Okay, I need to talk about something TikTok is doing that has me genuinely unsettled. And not in the usual "oh no, another algorithm change" way. More in the "they're quietly trying to absorb your entire business operation and hoping you won't notice because the projected ROAS looks nice" way.

Here's the deal: TikTok is piloting a managed-services programme in the US where they don't just help you sell on TikTok Shop. They run your TikTok Shop. The ads, the creative, the influencer relationships, the content production — including hundreds of AI-generated videos — and your listing optimisation. All of it. For a flat fee of around $10,000 plus ten to twenty per cent commission on every sale.

What do you, the seller, get to keep doing? Listing your products and sending free samples to influencers. That's it. That's your job now. You've been promoted to intern in your own business. Congratulations.

(I wish I was exaggerating.)

To understand why this is different from, say, hiring an agency, you need to follow where the money and the decisions actually sit. Because once you trace the flow, the picture gets… uncomfortable.

TikTok already requires Shop brands to run their ads through GMV Max, its automated ads system. Manual campaign control? Gone. That was the prerequisite for selling on the platform. The managed service is simply the next logical step: TikTok now runs GMV Max on your behalf, on TikTok's own inventory, at prices TikTok sets, against a commission TikTok collects, with performance measured by — you guessed it — TikTok.

Every single toll booth on the road belongs to the same operator. They're selling you the car, charging you for the petrol, deciding the route, and then handing you a report card they wrote themselves about how great the journey was.

In a normal agency relationship, even a messy one, there's friction between parties that produces information. Your agency has an incentive to prove its media buying beat the platform's defaults. Your finance team scrutinises blended margin against reported returns. Those tensions are annoying, but they generate something valuable: a second opinion.

When the platform runs the ads, makes the creative, supplies the audience, and reports the outcome? That second opinion vanishes entirely. The entity telling you the campaign worked is the only entity in a position to know — and it's also the entity getting paid whether or not it actually did.

This is the advertising equivalent of asking the fox to audit the henhouse and then publish the results. Bold strategy.

It's worth separating this from something we've been tracking from the other direction, because the two look similar on the surface but mean almost opposite things.

Over the past year, a wave of independent players — performance agencies, marketplace operators managing brands across dozens of channels — have been building proprietary AI execution stacks. Systems trained on years of accumulated marketplace data that monitor pricing, advertising, and content continuously and act on them autonomously. The logic was defensive: faced with platforms accumulating gravity, these operators built data moats deep enough to justify their position between you and the marketplace. They were racing to remain necessary.

TikTok's managed-services pilot is the platform's answer to that race, and the answer is brutal in its simplicity. Rather than compete with the operator layer on execution quality, TikTok is offering brands a version of the same thing — autonomous ads, generated content, creator sourcing, listing optimisation — from inside the platform that owns the demand.

The independent operator's moat was proprietary data and execution capability. The platform's advantage is that it doesn't need a moat, because it owns the ground the moat was dug into.

It's like building the world's most sophisticated lock-picking set, only to discover the building's landlord just changed the front door to one that only opens with his face. Technically impressive work on the lock picks, though.

Here's where things get genuinely absurd, and I mean that in the most literal sense of the word.

At the exact moment TikTok is preparing to sell brands a managed service that includes hundreds of AI-generated promotional videos, AI-generated content is already flooding TikTok Shop from the bottom up. Affiliate creators are using avatars and digital twins of themselves to pump out shoppable clips at volume — some promoting products for brands that have no relationship with the creator and no idea they're being represented this way. One beauty brand reportedly found AI duplicates selling its products through the open affiliate programme, eligible for commission, with zero connection to the company.

Some brands are already cracking down. A well-known home-appliance maker recently told its affiliate community that AI-generated content is banned and any affiliate using it will have their commissions stripped. The reasoning was specifically about consumer trust — they didn't want a synthetic version of their product cleaning a synthetic floor in front of buyers who are increasingly learning to spot (and discount) machine-made content.

