If you sell a US TikTok Shop seller sending hundreds of samples a month and judging the programme by how many parcels went out rather than by how many creators came back selling, the constraint that decided this year was not demand and not competition. It was how long your money stays gone. Five hundred samples out and ten videos back is not bad luck. It is a screening formula that was never written down, and the marketplace is large enough that this is easy to miss for a whole quarter. In the US, affiliate eligibility starts at 5,000 followers, the cost of a sample lands between 20 and 33 dollars once 15 to 25 dollars of product cost is added to 5 to 8 dollars of shipping, and a new programme is generally advised to budget at least 3,000 dollars a month, which is 90 to 150 parcels. The benchmarks that parcel count is supposed to produce are not mysterious: 40 to 60% of sampled creators post within fourteen days, and of those who post, roughly 20 to 30% generate measurable sales. So a hundred samples should return forty to sixty videos and eight to eighteen creators who actually sell. Five hundred returning ten is not the market being difficult. It is two of those three funnels having been left wide open.
Convert the story into a unit cost before deciding what to fix, because the two diagnoses lead to opposite actions. Five hundred samples at 28 dollars each is 14,000 dollars. At the benchmark funnels — 55% posting, 25% of posters selling — those five hundred parcels should have produced about 275 videos and roughly 69 selling creators, putting your cost per selling creator near 203 dollars. What actually happened: ten videos is a 2% posting rate, and if two or three of them sold, each selling creator cost between 4,600 and 7,000 dollars. Now put that next to what a creator is worth. A micro creator in the ten-thousand to hundred-thousand band, which is repeatedly the best ROI tier in 2026, might move forty orders a month at a 32-dollar average order value: about 1,280 dollars of GMV. At 15% commission and a 40% gross margin, that is roughly 320 dollars of monthly contribution after the creator is paid. A 203-dollar acquisition cost pays back inside the first month. A 5,000-dollar one does not pay back at all, and the reason is entirely upstream of the creators — it is in who got a parcel.
A sample is customer acquisition cost, not marketing spend
The accounting error is subtle and it drives everything else. Sellers book samples as marketing spend, which has no natural denominator, so it gets judged by volume: how many parcels went out this month. But a sample is an acquisition cost, and acquisition costs have a denominator — the thing you acquired. Price it that way and the arithmetic becomes uncomfortable in a useful way. Twenty samples at 12 dollars of product plus 5.50 of shipping is 350 dollars a month. Spread over 300 units sold that is 1.17 dollars per unit; spread over 50 units it is 7.00. Same programme, six times the overhead, and the only thing that changed is whether the creators sold anything. This is why sending more samples is not automatically scaling: if the denominator does not grow with the numerator, you have made the unit economics worse while the activity report looks better.
One structural fact about the US market is worth keeping in view while you do this, because it changes where you should concentrate. Affiliate-driven sales account for roughly 42% of US platform GMV, and the revenue is heavily concentrated at the top: in a well-run programme the top 5% of creators typically drive 60 to 70% of affiliate revenue. That cuts both ways. It means the long tail is not where the money is, so hiring broadly is expensive. It also means the screening step, not the recruiting step, is where a programme is won — and screening is nearly free while recruiting costs a parcel each time.
Step 1 — Price one selling creator before you send the next batch
Run the calculation with your own numbers before the next batch leaves. Take the cost of one sample including shipping, divide by the share of sampled creators who post, and divide again by the share of posters who generate sales. At 28 dollars, 55% and 25%, a selling creator costs about 203 dollars. Then work out what that creator is worth to you: expected orders a month times average order value, times your gross margin, minus the commission you pay them. If the payback is inside one month, the programme is healthy and worth scaling. If it is inside three, it is worth continuing with tighter screening. If you cannot make the payback land at all, no amount of extra parcels fixes it, because you are buying an asset that does not return its cost.
Set the commission band before you run this, because it sits inside the payback calculation. The 2026 tiers have settled into a recognisable shape: nano creators at one to ten thousand followers work at 10 to 15%, micro at ten to a hundred thousand at 12 to 18% and widely considered the best ROI band, mid-tier at a hundred to five hundred thousand at 18 to 25%, and macro above that at 25% or more, often with a flat fee on top. The US open-plan average sits around 13%. Pick the band first, then price the creator against it, not the other way round.
The cost-per-selling-creator calculation is one row of arithmetic, and the reason it does not get done is that nobody owns it. register with DAMI and ask the team for the US sampling payback worksheet so the thresholds and the payback test sit in one sheet before the next batch is approved.
Step 2 — Diagnose which of the three funnels actually broke
Three funnels sit between a parcel and a profitable creator, and only one of them is usually the problem. The first is posting: 40 to 60% is the benchmark within fourteen days, and industry reads put the share who post nothing at 30 to 50%. If yours is 2%, the leak is screening, not the creators. The second is selling: of those who post, 20 to 30% should generate measurable sales. If posting is healthy and selling is not, the leak is fit — you are reaching people who make videos but do not move product in your category. The third is amortisation: monthly sample cost divided by units sold. If both rates are fine and unit economics still hurt, your volume is too low to carry the programme, and the answer is fewer creators at higher quality rather than more at lower.
