UGC scripts for TikTok Shop: building an evidence-led brief

1. Introduction

Short answer for a brand deciding what to brief: build each UGC script from three parts (a hook, a way of showing the product, and an explicit call to action) and treat the hook as the first thing to test. In our US sample, price or deal hooks, bold claims, review or testimonial formats, voice-over B-roll and an on-screen link or cart call to action were each associated with higher sales per view than the market average, while skits, story openers and videos with no call to action were associated with lower. None of this shows that a script change causes more sales.

The question matters because creators are often briefed on style alone, and brands lack a shared vocabulary for what to ask for. Prior work on short-video popularity suggests that who posts matters a great deal: the winning solution to a recent popularity-prediction challenge found creator features matter most [1], and multimodal benchmarks show that content features add signal but remain hard to isolate [2]. Attention is also long-tailed and early popularity predicts later popularity [3]. Measuring the returns to advertising from observational data is hard in general [4].

We therefore ask a narrower question: among videos that sold on US TikTok Shop, how do sales per view differ by hook, format and call to action, and what does that suggest for a testable script brief?

2. Data

The window runs from 19 September to 18 October 2026. Our analysis tracked 51,843 videos across 29 categories on TikTok Shop in the US. Our AI analyst watched and analysed 9,285 of them in depth. That figure counts videos, not views.

For each analysed video we recorded the hook type, the format and the call to action, and linked it to the video's sales and views. Each of the three tables below covers a subset of the analysed videos, so group counts differ between hooks, formats and calls to action and should not be added across tables. The findings describe US TikTok Shop observations only; they do not describe UGC on other platforms or in other countries.

Daily rankings of top-selling TikTok Shop videos in the US were collected from several independent sources and merged into one record per video. Where sources overlap, their sales estimates are cross-checked against each other, and videos whose estimates disagree by more than half are flagged. An AI analyst watched each video and recorded its opening, format, angle, production style, use of AI, hook source, length, pacing, what appears in the first second and how it is presented (people on screen, voice-over, setting, UGC or studio feel), without seeing how the video sold. Sales per view is a video group’s revenue per view divided by the market’s, shrunk toward the average for small groups, with 90% ranges. This paper covers 2026-09-19 to 2026-10-18: 51,843 videos tracked, 9,285 of them watched and analysed.

3. Methods

The statistic reported throughout is sales per view: a group of videos' revenue per view divided by the market's revenue per view, so 1 equals the market average. Estimates for small groups are shrunk toward the average by empirical Bayes [5], which pulls thinly populated groups toward 1. We report 90% intervals computed on the log scale.

Share of sales is each group's share of the sales of the watched videos. A group can hold a large share of sales simply by being common, so we read share and sales per view together: share shows where volume sits and sales per view shows relative efficiency.

We do not adjust for category, price, creator or paid promotion in these tables. No predictive model is reported in this paper.

4. Results: hooks

Price or deal hooks had the highest sales per view at 1.5× the market (90% range 1.37–1.58), on 492 videos and 5.6% of sales. Bold claims followed at 1.2× (1.19–1.29) with 1,361 videos and 17% of sales, and result-first hooks at 1.2× (1.07–1.31) with 3.7% of sales. Social proof, at 1.10× (1.00–1.22), had a range that reaches down to the market average.

The largest groups by share sat close to average. Relatable POV held 18.5% of sales at 0.98× (0.95–1.02) and problem call-outs held 13.5% at 0.98× (0.94–1.02). Straight into the demo was lower at 0.92× (0.87–0.98), and story openers at 0.90× (0.81–1.00). Question hooks (1.03×) and unboxing reveals (1.01×) had ranges spanning 1.

The ordering is informative, but the intervals for many middle hooks overlap, so we would not rank them finely.

5. Results: formats

Voice-over B-roll sold at 1.2× per view (1.16–1.33) on 561 videos, unboxing at 1.2× (1.10–1.32) and review or testimonial at 1.2× (1.12–1.22), the last carrying 14.1% of sales. Talking head was 1.10× (1.05–1.16) with 12.1% of sales.

