If you sell a TikTok Shop US seller whose performance score has slipped below the threshold, with creator content driving the returns and negative reviews that pushed it down, creator content is probably your largest source of orders and, quite possibly, your largest source of refunds at the same time. The US performance score is one of the few platform metrics that changes your costs directly rather than your reach. Above the threshold, return shipping is largely not yours and settlement clears in a matter of days. Below it, the return shipping share moves onto you in full and the money sits out longer. Both of those hit the cash line in the same month, which is why a drop in the score feels like a margin problem rather than a traffic problem. And for most sellers running creators, the inputs that moved it are content inputs: negative reviews and returns, both of which are overwhelmingly produced by what the video promised relative to what arrived.
Put the fall on a monthly number before deciding how much work it deserves. Take a shop doing 8,000 orders a month at an 8% return rate: that is 640 returns, and at roughly $7 of return shipping each that is about $4,480 a month. Above the threshold you carry about a fifth of it, call it $900. Below it you carry all of it, so the drop costs roughly $3,500 a month in return shipping alone. Then the cash side: on $200,000 of monthly volume, moving settlement from a handful of days to around a month means close to $170,000 of your own money is sitting outside the business at any point in time, which you either finance or you do not buy inventory with. Those two lines are the real cost of the score, and neither of them appears anywhere in the content plan that caused them.
What falling below the threshold actually costs per month
The score is not one number, and treating it as one is why so much recovery work fails. It is built from several inputs, and the ones a creator programme can actually move are the review and return signals, not the fulfilment ones. A seller whose score fell because of late dispatch will spend a month rewriting briefs and see nothing change, because the input that moved it was never in a brief. Before you touch a single script, look at which input moved and by how much. If reviews and returns account for most of the fall, content is the lever and this article applies. If they do not, the fix is in the warehouse and no amount of brief rewriting will move the number.
There is a timing asymmetry worth planning around. The inputs that push the score down register quickly, but recovery does not: you have to produce a stretch of clean orders and clean reviews before the number moves back, which means the fix has to start before the cost becomes unbearable and continue after it stops hurting. Sellers who treat it as a one-week push usually relapse, because the briefs revert as soon as the score recovers and the same creators put the same claims back in.
Step 1 — Find out which input moved the score before fixing anything
Pull the last sixty days and split three things by week: return rate, negative review rate, and late dispatch rate. The one that moved first, and the one that moved most, tells you which lever you have. Then narrow it further — split return and review rate by traffic source, and specifically by creator. What you are looking for is concentration. If one or two creators account for a large share of the returns, this is a creator-specific problem and the fix is a conversation and a rewritten brief. If returns are spread evenly across creators, the problem is in the product or the listing, and rewriting briefs will spread the same result across more videos.
Watch the review text, not just the rating. A one-star review that says 'not as described' and one that says 'arrived late' are the same number and completely different problems, and only one of them is a content fix. Read thirty reviews and sort them into buckets by hand. It takes an hour and it is the fastest way to know whether this article is the right one to be reading.
Splitting returns and reviews by creator is the step that decides everything downstream, and it is also the step that needs the most joining. If you want the structure ready, register with DAMI and ask the team for the free SPS content recovery checklist: return rate and review rate by creator in one view, with the threshold columns already built.
Step 2 — Replace 'looks like' with 'is' in every creator asset
The gap between what a video shows and what arrives is the single largest producer of both returns and 'not as described' reviews, and it is almost always a depiction problem rather than a lying problem. Three things close it. Show the product next to a common object so scale is unmistakable. Show the material in a close-up under normal light, not under a ring light that flattens colour and hides texture. And stop smoothing the footage — a filter that makes fabric look heavier or skin look better is a filter that produces a return. The test is simple: if a viewer paused on a frame and measured it, would they get the same product that ships.
This is the one requirement that creators resist most, because it makes the content less pretty. The way to get it is to stop framing it as a restriction and start framing it as the reason the video converts: buyers who know exactly what they are getting do not return it, and creators whose audience does not return things keep their own standing with the platform intact. Say that out loud in the brief, with the number attached.
