TikTok Shop in-house or agency? Most build-vs-buy math ignores half the P&L.

If TikTok Shop is on your plan for the next two quarters, someone has built the spreadsheet. It has two columns. On the left, headcount: a TikTok Shop manager, a creator recruiter, someone on paid, tools, maybe a coordinator. Call it $35,000 to $45,000 a month once it is fully staffed. On the right, an agency retainer, which lands in a similar range.

The usual conclusion is that the numbers are close, so the deciding factor is ramp time — an in-house team takes six to twelve months to become effective, an agency is live on day one. That is a fair argument and it is roughly true.

It is also incomplete in a way that matters more than the ramp, because both columns are costing out a TikTok Shop function, and TikTok Shop is not where most of the money lands.

The line that is missing

TikTok Shop creates demand. A meaningful share of that demand does not convert on TikTok. It converts wherever the shopper goes to check you out before spending money, and for CPG in the US that is overwhelmingly Amazon.

We have watched this on our own accounts: a brand's Amazon branded search sat flat near 5,600 a week through ten weeks of TikTok posting, then tripled to 17,134 once affiliate volume hit critical mass. Same brand, same 90 days, Amazon sales up 39.7% and profit up 53%. That lag is its own planning problem, and I wrote it up separately in does TikTok Shop actually lift Amazon sales.

Now put that back into the spreadsheet. The demand your new TikTok team generates is going to arrive on your Amazon detail page. So the question is not "what does a TikTok Shop team cost." It is:

When the demand arrives, who is accountable for whether it converts?

If the answer is "nobody," or "our Amazon agency, separately," you have a structural problem that neither column priced.

The three ways this actually breaks

One: you build TikTok in-house and the demand leaks. Your new team does good work. Videos go out, branded search climbs. It arrives at a detail page with a main image built for search intent, thin reviews on the SKU the creators are pushing, and a competitor bidding on your brand name. Most brands we audit are losing around half their Amazon PPC budget to branded search they already earned. A TikTok program pointed at that page makes the leak bigger, and does it with your money. This is what listing optimization is actually for, and why we do it before scaling demand.

Two: you hire an agency for TikTok and keep your Amazon agency. Now you have two vendors, one channel of demand, and no agreement about attribution. When branded search triples, the TikTok agency reports a halo win and the Amazon agency reports an organic rank win. They are describing the same shoppers. You are paying two performance fees on one result, arbitrating a fight you cannot resolve because neither party can see the other's data.

Three: you build TikTok in-house and it is genuinely slower than the platform. This is the ramp argument, and it is real, but the mechanism is worth naming. TikTok Shop rewards speed more than any channel before it — a creator goes quiet, a format stops working, a video breaks out and needs paid behind it within days. In-house those decisions cross three people and an approval. That is not a criticism of the hires. It is what happens when a fast channel is run through a normal org.

When in-house genuinely wins

I run an agency, so discount this accordingly. But there are cases where building is clearly right and it is worth saying so:

• You are above roughly $50M and TikTok Shop is strategic, not experimental. At that size the function should be yours, and you can absorb a two-quarter ramp without the channel needing to pay for itself in month three.
• You already have a strong internal Amazon team. The coordination problem disappears when both channels report to the same person. That is most of the advantage an agency has.
• Your category needs deep product knowledge to brief creators. Regulated categories, technical products, anything where a wrong claim is a real liability. Outsourced briefing is genuinely worse here.
• You have someone in-house who has already done it. The playbook is the expensive part, not the headcount. If you have already bought it, build.

If two or more of those are true, build. The honest version of the agency pitch is that you are renting a playbook and a creator network, and if you already own both, you are renting nothing.

The comparison that is actually useful

Not "team versus retainer." Try this instead:

That last row is the real trade and it goes against agencies. If you build, you own the capability permanently. If you buy, you are renting it for as long as you pay. Anyone who tells you otherwise is selling.

The row above it is the one people skip, and it is where the money is.

How to run the math properly

If you do nothing else from this post, add two lines to the spreadsheet:

1. What is your current branded-search capture rate on Amazon? Pull Search Query Performance, look at your branded terms, and check your share of clicks and purchases. If you are leaking, quantify the leak at your current demand level, then multiply it by whatever lift you are forecasting. That is the cost of running TikTok without fixing Amazon.
2. Who signs off on the Amazon detail page? Write the name down. If it is a different company from the one generating the demand, add a line for the coordination cost — in meetings, in delay, and in the attribution argument you will have in month four.

Then compare. The columns usually stop looking close.

The short version

TikTok Shop is not a channel you can evaluate on its own, because a large share of its return arrives somewhere else. Build it, buy it, or split it — but price the Amazon side either way, and make one party accountable for the whole path from a video to a shipped unit at margin.

The spreadsheet with two columns is answering a smaller question than the one you are asking.