Demo or tutorial was the largest format by far, with 3,163 videos and 36% of sales, but sold at 0.96× (0.94–0.99). Skits were lower at 0.89× (0.82–0.96) with 5.5% of sales. Routine / GRWM (1.00×) was at the market average.

Two formats are thin: day in the life (16 videos, 1.4×, range 0.86–2.21) and live clip (35 videos, 1.2×, range 0.82–1.73) both have wide ranges that include 1 and medium confidence, so we draw no conclusion from them.

6. Results: calls to action

Orange-cart calls to action were the most common, with 4,429 videos and 53.9% of sales, and sold at 1.07× (1.05–1.10). Link or showcase calls to action sold at 1.11× (1.04–1.19) with 6.3% of sales. Comment prompts were at 1.10× (0.96–1.26) on 104 videos, a range that includes 1.

Videos with no call to action sold at 0.90× (0.87–0.93) but still held 36.5% of sales, so a call to action is not a precondition for selling. Limited-offer calls to action were also at 0.90× (0.79–1.03) on 124 videos and 1.7% of sales, with a range that includes 1.

The differences here are smaller than for hooks. The gap between the orange cart and no call to action is roughly 1.07× against 0.90×.

7. Discussion

Read together, the tables suggest a brief that names a specific price or deal, or a bold claim, in the opening line, shows the product through a review, voice-over B-roll or unboxing, and ends with an explicit link or cart prompt. This is an association: it is consistent with, but does not show, that these choices cause higher sales per view. Alternative explanations include product and price differences, category mix, creator audience size and reputation (which prior work finds matter most [1]), and paid promotion, which we did not control for. Observational comparisons like these are known to be hard to interpret as returns [4].

Recommendation, not a finding: an annotated script skeleton. The wording below is illustrative only and makes no product claim. Hook (first seconds): one line carrying a price or deal, or a bold claim the creator can substantiate on screen. Proof: show the product being used or reviewed, in the creator's own voice, in line with the review, voice-over B-roll or unboxing formats. Demonstration: show the evidence a buyer needs. Test duration separately; the format comparison does not establish an optimal video length. Call to action: one explicit prompt to the cart or link in the final moments.

Recommendation, not a finding: a test plan. (1) Keep product, creator type, length and offer fixed. (2) Vary only the hook, for example price or deal against relatable POV. (3) Brief several creators per variant so creator effects do not drive the result. (4) Compare sales per view over the same days, and record whether paid promotion was used. (5) Repeat before changing the brief. Because the hook ranges for many types overlap, treat our ordering as a starting hypothesis.

8. What this means for brands and founders

• Test a price or deal hook first: these hooks sold at 1.5× the market average per view (90% range 1.37–1.58), though on only 5.6% of sales.

• Put a bold claim you can show on screen in the brief: bold-claim hooks sold at 1.2× per view (1.19–1.29) and carried 17% of watched-video sales.

• Ask for a review, voice-over B-roll or unboxing format rather than defaulting to a demo: demo or tutorial videos sold at 0.96× per view (0.94–0.99) despite 36% of sales, while review or testimonial sold at 1.2× (1.12–1.22).

• Include an explicit link or cart call to action: videos with none sold at 0.90× per view (0.87–0.93), against 1.07× for the orange cart (1.05–1.10), though 36.5% of sales still came without one.

• Change one script element at a time: many hook ranges overlap, so a hook-only test across several creators is a safer way to learn what works for your product.

9. Conclusion

In a US TikTok Shop sample of 9,285 videos analysed in depth, price or deal hooks (1.5×), bold claims (1.2×), review or testimonial formats (1.2×) and link or cart calls to action (1.07× to 1.11×) were associated with higher sales per view than the market average. Skits, story openers and no call to action were associated with lower.

The evidence supports a testable brief rather than a guaranteed formula. Several groups are thin or have ranges that include the market average, and we did not adjust for creator, category, product or paid promotion. We would treat the framework above as hypotheses for a controlled, hook-only test.