Deal-led openings and urgency angles sell more per view than the demo-heavy mainstream of TikTok Shop

1. Introduction

Sellers on TikTok Shop choose how a video opens, what format it takes, which persuasive angle it leads with and how it is produced. These choices are cheap to change, so knowing which ones are associated with more sales for the attention they receive is useful to creators and brands alike. We focus on sales per view rather than total sales, because total sales mostly reflect how many people a video reached.

Prior work shows that attention to online content is long-tailed and that early popularity predicts later popularity [1], so raw sales are dominated by a few heavily viewed videos. Research on short-video popularity prediction has built multimodal benchmarks [2], and the strongest recent systems find that creator features matter most [3]. This suggests creative choices explain only part of performance, a point we return to when interpreting our results.

We therefore ask a narrower question: holding reach aside, which creative choices sell above or below the market average per view, and how much of the market's sales currently flows through each?

2. Data

The sample is 13,309 top-selling TikTok Shop videos across 29 categories, observed between 2 September and 1 October 2026. Of these, 3,340 were watched and analysed in depth. This is a single 30-day window, so seasonal effects cannot be separated from stable patterns.

For each watched video we recorded the hook type (how the video opens), the format, the main persuasive angle and the production style, alongside sales and views. Not every watched video received a label in every dimension, so group counts within a dimension do not sum to 3,340.

Because the sample consists of top-selling videos, all comparisons are among relatively successful content. Results describe what separates stronger from weaker sellers within that group, not what separates sellers from non-sellers.

Daily rankings of top-selling TikTok Shop videos, products and categories 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 and hook source, 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 2026-09-02 to 2026-10-01: 13,309 videos tracked, 3,340 watched and analysed in total. Predictions come from a logistic model retrained daily and tested on recent weeks it had not seen.

3. Methods

For each creative choice we compute a sales-per-view index, where 1 equals the market average. Groups with few videos produce noisy raw estimates, so we apply empirical-Bayes shrinkage that pulls small-group estimates toward the market mean in proportion to their uncertainty [4]. We report each index with a 90% interval and a confidence grade; groups with very few videos are flagged as low or medium confidence.

Separately, we trained a predictive model on 2,667 rows to estimate whether a video falls among the stronger sellers, using its creative features. The model was tested on a time split, training on earlier videos and evaluating on later ones, and scored by the area under the ROC curve (AUC), which measures how well it ranks stronger above weaker videos [5]. The AUC range comes from bootstrap resampling [6]. The model was not restricted to analyses that never saw sales (cleanOnly is false), so some leakage from outcomes into features cannot be ruled out and its score may be optimistic.

The model achieved an AUC of 0.64 (90% range 0.57–0.70). This is above chance (0.5) but modest, and the drivers it reports, expressed as odds ratios, should be read as associations rather than levers.

4. Results: openings

Three hooks sell clearly above the market per view: price or deal openings at 1.53× (90% range 1.37–1.71, 193 videos), bold claims at 1.28× (1.20–1.37, 478 videos) and questions at 1.27× (1.13–1.43, 184 videos). Bold claims combine efficiency with scale, holding 16.1% of watched-video sales.

The most common hook, the relatable POV, holds the largest share of sales at 20.2% but sells at roughly market rate, 1.05× (0.99–1.11). Problem call-outs, with 14.2% of sales, sit at 1.00× (0.93–1.08).

Two openings sell below the market. Going straight into the demo sells at 0.89× (0.80–0.98, 242 videos), and unboxing-reveal openings at 0.88× (0.76–1.02), though the latter's range touches the average. List or countdown openings (20 videos) and the residual other category (24 videos) are too thin to call.

5. Results: angles

Angle shows the widest spread of any dimension. Urgency or scarcity sells at 1.64× (1.40–1.92), but only 83 videos use it and it holds 2.3% of sales. Value or deal angles sell at 1.28× (1.19–1.37) across 515 videos and 16.7% of sales, the strongest combination of efficiency and scale. Social proof angles sell at 1.26× (1.06–1.51), from 66 videos.

Problem-to-solution is the dominant angle, used by 989 videos and holding 33.5% of sales, yet it sells at market rate, 1.01× (0.96–1.06). Lifestyle, transformation and education angles are also indistinguishable from the average.

Three angles sell clearly below the market: gifting at 0.81× (0.71–0.92), curiosity at 0.67× (0.59–0.75) and humour at 0.61× (0.50–0.75). Their intervals sit entirely below 1. The gifting result may partly reflect timing, since the window falls before the main gifting season.

6. Results: format and production

Unboxing is the most efficient format at 1.37× (1.20–1.57, 135 videos), followed by voice-over B-roll at 1.29× (1.18–1.41), review or testimonial at 1.27× (1.17–1.37) and talking head at 1.18× (1.09–1.28). Together these four hold about 39% of sales. Demo or tutorial is by far the most common format, with 1,183 videos and 33.8% of sales, but sells at 0.92× (0.88–0.96). Skits sell at 0.86× (0.77–0.97).

Unboxing as a format sells well above average, while unboxing as an opening hook sits slightly below it. This suggests the value lies in the full reveal rather than in leading with it, though the two labels capture different things and the comparison is indirect.

Filmed live action dominates production, with 2,054 videos and 92.9% of sales at 1.04× (1.00–1.07). Photo slideshows sell at 0.67× (0.60–0.75, 147 videos), and mixed media at 0.68× (0.48–0.98, 24 videos). AI-generated footage shows a high point estimate of 1.40×, but it rests on 17 videos and its range (0.99–1.99) includes the average. AI avatar presenters (5 videos) cannot be assessed. The model's largest driver is likewise filmed live action, with an odds ratio of 2.54.

7. Discussion

For creators and brands, the pattern is that the market's default choices, namely demo-led formats, problem-to-solution angles and relatable openings, carry most sales but at or below average efficiency. Less common choices that put price, value, urgency or other buyers' experience up front sell more per view. Reviews, unboxings and voice-over B-roll are associated with higher sales per view, while slideshows, humour and curiosity are associated with lower. The model points the same way: a visual pattern-break within reviews has an odds ratio of 1.92. By contrast, showing the product in the first second (0.62) and education angles (0.59) are associated with lower odds of being a stronger seller.

These results are consistent with prior findings that creative content explains only part of performance [3]. Our model's AUC of 0.64 (0.57–0.70) indicates that creative features rank videos only modestly better than chance.

Several alternative explanations apply. Deal and urgency content may coincide with real discounts, so the price rather than the framing may drive sales. Paid boosting appears among the model's drivers (odds ratio 1.59), and boosted videos may differ in both creative and reach. Creator audience and category mix are not controlled for in the indices. Finally, the most efficient choices are rare, and their efficiency may not hold as more sellers adopt them.

8. Conclusion

Across 29 categories in one 30-day window, the creative choices that carry most TikTok Shop sales are not the ones that sell most per view. Urgency angles (1.6×), price or deal hooks (1.5×) and unboxing, review and voice-over formats (1.3–1.4×) sell above the market. Photo slideshows, humour and curiosity sell well below it.

These are associations among top-selling videos, measured over a short window, with a model that offers only a modest predictive signal and was not restricted to sales-blind analyses. They are best used as hypotheses for creative testing rather than as rules, and should be re-checked as the window lengthens.