Step 3 — Put size on the screen, not in the description
US sizing is the second largest producer, and it is a disclosure problem rather than a sizing problem. The buyer is not wrong about the size, they were never told the measurement. Put three things on screen as text, not in the description: the actual measurement in inches at the points that matter, the model's height and weight, and the size the model is wearing. A creator saying 'true to size' is telling the buyer nothing, because true to whose size. On-screen text is what works, because a large share of viewers watch without sound and will never hear the sentence, and because a claim on screen is a claim you can check before the video posts.
Add the one line that filters hardest: who this size does not fit. A creator who says 'if you are between sizes, take the larger one, and if you are above 5'9" the length will be short on you' will lose some orders in the video and gain them back in the review section. That trade is almost always worth making while your score is below the threshold, because one avoided return is worth more than one marginal order.
Step 4 — Set the delivery expectation inside the video
The third cause is expectation, and it sits entirely inside the video. Buyers who were told the parcel arrives in three days and get it in nine leave a review about the product. Buyers who were told what is in the box and find fewer items leave a review about the product. Neither complaint is about the product, and both land on the score. Put two lines in the brief as required on-screen text: the delivery window you can actually hold, stated as a range rather than a promise, and the exact contents of the parcel. Underselling both is the only version of this that protects the score, and it costs nothing — the buyer who would have waited nine days is not deterred by being told nine.
Once the score is back above the threshold, keep exactly two of the four requirements permanently: the on-screen measurement and the contents line. Those two cost a creator nothing after the first video and they are the ones that stop the relapse. The delivery window line can go back to whatever you were doing. Keeping all four forever is how briefs become long enough to be skipped.
How DAMI reads returns and reviews back to the creator who produced them
DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, sample management, campaign management, store analytics, and multi-store team dashboards into one place. All four fixes depend on one thing: knowing which creator produced which return and which review. Three steps make that a read instead of an investigation.
Step 1 — Split returns and reviews by creator
Read return rate and negative review rate per creator as their own columns rather than as shop totals. Output: concentration becomes visible in one sort — you can see whether two creators are driving the fall or whether it is spread across the roster.
Step 2 — Put the four requirements on the creator's record
Attach the depiction, measurement, contents and delivery expectations to the creator's brief with a version number. Output: every creator is working from the same four requirements, and you can see which brief version was live when the score moved.
Step 3 — Watch the two inputs weekly while recovering
Track return rate and negative review rate weekly until the score recovers, then monthly. Output: the fix is measured against the inputs that caused the fall, not against the score, which moves slower than either of them.
Frequently asked questions
How long does it take for the score to recover once the content is fixed?
Longer than it took to fall. The inputs that pushed it down register immediately, but recovery needs a run of clean orders and clean reviews before the number moves back. Plan for weeks rather than days, and keep the requirements in place after the score recovers, because that is where most sellers relapse.
Should I stop working with the creators who produced the returns?
Not immediately. Concentration tells you where to look, not who to fire. Give the creator the rewritten brief and one cycle to see whether the number moves; a creator whose returns fall after a proper brief is more valuable than a new one you have not measured. Remove them only if the second cycle looks like the first.
Is the return shipping share the only cost of a low score?
No, and it may not be the largest. The settlement delay is often bigger: on meaningful volume, moving from a few days to around a month keeps a large share of your own cash outside the business, which you either finance or do not spend on inventory.
Can I track this in a spreadsheet?
For one cycle, yes. It breaks when the question becomes per creator and weekly, because returns, reviews and brief version have to be refreshed on the same rows every week. That is the join nobody redoes, and it is exactly the join the fix depends on.
Conclusion
Every cost line in this article moves without anyone deciding to move it, and every one of them is cheaper to fix before the next campaign than after it. DAMI brings an 8-million-plus creator database, competitor and competitor-store creator search, automated outreach tasks, AI-assisted outreach, sample management, campaign management, store analytics, and multi-store team dashboards into one place.
Fifteen minutes today: pull the last sixty days of negative reviews and sort them into two buckets — 'not what I expected' and 'arrived late'. If the first bucket holds most of them, the fix is in your briefs and you can start with the on-screen measurement line on your three highest-volume SKUs. If the second bucket holds most of them, stop reading this and go to the warehouse.
Start with DAMI and run the next cycle from numbers you can defend.
