What sells in Health on TikTok Shop: a mid-priced, affiliate-led category with growing LIVE sales

1. Introduction

Brands and creators in Health must decide what to sell and what to film, usually with little category-level evidence. This paper asks where Health stands in the TikTok Shop market, how it sells, at what price and commission, and which video openings and formats are associated with higher sales per view.

Short-video research has mostly asked what predicts popularity. Early popularity predicts later popularity, and attention is long-tailed [1]. Current prediction work finds that creator features matter most [2] and relies on multimodal benchmarks [3]. Popularity is not sales, though, and the returns to advertising are hard to measure from observational data [4]. We therefore describe associations in a single category and do not estimate effects.

Prior work also bears on the commercial side. Live selling builds trust in social commerce [5], affiliate referral fees are set strategically and vary [6], and online retail sells a long tail beyond the bestsellers [7].

2. Data

The window runs from 2 September to 1 October 2026. Our analysis tracked 351 Health videos. Our AI analyst watched and analysed 153 of them in depth; this counts videos, not views. Marketplace figures rest on the 100 top-selling Health products. Market comparisons use all 28 categories tracked, and 3,342 videos were watched across the market.

For each watched video we recorded its opening (hook), format and production style, and its sales. For products we recorded price, affiliate commission rate, launch date and sales channel. For the category we recorded weekly revenue over the past year and counts of shops and active products. Tagged counts differ slightly between breakdowns, because not every video carries every tag.

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 and what appears in the first second, 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. Weekly category figures for the US (estimated revenue, its split between video, LIVE and the shop tab, affiliate revenue, shops and active products, and a year of weekly revenue) and daily rankings of each category’s top-selling products (unit price, commission rate, launch date and revenue by channel) were collected from several independent sources. Medians of price and commission are weighted by revenue; momentum compares the latest weeks with the same number of weeks before. This paper covers 2026-09-02 to 2026-10-01: 351 videos tracked in Health, 153 of them watched and analysed; the market baseline uses all 3,342 watched videos; 100 top-selling products in Health. Predictions come from a logistic model retrained daily and tested on recent weeks it had not seen.

3. Methods

Sales per view is a group of videos' revenue per view divided by the market's, so 1 is the market average. Small groups are shrunk toward the average by empirical Bayes [8]. We report 90% intervals on the log scale and label confidence low, medium or high by group size.

Momentum compares estimated revenue in the latest 4 and 13 weeks with the same number of weeks before. Channel shares split the latest week's revenue into shoppable video, LIVE and the shop tab, plus the share credited to affiliate creators. Prices and commission rates are for the top-selling products in the latest week. Medians are weighted by revenue, and revenue per product by band is set against the average product. Product age is measured in days from launch. Competition uses shop and product counts, average revenue per shop, and the top ten products' share of top-product revenue with the Herfindahl–Hirschman index [9].

Seasonality compares each month's revenue with the category's average month over the past year, excluding partial months; separating season from trend follows [10]. The predictive model is a logistic model of top-selling videos, retrained daily and tested on recent weeks it had not seen. We report AUC [11] with a bootstrap range [12].

4. Results: momentum, channels and competition

Health took 6.2% of market revenue in the latest four weeks and ranked 4th of 28 categories. Revenue in the latest four weeks was 9.4% above the previous four. Over 13 weeks it was 8.7% lower, against an 11.1% fall for the market. The category has therefore held up better than the market, although the short-run rise follows a softer quarter.

Video accounts for 64.3% of Health revenue, almost the same as the market's 64.9%. LIVE accounts for 25.1%, against 17.1% for the market, and the shop tab for 10.6%, against 18.0%. Creators in the affiliate programme are credited with 83.6% of category revenue.

Health has 3,688 shops and 18,535 active products, against a median of 4,461 shops across categories. Average revenue per shop is $6,166, against a median of $2,401. The top ten products take 26.7% of top-product revenue, the same as the median category, and the Herfindahl–Hirschman index among top products is 145. Revenue is spread across many products rather than a few.

