A new shapewear SKU, from angle to review
Updated 2026-08-15 · Risk-term list version 2026-08-15c
The creative process for a new SKU has five steps and the order is not optional: pick the angle → write both scripts → scan the claims → write the brief → review after the flight. Reversing any of it produces the most expensive kind of rework — finding out after the shoot that the line cannot be said.
Three of the five have explicit decision rules: the angle from the category prior plus the product’s own fields, the claims from level plus context, the review from "did the angle win or the hook".
Each step below: what to do, what the rule is, and what going wrong looks like.
Step 1: pick the angle (30 minutes)
The three angles that pay in shapewear are Pain / Solution (prior 86), Myth-bust (80) and Social proof (76), for the same reason: buying here is about solving one specific wearing frustration, not about liking the design.
The prior is a starting point, not the answer. Your product’s own fields move it: review volume (social proof runs on it), a genuine strike-through price (price anchor runs on it), how many spec fields exist (review-style angles run on those). Running social proof with 400 reviews is bringing the thinnest ammunition to the angle that needs it most.
| Angle | Shapewear prior | When not to use it |
|---|---|---|
| Pain / Solution | 86 | Almost always usable; skip only when there is no clear wearing frustration |
| Myth-bust | 80 | When you have no defensible counter-evidence — inventing one to manufacture conflict backfires |
| Social proof | 76 | Under 500 reviews; not enough ammunition |
| Unboxing / Review | 74 | Thin spec data makes it hollow |
| Day-in-the-life | 70 | Cold start — the path to conversion is long; use it for retargeting |
| Price anchor | 66 | On its own it drags you into a price war; better as a closing beat |
Step 2: write both scripts (1 hour)
Not a translation — two originals. The differences go past vocabulary: Chinese voiceover runs fast and dense with no throat-clearing in the first three seconds; English is more conversational, allows pauses and self-deprecation, and a read-aloud script is a conversion killer. CTA conventions differ entirely.
One concrete gap: US commercial content needs the commercial relationship disclosed. Chinese-language copy has no such habit, and a straight translation drops it.
Three hooks per angle, one per video — those three become three cells of the A/B. Lock the hooks verbatim; let the creator put the middle in their own words.
Step 3: scan the claims (10 minutes, and it saves the most money)
This step scans your listing, not your script. That is where the risk originates: sellers copy the listing’s strongest line straight into the voiceover and the creator brief.
Order: listing → script → captions → creator brief → comment replies. Then act by level: high risk gets rewritten; needs review depends on context — "a size smaller" in sizing advice is fine, "look a size smaller" is a quantified body-change claim.
Currently 23 Chinese and 22 English phrasings, 19 mapped across. Full list: flagged terms. Paste and scan: the checker.
Step 4: write the brief (30 minutes)
One page, carried by two sections: the one required shot and the phrases that must not be said. The second is pasted from step 3.
The most persuasive shapewear shot is "same outer garment, only the layer underneath changes" — one variable. More variables and nobody believes it, and this is exactly where a creator improvises (new pose, new light, different dress). So the brief has to nail it down: same camera, same outfit, no cut.
Full structure and template: how to brief a creator.
Step 5: review after the flight (30 minutes)
Default matrix: 3 angles × 2 hooks = 6 cells, $30 minimum per cell, 5-day window. Set the thresholds before the flight — deciding them after seeing the numbers is self-deception.
The review answers one question: did the angle win or the hook. The test is mechanical: both hooks on an angle up → the angle won; only one → the hook won. Which is why no two cells may test the same sentence — a collision destroys the test.
| What the numbers look like | What it means | What to do |
|---|---|---|
| 3s view-through < 25% | Hook problem | Swap the hook, keep the angle |
| 3s fine, CTR < 0.8% | The middle does not convince | Fix the proof beat with concrete evidence |
| CTR fine, CVR < 1% | Not a creative problem | Landing page, listing or price — stop making creative |
| $30 spent, no orders | Dead cell | Stop; move the budget to the other hook on that angle |
| ROAS > 1.2× target at minimum spend | It works | +50% budget and produce two new hooks on that angle immediately |
The three sub-ranges differ
The beachhead splits into three sub-ranges with different centres of gravity. One playbook does not cover all three:
- Bodysuits and tummy-control shorts: lead on "you cannot tell it is there" and "it did not roll after a full day". The required shot is one outer garment, one changed layer.
- Shaping leggings: the only hard metric is the squat test. This audience cares about one thing; everything else is a bonus.
- Sport compression and back support: moving the setting to training or desk work pulls the viewer out of the shapewear frame and lowers the odds of tripping health-category review. Support pieces may describe the sensation ("my shoulders had something to lean on") and never the outcome ("corrects posture").
Related
FAQ
How many assets does a new SKU need?
Six on the default matrix (3 angles × 2 hooks). Fewer than three angles tells you nothing about angle; more than three thins each cell below significance.
How long do the five steps take?
Half a day to a day by hand. The time sink is not the steps — it is the rework that follows a wrong order. Discovering after the shoot that the line cannot be said writes off the whole sample round.
Do semi-managed Temu/SHEIN sellers need all five?
Steps 2 and 3 are enough. Without ad spend, A/B and review are meaningless — but the listing copy carries the same claim risk, and it is harder to self-audit.
What does a wrong angle look like in the data?
Decent 3-second view-through with a CTR that will not lift: they watched the opening and were not convinced. Fix the proof beat, not the hook.
Why are the priors not derived from campaign data?
Because we do not have that data and will not pretend we do. They are hand-written starting points stored next to their reasoning, reviewable and diffable. Once enough review sheets come back, real win rates replace them — that is where the value actually accrues.
Sources
- TikTok 广告政策 · 误导与虚假内容（页面自标 Last Updated: April 2026） （checked 2026-08-15）
- TikTok 竞价 In-Feed 广告规格（页面自标 Last Updated: June 2026） （checked 2026-08-15）
- Meta 广告标准 · 健康与保健 （checked 2026-08-15）