Social Short-Form Video Marketing Research 05/10 | Research article
An actor sits in a bedroom, faces a phone camera and says, “I have used this for three months.” Another creator, who has reviewed products in the category for years, clearly states that the brand paid for the video and demonstrates the key function from beginning to end. Which one is more “authentic”? Handheld framing, available light, conversational delivery and rough editing may make the first look more like a user post. They cannot tell us which claim deserves belief.
Our central argument is that creator trust is not a UGC look. It is an inspectable content contract. The contract has at least four conditions: provenance must be clear; the creator must be qualified to discuss the product; the product must fit the creator’s established content logic; and the commercial relationship must be disclosed intelligibly. Closeness can reduce communication distance, but it cannot replace expertise. Follower count describes scale, not incremental sales. Looking like an ordinary user cannot turn a brand-produced advertisement into an unsolicited testimonial.
Our search cutoff is 5 October 2026. We compare recent peer-reviewed studies, a dynamic theoretical model, a non-peer-reviewed longitudinal platform preprint, the public abstract of a travel-content experiment, and current official guidance from the US Federal Trade Commission and YouTube. These sources examine different markets, platforms and estimands. We therefore do not treat TikTok, Douyin, Instagram and YouTube as one environment, or engagement, credibility, purchase intention and observed sales as interchangeable outcomes.
Start by separating three content identities
Project teams often use “UGC” to mean both content provenance and a production language. That ambiguity damages the decision before filming begins.
| Content identity | Actual control relationship | What can be claimed honestly | Main risk |
|---|---|---|---|
| Genuine user post | The user publishes voluntarily; the brand has no material control or connection | This is the user’s own public expression | Reuse may remove the original context after permission is obtained |
| Brand-produced UGC-style ad | A brand or agency funds, scripts, casts and approves the work | This is an advertisement made in a personal, everyday idiom | An actor is presented as an ordinary customer with a fabricated history |
| Paid creator content | The creator has an existing audience and persona; the brand supplies payment, gifts or another benefit | This is a commercial collaboration, with truthful experience where relevant | The material relationship is hidden, or familiarity substitutes for product evidence |
All three can be rough or polished. The distinction is not grain, camera shake or bedroom styling. It is who initiated the message, who controls it, and who is responsible for its claims. ONCE’s short-form video production service makes the same distinction: when actors or models deliver UGC-style advertising, it should be described as produced advertising, not an unsolicited customer review. This is not semantic fussiness. It determines which first-person experiences a script may legitimately contain.
Closeness is not qualification
Familiarity with a creator is an important source of influence. Repeated viewing lets audiences learn how someone speaks, what they value, and what they normally reject. A commercial message therefore does not begin from zero. Yet this relationship is closer to “I know how this person tends to judge” than “this person can validate every product claim.”
Nistor, Selove and Villas-Boas’s Influencer Authenticity: To Grow or to Monetize was published online in September 2024 and appeared in Management Science in 2025. It is a dynamic theoretical model, not an advertising experiment. The authors operationalise authenticity as whether influencers endorse products consistent with their personas and preferences, then model how growth and monetisation incentives can change that choice. It helps explain why an incongruous endorsement may damage an established relationship, but it does not prove that every audience responds identically.
Cheng and Zhang’s Reputation Burning: Analyzing the Impact of Brand Sponsorship on Social Influencers, published online in October 2024 and in Management Science in 2025, is closer to observed platform behaviour. The study uses user-generated YouTube videos from prominent English-speaking beauty and style influencers. Distribution-free reweighting matches sponsored videos with comparable organic videos. Sponsored posting is associated with an average subscriber loss of about 0.19%, with a stronger relationship among larger-audience influencers. The negative relationship is mitigated when the sponsored content fits the influencer’s usual content, or when the promoted brand is less well known.
Two limits matter. First, this is a matched observational study, not an experiment that randomises sponsorship; unobserved factors may remain. Second, subscriber change on beauty YouTube does not equal sales on today’s Douyin. The more defensible lesson is that fit is not a demographic label attached to a creator. It is whether the commercial claim continues the creator’s established way of making judgements. If a creator known for ingredient analysis suddenly offers only mood, or a rigorous tester merely recites copy, the audience may perceive a temporary rewrite of the creator’s standards.
