UGC with or without a creator on camera

1. Introduction

Short answer for a brand deciding between creator-on-camera UGC and hands-only UGC: in our US TikTok Shop observations, videos with the creator's face in the first second sold at a slightly higher rate per view than videos with the creator not on camera, but the gap is modest and format indices vary more widely than on-camera status. We therefore suggest treating on-camera as a hypothesis to test on your own product with a controlled comparison, not as a rule.

The question matters because a face-to-camera brief and a hands-only brief differ in cost, creator availability and the kinds of products they suit. Brands and creators have to choose before they know which video will sell, and attention online is long-tailed, with a few items taking most of the response [1]. Face-in-first-second videos carried 62.2% of sales among the watched videos; this group share does not itself demonstrate a long-tailed distribution.

Prior work on short-video popularity suggests that creator-level features matter a great deal [2] and that multimodal content features can add predictive signal [3]. Observational data also make the returns to advertising hard to measure [4]. We therefore report associations only, say where confounding is likely, and separate what we measured from what we recommend. These are US TikTok Shop observations, not all UGC everywhere.

2. Data

The window runs from 19 September to 18 October 2026. We tracked 51,843 videos across 29 categories. Our AI analyst watched and analysed 9,285 of them in depth; this is a count of videos, not of views.

For the watched videos we recorded whether the creator was on camera and, if so, whether the face appeared in the first second or later, along with the video's format. We also recorded each video's revenue and views, which give sales per view and share of sales. Shares of sales refer to sales of the watched videos.

The on-camera groups are those with a recorded on-camera status in the evidence. The evidence has no breakdown of on-camera status by category, product price or paid promotion, and it does not cross on-camera status with format.

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 of videos' revenue per view divided by the market's revenue per view, so 1 means the market average. Estimates for small groups are shrunk toward the average by empirical Bayes [5], and we report 90% intervals computed on the log scale.

Share of sales is a group's share of the watched videos' sales. We read an interval that excludes 1 as evidence that the group differs from the market average, while treating comparisons between groups as descriptive rather than formal pairwise tests. We label confidence as high or medium as given in the evidence, and we flag groups with few videos as thin. All comparisons are descriptive. We have not adjusted for product, category, price, creator or paid reach.

4. Results: on-camera status

Videos with the face in the first second (4,320 videos) sold at 1.045 times the market average per view (90% range 1.021–1.069) and made up 62.2% of sales. Videos where the face came later (1,055 videos) sold at 1.004 (0.959–1.05) and made up 13.7% of sales. Creator-not-on-camera videos (2,203 videos) sold at 0.899 (0.868–0.932) and made up 24.1% of sales. Confidence is high for all three groups.

The reported intervals for face-in-first-second and creator-not-on-camera are separated in this descriptive comparison. These are relative revenue-per-view indices, not conversion-rate percentage points or a formal pairwise causal test. The face-later interval includes the market average.

The not-on-camera group is broad. It combines every video in which the creator is absent, so it is not the same as a hands-only demonstration, which this evidence does not isolate.

5. Results: format

Among formats with ample data, voice-over B-roll sold at 1.238 (1.158–1.325; 561 videos, 5.8% of sales), unboxing at 1.201 (1.096–1.316; 309 videos), review or testimonial at 1.17 (1.121–1.22; 1,071 videos, 14.1% of sales) and talking head at 1.102 (1.048–1.159; 987 videos, 12.1% of sales). Demo or tutorial, the largest format at 3,163 videos and 36% of sales, sold at 0.961 (0.935–0.987). Skits sold at 0.888 (0.824–0.957).

Three formats rest on thin data and wide ranges. Day in the life (16 videos) sold at 1.379 (0.861–2.207), and live clips (35 videos) at 1.19 (0.821–1.726); both are medium confidence. Green screen (65 videos) sold at 1.049 (0.83–1.326). We do not draw conclusions from these three.

Format gaps are wider than the on-camera gap. Voice-over B-roll is among the highest, while creator-not-on-camera overall is below average, so absence from the frame cannot be the whole story. We cannot say from this evidence how many voice-over B-roll videos show no creator.

6. Discussion

For a brand, the measured pattern is that face-in-first-second videos sold slightly above the market average per view, and creator-not-on-camera videos slightly below it. Face-forward formats such as review or testimonial and talking head were also above average. Demo or tutorial, the format most associated with showing a product in use, was slightly below average. None of this shows that putting a face on camera makes a video sell more.

Several alternative explanations fit the data. Product and category: some products may suit a face-to-camera pitch, others a hands-only demonstration, and the evidence does not break results down by category or product. Paid reach: videos with paid promotion can reach different audiences, and observational data make the returns to advertising hard to separate from selection [4]. Creator: creator-level features carry much of the signal in popularity prediction [2], so on-camera creators may simply be stronger sellers. Shrinkage and 90% ranges guard against small-group noise, not against these confounders.

Recommendation, not a finding: a test plan for your own product. (1) Fix the product, offer, price, script and call to action. (2) Brief two variants, one with the creator's face in the first second and one hands-only, ideally from the same creator. (3) Post in the same window, with the same paid-promotion setting for both or none for both. (4) Decide in advance that the metric is revenue per view, and the rule for calling a result. (5) Record category, price point and format for every video. (6) Repeat with more than one creator before changing your default brief. (7) Treat a small difference as inconclusive, since the observed gap in our data is itself small.

7. What this means for brands and founders

• Test the opening before you change the brief: videos with a face in the first second sold at 1.045 times the market average per view (range 1.021–1.069), against 0.899 (0.868–0.932) with the creator not on camera, a small gap worth checking on your own product.

• Do not treat hands-only as one format: voice-over B-roll sold at 1.238 times the average per view (1.158–1.325), while demo or tutorial sold at 0.961 (0.935–0.987) across 3,163 videos.

• Consider a review or testimonial brief as one arm of your test: this format sold at 1.17 times the average (1.121–1.22) across 1,071 videos and carried 14.1% of watched-video sales.

• Keep paid reach equal across test variants: our figures do not separate paid from organic, so an unequal budget could look like an on-camera effect.

• Do not build a plan on thin formats: day in the life (16 videos) had a range of 0.861–2.207, too wide to act on.

8. Conclusion

In US TikTok Shop videos over 19 September to 18 October 2026, a face in the first second was associated with 1.045 times the market average sales per view (1.021–1.069), against 0.899 (0.868–0.932) when the creator was not on camera. The gap is small, and format differences are larger.

The evidence does not isolate hands-only demonstrations and does not control for product, category or paid reach. A matched test on your own product is a better basis for the decision than these averages.