5. Results: price, commission and product age

The revenue-weighted median price of top products is $29, and the middle half sold between $19 and $39. The $20–35 band holds 45 products and 42.3% of revenue, but its revenue per product is 0.94× the average. The $10–20 band reaches 1.06×. The $60–100 band reaches 1.6×, but it has only 4 products (6.4% of revenue), so we treat that as thin evidence. Only 1 product is priced at $100 or more.

The revenue-weighted median commission is 20%. The 20–30% band holds 43 products and 51.0% of revenue, at 1.10× the average product. The 15–20% band holds 30 products at 0.84×. The 10–15% band is at 1.06×. Video carries 81–83% of revenue in each of these three bands. The other bands hold 1 or 2 products each, so we draw no conclusions from them.

Products launched in the last 90 days produce 3.8% of top-product revenue. The 30–90 day band has 5 products at 0.76× the average. Products older than six months sit at 1.01–1.04×. Top sellers are mostly established products, though a short window and small counts limit what this shows.

6. Results: openings, formats and the model

Among hooks, social proof (1.50×, 90% range 1.07–2.10, 5 videos) and price or deal (1.46×, 1.05–2.03, 9 videos) have the highest sales per view, but both are low confidence. Bold claims reach 1.35× (0.99–1.85, 18 videos). Problem call-outs are the largest group, with 36 videos and 28.2% of sales, at 1.07× (0.83–1.39, high confidence). Visual pattern-breaks are the weakest at 0.72× (0.55–0.94, 16 videos). Comparison hooks reach 0.74× (0.45–1.20, 9 videos), a low-confidence result.

Among formats, review or testimonial videos reach 1.53× (1.09–2.14, 10 videos) and voice-over B-roll 1.46× (1.02–2.09, 11 videos), both medium confidence. Talking-head videos are the bulk of the category, with 74 videos and 50.4% of sales, at 1.04× (0.89–1.22). Skits are at 0.81× (0.58–1.14, 18 videos). All 78 videos tagged for production were filmed live action, at 1.08× (0.91–1.27), so production style does not separate videos in this sample.

The predictive model separates top-selling videos from others modestly: AUC 0.64 (90% range 0.57–0.70), trained on 2,667 rows. It was not trained only on analyses that never saw sales (cleanOnly is false), so it should not be read as a pure test of what can be known before publishing. Its leading drivers include a before-and-after treatment and boosting with ads, whose odds ratios of 1.66 and 1.59 are associations.

7. Discussion

For brands, the evidence points to a crowded, affiliate-led category. Mid-priced products ($20–35) hold the largest share of revenue, and commissions of 20–30% are common among the top sellers. Average revenue per shop is high relative to other categories, but the 3,688 shops and 18,535 active products show how much competition there is. LIVE accounts for a quarter of revenue, so a LIVE plan may suit this category better than the market average implies, which is consistent with the trust-building role described in prior work [5].

For creators, review, testimonial and voice-over formats and social-proof, price or deal and bold-claim hooks are associated with more sales per view. These groups are small, and some ranges cross 1. Talking-head videos and problem call-outs carry most of the category's sales at roughly average sales per view, so these are the safe baseline rather than a differentiator. Visual pattern-breaks are below average.

The season looks uneven rather than sharply peaked. Over the past year June ran at 1.19 of a normal month and March at 1.11, while August was 0.81. Halloween is under way, Black Friday / Cyber Monday starts in 40 days, holiday gifting in 45 and New Year resolutions in 92. We have only one year of monthly data and cannot separate season from trend. Alternative explanations also apply: creator size, ad spend, product quality and promotions could drive sales as much as format [4]. The model's modest AUC is consistent with this.

8. Conclusion

Health is a large and relatively resilient TikTok Shop category. It ranked 4th of 28 and held up better than the market over 13 weeks. Sales run mostly through affiliate creators, with a larger LIVE share than the market, and top products cluster around $29 and a 20% median commission.

On video, review, testimonial and voice-over formats and social-proof, deal and bold-claim hooks are associated with above-average sales per view, with the caveats of small samples and wide ranges. We recommend testing these against a talking-head baseline rather than treating them as established rules.