The product must participate in persuasion
Choosing an apparently appropriate creator is still not enough. Yang, Zhang and Zhang’s Engagement That Sells: Influencer Video Advertising on TikTok was published online in December 2024 and in Marketing Science in 2025. It links influencer video ads on the original Chinese TikTok, Douyin, with Taobao sales from May to November 2019. The authors construct a product-engagement score from visual salience and alignment with product placement. Their model sample contains 16,951 video ads; the sales panel includes 259 treated products.
Using posting-time variation in a difference-in-differences design, the study finds that a higher product-engagement score predicts a larger sales lift in that setting. Overall likes, comments and shares, and raw amounts of product placement, do not offer the same explanatory value. Follower count, average views and influencer fees are not significant in the reported sales model. This is not proof that followers never matter. The setting is China’s 2019 Douyin–Taobao ecosystem, treatment is not random, and explanatory power is limited. It nevertheless exposes a more productive question than “large creator or small creator?”: does attention to the creator actually attach to a product action and a reason to buy?
A creator script should therefore do more than borrow a familiar face at the opening and insert a product into the closing line. The product can enter the creator’s most credible act of judgement: a reviewer compares failure conditions; a craftsperson reveals a step in the process; a commuter states the constraints of the situation; a camera creator makes a visual difference visible within the task.
ONCE’s published Logitech MX Brio product-video case organises a brand-produced film around a creator’s late-night task. The page demonstrates how a commercial can use a creator mission to structure a product story. It is not presented as an independent endorsement to that person’s existing followers, and it reports no sales experiment. Precisely because its identity is legible, it helps show that a creator viewpoint and a creator endorsement are not the same thing.
Disclosure activates persuasion knowledge, but concealment is not a solution
Commercial disclosure is often treated as either a legal footnote or an automatic performance penalty. The evidence is more complicated.
Bertaglia and colleagues’ Influencer Self-Disclosure Practices on Instagram: A Multi-Country Longitudinal Study is a July 2024 preprint and has not been peer reviewed. It covers more than ten years, over one million Instagram posts and 400 creators across the United States, Brazil, the Netherlands and Germany. Sponsored posts receive lower engagement on average, but proper disclosure does not reduce engagement further beyond sponsored status. The design is observational rather than a random assignment of identical content to disclosed and undisclosed conditions, so it cannot prove that disclosure has no cost.
By contrast, Jhawar’s 2025 online article in the Journal of Vacation Marketing reports three between-subject experiments in its public abstract. Across fictitious and real travel-influencer settings, explicit disclosure increased avoidance of the influencer and destination, mediated by expectancy violation. We could access the formal abstract, not the restricted full text, so we do not extend its methods or effects beyond what that abstract reports. Travel context, culture and prior expectations may shape the reaction. The apparent conflict is informative: one study follows platform behaviour over time and another manipulates a specific commercial cue; one observes engagement and another measures avoidance.
The US Federal Trade Commission’s current Disclosures 101 for Social Media Influencers treats payment, employment, family relationships, and free or discounted products as material connections. It asks that disclosures be clear, hard to miss and placed with the endorsement; burying a video disclosure only in a description is generally insufficient. YouTube’s official guidance requires creators to select the paid-promotion option so that YouTube can display a disclosure at the start, while stating that creators and partners remain responsible for applicable law. These are US and YouTube requirements, not a recommendation-system formula or legal advice for every market.
The creative question is not “how can we prevent people from noticing this is an ad?” It is “how can the commercial reason remain consistent with the creator’s established standard of judgement?” A creator can disclose the collaboration at the opening and immediately explain why it was accepted, what was tested, and which conclusions are limited. Brand claims that cannot be independently verified can remain clearly attributed to brand materials rather than being converted into personal experience. Transparency makes persuasive intent visible. Concealment risks the creator’s entire content identity when the connection is discovered.
