UGC calls to action for TikTok Shop: a testing guide

1. Introduction

Short answer: in our US TikTok Shop sample, videos that prompted viewers toward the orange cart sold at higher sales per view than videos with no call to action, so a clear cart prompt is a reasonable first thing to test. Deal-led and urgency-led framing also sold above average, but you should test them only where the deal, deadline or stock limit is real. Test one prompt at a time against a no-prompt version rather than copying a pattern from this paper.

The question is which purchase prompts in creator-style (UGC) videos are associated with more revenue per view, and how a brand or creator can test them honestly. It matters because a purchase prompt can be varied in a brief while keeping product, creator and budget comparable.

Prior work gives reasons for caution. Attention to online content is long-tailed and early popularity predicts later popularity [1]. Recent short-video popularity work finds creator features matter most [2], so differences between prompts may partly reflect who made the video. Returns to advertising are also hard to measure from observational data [3]. We therefore describe associations only and treat the final test plan as a proposal.

2. Data

Our analysis covers US TikTok Shop from 19 September to 18 October 2026. We tracked 51,843 videos, and our AI analyst watched and analysed 9,285 of them in depth across 29 categories. "Watched" is a count of analysed videos, not of views.

For each analysed video we recorded the type of call to action (for example orange cart, link or showcase, comment prompt, limited offer, or none) and the creative angle (for example value or deal, urgency or scarcity, social proof, curiosity). Group sizes range from 104 videos (comment prompt) to 4,429 videos (orange cart) for call-to-action types. Every group in this paper carries a high confidence label, but the window is a single month and some groups are small.

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

Sales per view is a group's revenue per view divided by the market's, so 1 means the market average and 1.2 means 20% above it. Estimates for small groups are shrunk toward the average by empirical Bayes [4], and the 90% ranges are computed on the log scale. A range that includes 1 means we cannot distinguish the group from the market average.

Share of sales is a group's share of the revenue of the watched videos. Groups are not mutually exclusive across the two classifications: one video has one call-to-action type and one angle. The figures are not adjusted for creator, product, category, price or paid promotion.

4. Results: call-to-action type and sales per view

Link or showcase prompts sold at 1.11× the market's sales per view (90% range 1.04–1.19×, 532 videos) and orange cart prompts at 1.07× (1.05–1.10×, 4,429 videos). Comment prompts sat at 1.10× but with a wide range (0.96–1.26×, 104 videos), so that group is consistent with no difference.

Videos with no call to action sold at 0.90× (0.87–0.93×, 3,168 videos). Limited offer prompts were at 0.90× too (0.79–1.03×, 124 videos), with a range that includes 1, so we cannot say they differ from average.

The gap between the orange cart and no-prompt groups is small in size but its ranges do not overlap. It is an association, not evidence that adding a prompt to a given video would raise its sales.

5. Results: where the sales come from

Orange cart prompts accounted for 53.9% of watched-video sales and videos with no call to action for 36.5%. Link or showcase accounted for 6.3%, limited offer for 1.7% and comment prompts for 1.6%.

Most revenue in the sample therefore comes from videos that either prompt toward the cart or do not prompt at all. The smaller prompt types are thin: their estimates are less informative about what would happen if more videos used them.

6. Results: creative angle and sales per view

Among angles, urgency or scarcity sold at 1.35× the market's sales per view (90% range 1.22–1.49×, 235 videos, 3% of sales). Value or deal sold at 1.26× (1.21–1.32×, 1,325 videos, 15.8% of sales). Social proof was 1.21× (1.10–1.33×) and education 1.18× (1.06–1.32×).

Problem → solution, the largest group at 33.4% of sales, sat at 0.98× (0.95–1.00×). Seasonal moment (0.92×, 0.87–0.97×), gifting (0.87×, 0.81–0.95×) and curiosity (0.71×, 0.65–0.77×, 460 videos) sold below average. Humour (0.88×, 0.76–1.02×) is not clearly different from average.

Our classification records the angle of the video, not whether a stated deadline, discount or stock limit was genuine. We cannot tell from this evidence whether urgency worked because it was real, or whether genuine scarcity simply coincides with products that sell well.

7. Discussion

For brands and creators, the pattern is that explicit buying direction and concrete reasons to buy (a deal, social proof, education) are associated with more sales per view, while vague hooks such as curiosity are associated with fewer. This fits the general caution in prior work: creator features matter [2] and observational returns are hard to read [3], so these are leads for testing, not rules. Alternative explanations include creator quality, product price and category mix, paid promotion, and the fact that deal-led videos may be made for products already discounted.

Recommendation, not a finding: a test plan. (1) Pick one product and keep price, offer and creator brief fixed. (2) Change only one element, for example a spoken cart prompt versus no prompt, or a deal angle versus a problem → solution angle. (3) Use only real offers: state a discount, deadline or stock level only if it is true and will be honoured, and do not simulate urgency or scarcity. (4) Run variants over the same period and with similar paid and organic support, and record whether any paid promotion was used. (5) Compare revenue per view, not views alone, and note the number of videos per variant. (6) Treat small gaps or overlapping results as no difference and repeat before scaling. (7) Keep a no-prompt control in every round.

This plan is our suggestion. We have not measured it, and the data here does not show that following it will increase sales.

8. What this means for brands and founders

• Test a clear cart prompt: orange cart videos sold at 1.07× the market's sales per view (90% range 1.05–1.10×) against 0.90× (0.87–0.93×) with no call to action, so include a no-prompt control in your first test.

• Lead with a real deal where you have one: the value or deal angle sold at 1.26× (1.21–1.32×) across 1,325 videos, but only state discounts you will actually honour.

• Treat urgency as a test, not a template: the urgency or scarcity angle was at 1.35× (1.22–1.49×) from only 235 videos, so use it only for genuine deadlines or stock limits.

• Rethink curiosity-only hooks: the curiosity angle sold at 0.71× (0.65–0.77×), the lowest group, so pair any teaser with a concrete reason to buy.

• Don't read much into limited offer wording or comment prompts yet: they were at 0.90× (0.79–1.03×) and 1.10× (0.96–1.26×), and both ranges include 1.

9. Conclusion

In US TikTok Shop videos from 19 September to 18 October 2026, cart and link prompts were associated with higher sales per view than no call to action (1.07× and 1.11× against 0.90×), and urgency, deal, social proof and education angles were associated with above-average sales per view. Curiosity was associated with the lowest (0.71×).

These are descriptive results from one month, one market and an unadjusted comparison. They are a basis for honest, controlled tests of purchase prompts, not a guarantee that any prompt will lift sales.