, then inherently pay for things that they may have once not valued. Strategy, training, negotiating, planning all areas of the craft that are being devalued under the threat of AI.

The ANA has put $21.6 billion of unrealised value in programmatic. You don't need a trading desk to go and claim some of it. You need to know which questions to ask, and you need to be structurally allowed to ask them.

Principal Trading Is Making The Decision For You

This is the bit that has changed fastest, and it applies whether you're spending $2m or $200m.

The ANA's March 2026 study found 58% of advertisers said their agency used principal media in the past year, up from 47%. Ninety percent said their top concern was not knowing whether the principal media being recommended to them was in their own best interest, up from 79% two years earlier. Only 57% had any guidelines governing it. More than a third said their contracts either didn't address it or they weren't sure.

TrinityP3's Darren Woolley put the mechanism plainly in B&T: agencies can "sign up the client for a non-transparent principal media trading agreement, and then provide all the data and tech services that the client wants, but can't afford to make it appear (to their CFO) that they've paid for them."

The technology isn’t free. It's being paid for out of your media. The bill is going up, because everybody in this industry is now funding an AI and infrastructure build-out at the same time, and the money has to come from somewhere. It comes from the same place it always has.

I'm not saying principal trading is wrong and in most cases it’s fully complied with. It's a legitimate commercial model that some clients knowingly opt into for a lower headline cost. Whether they are seeing better quality ads, we highly question. What I'm saying is that it makes ownership a real consideration rather than a theoretical one. If more of your media is being converted into someone else's inventory to fund someone else's tech roadmap, then owning your own decision layer stops being a nice-to-have and turns into a straightforward commercial choice. It does that at $2m just as much as it does at $200m.

So Who's Actually Too Small?

On revenue alone, almost nobody. Which isn't the same thing as saying everybody should go and do it.

Take the brand this most often applies to. A DTC business turning over a few million, built off the back of capturing intent - you got very good at catching people who were already looking for you, and that was enough to build a company on. Then growth flattens, the cost of that same intent keeps climbing, and the honest read is that you've run out of people actively searching, and now you need channels that create demand rather than harvest it. Nobody on your team has ever bought one.

This is where I'll argue against my own side. Google and Meta got in-housed by brands like yours because they were built to be in-housed - self-serve, well documented and with a attribution loop short enough to learn from your own mistakes inside a fortnight. None of that knowledge really transfers to brand. Running a cross-functional brand campaign, where one strategy has to hold across TV, out-of-home, audio and digital and is built off actual strategy rather than a channel plan, is a completely different discipline. That is precisely where good agencies earn their money. Go and hire it and pay properly for it.

Which is the part that gets lost every time this argument comes up. Owning your stack was never an instruction to fire your agency. It's a question about what sits underneath whoever does the work, and who still holds onto it the day that relationship ends.

So the real test to in-housing isn't revenue. You're too small if nobody on your team actually wants to own it. Not has capacity for it but actually wants to own it. You can bring the expertise in from outside now, but you can't bring in somebody who cares, and if nobody internally does, all you've done is move the problem into your own building and put your name on it.

You're also too small if you can't name two or three decisions you'd make differently with full visibility. If the honest answer is "probably the same thing, just cheaper", that's a negotiation rather than an in-housing project, and you should go and have it with publishers.

Everyone else is big enough. A three-person team can hold the seat, the data and the decisions, and put the craft and the execution with people who are paid to do it rather than paid a percentage of it. That combination wasn't available in 2019, which is the only reason Lollback was right at the time.

The industry has spent a decade telling smaller brands that control is a privilege of scale, and it was never true - it was just expensive enough to disprove that nobody bothered arguing.

If you're reading this with questions you'd rather not ask out loud, you're not the first. But the time is now to start asking those questions, it may be the only path to take going forward.