ONCE’s four-part trust audit
From this evidence, we propose a limited framework for project approval and script review. It is not a validated universal scale. It is a set of four falsifiable questions.
| Audit | Question at approval | Observable material | What cannot substitute |
|---|---|---|---|
| Provenance | Who pays, writes, controls and finally publishes? | Contract, script authority, publishing account, reuse permission | Handheld camera or domestic styling |
| Qualification | Why is this creator able to discuss this claim? | Scope of use, demonstration, experience, failure conditions | Follower count and warmth |
| Fit | Which established content logic does the collaboration continue? | Past topics, judgement standards, audience expectations, category relationship | Age, gender or other surface labels |
| Commercial transparency | Can the material relationship and creative constraints be understood easily? | Spoken and visual disclosure, platform label, separation of experience from brand claim | Vague wording deep in a description |
The four items are not a score to total. If provenance is disguised, production quality cannot compensate. If disclosure is present but qualification is absent, the creator is merely transparent about making an unsupported claim. For Spark Ads and cross-platform reuse, rights must also be managed separately from trust. ONCE’s guide to integrating UGC-style videos with brand commercials treats content roles and brand production as design choices rather than pretending that every personal-looking scene has the same origin.
A minimum test that is allowed to fail
If a team wants to learn whether the problem lies in disclosure, creator fit or product evidence, it can develop three versions that hold creator, product, offer and media conditions as constant as practical:
- Version A discloses the partnership at the opening but provides only a spoken recommendation.
- Version B uses the same disclosure and places one observable product action inside the creator’s most characteristic task.
- Version C retains that product action but is posted by the brand with an actor and clearly labelled as produced advertising.
Randomise exposure where feasible. First test whether viewers correctly identify the content’s provenance. Then measure creator credibility, comprehension of the product claim and unaided brand recall. Treat clicks, additions to cart and purchases as separate outcomes. Completion is not comprehension; engagement is not incremental sales. If B outperforms A, product action rather than disclosure tone may be responsible. If C produces higher comprehension but lower relational trust, brand control and creator relationships are providing different value. If there is no difference, the sample may be weak, the claim undifferentiated, platform delivery may have selected different audiences, or the creator may never have fit.
This is a proposed experiment, not an ONCE client experiment or reported result. Real campaigns also face recommendation dynamics, frequency, competitive activity and cross-platform search.
Conclusion: replace “does it look real?” with “what is true?”
A UGC look can reduce formal distance without supplying genuine provenance. Creator familiarity can make people willing to listen without conferring professional qualification. Disclosing a partnership can change response, but that does not make concealment a valid optimisation tactic.
Trust in short-form video comes from relationships that can be questioned: who this person is, why they are qualified to say this, what the product actually does in the frame, and what material connection exists between brand and creator. The four-part audit converts authenticity from an aesthetic judgement into a production decision and makes failure more diagnosable. Not every film needs to resemble a customer selfie. Every film does need provenance, qualification, fit and transparency that agree with one another.
For a programme that must coordinate brand production, creator collaboration and platform versions under one identity policy, begin with ONCE’s short-form video production service. A useful brief first answers three questions: who is speaking, why are they entitled to make this claim, and how will the audience know that a commercial relationship exists?
Sources and reading scope
- Jeremy Yang, Juanjuan Zhang and Yuhan Zhang, Engagement That Sells: Influencer Video Advertising on TikTok, Marketing Science 44(2), 2025, DOI 10.1287/mksc.2021.0107. Formal full text, methods and results tables read.
- Mengjie Cheng and Shunyuan Zhang, Reputation Burning: Analyzing the Impact of Brand Sponsorship on Social Influencers, Management Science 71(7), 2025, DOI 10.1287/mnsc.2023.00193. Formal page and public abstract read.
- Cristina Nistor, Matthew Selove and J. Miguel Villas-Boas, Influencer Authenticity: To Grow or to Monetize, Management Science 71(6), 2025, DOI 10.1287/mnsc.2023.03479. Formal page and model scope in the author manuscript checked.
- Thales Bertaglia, Catalina Goanta, Gerasimos Spanakis and Adriana Iamnitchi, Influencer Self-Disclosure Practices on Instagram: A Multi-Country Longitudinal Study, arXiv:2407.09202, submitted July 2024. Preprint, not peer reviewed; abstract and version record read.
- Anand Jhawar, Examining the consequences of sponsorship disclosures on travel influencer-social media user relationships: A journey of expectancy violation, Journal of Vacation Marketing, published online 25 September 2025, DOI 10.1177/13567667251382467. Formal abstract only; we do not claim access to the restricted full text.
- US Federal Trade Commission, Disclosures 101 for Social Media Influencers, published November 2019, checked 5 October 2026; YouTube Help, paid-promotion disclosure guidance, checked 5 October 2026. They are used for the US and YouTube contexts, not as legal advice for China.
