What Is a Creative Machine in ecommerce?
A creative machine is a content production system built to generate, test and scale ad creative at volume, rather than a one-off campaign or a single hero asset. It treats creative as the primary growth lever in paid social, because platform automation has taken over most audience targeting. The goal is throughput with structure: enough new creative to keep feeding the algorithm, tested cleanly enough to know what actually works.usiness gets involved.
How Chief Nutrition Made Content the Engine
Chief Nutrition is an Australian health food brand built to replace the ultra-processed stuff on supermarket shelves. Justin Babet took it from five grand a month online to over a million, walking into food with no category experience and bringing the build, measure, learn habits from three previous tech businesses.
Here’s the part worth holding onto. Across four years of more than doubling revenue annually, the biggest investment Chief made wasn’t media spend, it was content production. At Chief, creative doesn’t support the growth, it drives it. Everywhere else Justin runs deliberately lean, outsourcing SEO, email and even a fractional CMO so the team can stay flexible while the business reshapes itself every few months. disciplined on the surface.
“We’ve more than doubled every year for the last four years and the business changes very regularly. The major investment has just been in content production.”
Creative Does the Targeting Now
The platforms have automated most of what used to make a good media buyer worth their fee. A few years ago, working out which segment should see your running shoe was a genuine skill. Meta does that part itself now, which leaves the creative as the signal that tells the machine who you’re chasing. Put an ad up and the creative effectively does the targeting: the right hook pulls in the right person, the right format lands with the right mindset. You’re still making a targeting decision, you’re just making it through the creative rather than the audience settings.
Simon Beard, who built Culture Kings to $600 million, showed the commercial side of this. He’d happily triple the spend on a lower-ROAS product carrying 80% margin and a high repeat rate over a high-ROAS product with thin margins. An agency chasing the platform’s ROAS number gets that call wrong every time, because they’re not close enough to the unit economics to see why it’s right. So it’s worth a hard look at where your attention and budget actually go right now. If it’s still mostly audience strategy, you’re running the old version of this.
A Content Machine Is a System. Find the Blocker.
Most teams already know they’re under-producing creative, and it’s almost never down to effort. The output stalls because something in the pipeline is jammed, and pushing harder just heaps more pressure on the part that’s already stuck.
Think of the pipeline as four stages: ideas, production, editing, distribution. When content isn’t getting made, one of those is broken, and the fix is usually a change to how you work rather than another hire or a bigger budget. Find where the output stops and solve for that.
The cleanest version of that is to remove the handoff altogether. At Ovira, Alice Williams traced her blocker to the gap between the person who understood what the algorithm was rewarding and the person making the content. By the time a brief crossed between them, the moment had passed. Her fix was to merge the two jobs into one person sitting in the ad account, running spend and making creative at the same time. Where that kind of restructure isn’t on the table, AI can pick up the slack, but only once you give it real context.
Mike Halligan at Scratch got nothing usable out of generic AI until his creative strategist wrote a context document spelling out what good Scratch creative looks like, the tone, the things to steer clear of. Usable output jumped from around 15% to close to 80%. AI didn’t need to invent the ideas, it just needed enough context to speed the engine up.
Volume Only Works If New Creative Gets a Genuine Test
Building a fast pipeline is one job. Making sure the new creative actually gets a fair test is another, and it’s the one most teams get wrong. You can pump out a mountain of content and learn nothing, because the algorithm keeps defaulting to what it already trusts. Drop new creative into the same campaign as your proven winners and Meta quietly funnels the spend toward the familiar, so the new work barely gets seen, underperforms, and gets binned before it ever had a chance.
The answer is a dedicated testing campaign. At Scratch, Mike keeps new creative quarantined from the proven winners, each piece on its own budget, away from the algorithm’s bias toward what’s already working. Anything that performs graduates into the main campaign, anything that doesn’t gets parked, often to be resurfaced later, since Meta tends to reward content that’s had a rest. Around 60 to 70% of Scratch’s sales now come from creative launched in just the last two to three months, which is what a system built to keep finding winners looks like in practice.
The same logic scales up. Paula Mitchell’s team at Freedom runs thirty to fifty variants per campaign using AI to generate the variations, same team, no extra production budget. Her one rule is equal starting conditions for every variant: equal budget, equal airplay. Skew the exposure and you get misleading signals, and it’s frighteningly easy to chase those signals down a rabbit hole. If you’ve gone to the trouble of building the machine, give the tests a genuine shot at telling you something true. The practical check is simple enough: is your new creative in its own campaign, or is it sitting next to the winners pretending to be tested?
“We simplified it a lot and put a lot more diversity into the creative. That created a big unlock. In the last three months we’ve doubled our new customer acquisition.”
The Takeaway
The brands pulling away on paid social aren’t spending more. They’re creating more and testing more, and they can do that because they’ve built the infrastructure for it. Whether we like it or not, retailers and brands are content machines now, and a content machine is a systems problem before it’s a creative one.
Build the pipeline. Clear the blocker wherever the output stops. Keep new creative separate so the algorithm gives it room to work. And keep feeding it, because the window when a piece of creative is fresh to the platform is the window that counts.
The content train isn’t slowing down. So when you go looking for your next batch of new customers, is your creative engine ready to find them?
Frequently Asked Questions
Why is creative more important than audience targeting on Meta now? Meta, Google and TikTok have automated most audience targeting, including lookalikes and bid optimisation, so the platform finds the right person for you. The creative you run is what now determines who sees the ad, which makes creative the primary lever you actually control. Audience strategy hasn’t disappeared, but it’s no longer where the edge is.
How do you build a content production system for ecommerce ads? Treat content as a pipeline with four stages: ideas, production, editing and distribution. When output stalls, find which stage is blocked and fix that, rather than adding more people or budget. Options include merging the media buying and creative roles to remove handoffs, or using AI with a detailed brand context document to lift usable output.
How should you test new ad creative without skewing the data? Run new creative in a dedicated testing campaign, separate from proven winners, so the algorithm doesn’t concentrate spend on familiar ads. Give every variant equal budget and equal exposure so the signal is reliable. Move winners into the main campaign and archive the rest.
Can AI-generated creative actually perform in paid social? Yes, but only when it’s given proper brand context. Generic AI output tends to be unusable, while AI guided by a document covering brand tone, what good creative looks like and what to avoid can lift usable output dramatically. Brands like Scratch and Freedom use AI to scale creative volume without expanding the team or production budget.
Based on Episode #629 of the Add To Cart Playbook with Justin Babet, Founder of Chief Nutrition.
In this Playbook:
- Creative does the targeting now
- A content machine is a system. Find the blocker.
- Volume only works if new creative gets a genuine test
Join Us!
Come behind the scenes with Add To Cart Community: Join our crew of e-commerce operators chasing growth, sharing wins, and staying sharp with deep dives, live events and no-BS inspiration.