So. The same category of content is being fought from below and sold from above at the same time. Brands are stripping commissions to keep AI slop out of their affiliate feeds, while the platform prepares to offer AI slop as a premium managed feature. Hundreds of videos, generated at scale, on the brand's behalf.

This is like a hotel charging you extra for the noise-cancelling room while simultaneously hiring a brass band to play in the lobby. You're paying to solve a problem the building is actively creating.

The trust drawdown we've talked about before — the slow depletion of consumer confidence as a shared resource — acquires a sharper mechanism here. The depletion is no longer just an externality from individual sellers cutting corners. It's becoming a line item the marketplace upsells.

The flat fee and the commission are the visible cost. The dependency is the one that doesn't appear on the invoice.

A seller who hands the entire operation to TikTok is, within a quarter or two, a business that lists products and posts samples. That's the whole capability set. The advertising know-how sits with TikTok. The creative is generated by TikTok. The creator relationships belong to TikTok. The first-party data on what sold, to whom, and why — that accrues to TikTok too.

Should the arrangement sour — the commission rises, the category economics shift, or the platform decides to favour its own managed accounts over independent ones — you have no capability to fall back on and nothing you've built to leave with. The exit cost isn't a contractual penalty. It's the absence of anything to exit to.

Now, whether that trade is sensible depends on who's making it. For a small seller with no marketing function, no data infrastructure, and no realistic path to building either — outsourcing the whole operation isn't obviously irrational. It's access to competence that would otherwise be out of reach, and the dependency it creates is arguably a dependency they already had.

But for a brand with equity worth protecting — a distinctive voice, customers who buy by name, a reason to exist beyond appearing in the feed at the right price — the managed service purchases short-term efficiency by dismantling the very capabilities that make it more than a supplier of interchangeable stock.

The convenience is real. So is the erosion. The question is which one you're going to feel first. (Spoiler: the convenience comes immediately. The erosion comes when you try to leave.)

None of this is being improvised, by the way. The model has a working prototype, and it's not in the US.

TikTok's approach mirrors its sister platform Douyin in China, where the operator running content, commerce, and creator relationships directly from inside the platform is an established, successful model rather than an experiment. TikTok's commerce leadership has increasingly been shaped by executives who built the Chinese business, and the managed-services pilot reads less like a novel idea and more like an import — a model proven in one market being carefully installed in another.

If that's the trajectory — and every signal suggests it is — the pilot isn't an isolated programme. It's an early move toward a fully verticalised platform where the independent seller is progressively compressed into a supplier role and the platform occupies everything above it.

The specifics will differ by market and regulation. But the direction is legible enough to plan against now, while you still have capability worth protecting.

The managed service will almost certainly produce sales. Automated systems trained on a platform's own conversion data, pointed at that platform's own audience, with content generated to that platform's own specifications — these are well designed to move product in the short term. I don't doubt that for a second.

But "it worked" means something very different when the only party keeping score is the one being paid to say so.

There's a meaningful difference between a tool that extends what your business can do and a service that stands in for the business doing it. One leaves you stronger and more capable of independent action. The other leaves you fluent only in placing the order. TikTok's managed service is firmly the second kind — and it's being offered at the level of the entire commercial operation, not a single task within it.

You can rent the appearance of a marketing operation for a flat fee and a share of revenue. What you can't rent is the thing that made having one worthwhile: the capability, the data, and the control that let you decide, independently, what to do next.

And if that sounds like a small distinction now, wait until you try to leave.

🌐 Framer - Design and launch fast websites with AI, no code required.

🖥️ n8n - Open-source, self-hostable workflow automation with custom code nodes.

📹 Kling AI - Filmmaker-friendly AI video with strong character consistency and motion control.

🔌 Langfuse - Trace, evaluate and debug your AI agents and LLM apps in production.

🤖 Replit - Cloud IDE with an autonomous agent that builds, hosts and deploys apps.

😍It was great, all I needed

🙂Good, not great

🙄It sucked