Diagnose in that order and resist the urge to fix all three at once. Sellers who change the commission, the brief and the creator tier in the same month learn nothing, because the next batch cannot tell them which change worked. Fix the funnel that is furthest from benchmark, hold the other two constant, and read the next batch against the same numbers.
Step 3 — Write the screening thresholds down and stop waiving them
Write the thresholds down and stop waiving them per creator. Four checks cover most of the damage: at least 5,000 followers, because that is the platform's affiliate eligibility floor and a parcel below it cannot produce an attributable sale; average views above 5,000 on the last ten posts, because follower counts are inflated far more easily than view counts; at least three videos in the last thirty days, because an inactive creator will not produce timely content no matter how good the sample is; and genuine category alignment rather than occasional adjacency. The first three are mechanical and take under a minute each. Waiving them is where 14,000 dollars goes.
Two notes on reading those numbers honestly. Average views should be taken over the last ten posts rather than over the whole profile, because one viral video from two years ago will carry a lifetime average and hide six months of decline. And category alignment should be checked on the last twenty videos, not on the bio: a lot of profiles describe a niche they stopped posting in some time ago.
Step 4 — Batch, follow up at fourteen days, cut at twenty-one
Batch the sends and build the follow-up into the batch rather than into someone's inbox. One frozen list per week, dispatched on the same morning, so every parcel shares a dispatch date — that date is what makes the batch readable three weeks later. Then run two rules without exception: if nothing is posted fourteen days after delivery, send one follow-up; if there is still nothing at twenty-one days, flag the creator and stop sending. The platform has its own sample integrity policy that restricts creators who repeatedly receive samples without posting, so your cut-off supports a standard rather than just protecting your budget.
Read the batch back as three numbers, not as a feeling: parcels sent, videos posted, and creators who generated measurable sales. That last one is the only one that belongs in a budget conversation. A batch that sent forty and produced eleven videos sounds acceptable until you notice that two of them sold, which puts the batch at roughly 560 dollars per selling creator against the 203 you modelled — and that is the moment the screening thresholds get taken seriously.
How DAMI turns sample spend into a per-creator return rate
DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, AI video creation, sample management, campaign management, social media management, store analytics, and multi-store team dashboards into one place. Three actions turn sample spend into a per-creator return rate, and each one produces something you can look at rather than something you have to take on trust:
Source from creators already selling in your category instead of browsing a database
Open competitor and competitor-store creator search in DAMI and pull every creator who posted about three comparable US stores in the last thirty days. The output is a list with demonstrated category fit, which removes the funnel leak that costs the most money.
Screen mechanically before a parcel is approved
Filter that list inside DAMI on the eligibility floor of 5,000 followers, average views on recent posts, and posting frequency in the last thirty days. The output is a shortlist with an expected posting rate, which is the input your cost-per-selling-creator calculation needs.
Record every parcel against a batch and read it back at three weeks
Manage samples in DAMI against a batch date, then let the fourteen and twenty-one day rules run. The output is three numbers per batch — sent, posted, sold — which is the only reporting that survives a budget review.
Frequently asked questions
Is 40 to 60% posting really the benchmark, or is my category different?
The 40 to 60% figure within fourteen days is the commonly cited range, with another industry read putting the share who post nothing at 30 to 50% — the same thing stated from the other side. Categories differ somewhat, but not by the factor you would need to explain a 2% rate. If your number is dramatically below the band, assume screening before you assume category.
How much should a new programme budget for samples?
A commonly advised starting point is at least 3,000 dollars a month, which at 20 to 33 dollars per sample is 90 to 150 parcels. Treat that as a ceiling you earn the right to raise rather than a floor: scale it only after a batch has returned a cost per selling creator you are happy with, not before.
Should I pay the bigger creators to guarantee output?
It buys a different thing than you might expect. Macro creators above 500,000 followers typically command 25% or more and often a flat fee, and their conversion does not reliably justify it against a micro creator with forty thousand followers in one category — a band that repeatedly shows the best ROI. Guaranteed fees guarantee a post, not a sale, and the funnel you are fixing is the one that produces sales.
Does a returned order claw back the commission I already paid?
It depends on when the return happens relative to the roughly 15 to 31 day settlement window. A return inside the window cancels the pending commission; a return after payout is deducted from a future balance rather than refunded to you. Either way the commission follows the sale, so a creator whose audience returns at a high rate costs you more than their commission line suggests.
Can I run this in a spreadsheet?
For the first two batches, yes, and you should — it forces you to look at the numbers. It stops working when you are screening hundreds of names against four thresholds, running follow-up rules keyed to dispatch dates, and reading multiple batches at once. The sheet does not break; it simply stops being opened by the person who needs it on the day the batch is approved.
Conclusion
Cash is the one input that decides how large you are allowed to become, and everything above is about making the same money cover more of the business than it did last quarter. DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, AI video creation, sample management, campaign management, social media management, store analytics, and multi-store team dashboards into one place.
Take your last sample batch, write down three numbers — parcels sent, videos posted, and creators who generated sales — then divide the batch cost by that last number. Fifteen minutes, and you have the cost per selling creator that everything above is built on.
Start with DAMI and find out how much larger your next cycle can be without a larger balance behind it.
