Impact of internet celebrities’ short videos on audiences’ visit intentions: Is beauty power?

Social media is ubiquitous in tourism, and short videos on social platforms influence travel decisions. Consumers often seek content from opinion leaders, with marketers turning to online influencers. Internet celebrities act as opinion leaders in fan communities, fostering engagement and content co-creation. Mobile short-form videos enhance immersion, social presence, and entertainment, reshaping destination marketing toward a consumer-oriented model. In China, destinations have rapidly gained fame via influencers’ short travel videos (e.g., Zibo Barbecue), though some videos receive lukewarm responses, raising questions about effectiveness. Internet celebrities differ from traditional celebrities: they have limited public credibility, deeper field engagement, and often use beauty filters, highlighting characteristics of trustworthiness, expertise, and attractiveness. Short video dissemination is monetized based on popularity metrics, prompting influencers to manage content and attempt congruence with destinations; however, some popular videos do not translate into visitation, complicating marketing outcomes. Prior tourism research has focused on traditional celebrity endorsements or official promotional videos; the third-party role of internet celebrities remains underexplored. This study combines source credibility theory and the match-up hypothesis to model how internet celebrities’ characteristics influence audiences’ attitudes toward short videos and destinations, and their visit intentions, and tests which characteristics (trustworthiness, attractiveness, expertise, and celebrity–destination congruence) significantly affect visit intention, aiming to inform theory and provide practical guidance for DMOs.

The literature review situates internet celebrities (social media influencers) as distinct from traditional celebrities, gaining followers through stylized content and integrating endorsements within personal narratives. Short travel videos with narrative and destination information shape audiences’ cognition, evaluation, and aspiration toward destinations and can drive willingness to visit or recommend. Internet celebrities help spotlight obscure destinations and generate abundant user information and eWOM effects on destination image. Beyond visual and audiovisual elements, the endorser dimension requires exploration. Source credibility theory posits that credibility of the information source drives persuasion; in tourism, credibility is crucial due to intangibility and risk. Source credibility comprises trustworthiness (objectivity and honesty), expertise (knowledge and experience), and attractiveness (physical and other attributes). While attractiveness is often operationalized as physical attractiveness in practice, online filter use complicates its role. The review proposes: H1a–c (trustworthiness positively affects attitudes toward short video and destination, and visit intention); H2a–c (expertise positively affects the same outcomes); H3a–c (attractiveness positively affects the same outcomes). The match-up hypothesis suggests endorsements are more effective when endorser image aligns with product/brand; applied to tourism, internet celebrity–destination congruence may drive attitudes and intentions. Thus H4a–c propose positive effects of congruence on attitudes and intentions. Drawing on the dual mediation hypothesis and the Destination Advertising Response (DAR) model, attitudes toward the short video should influence attitudes toward the destination (H5a), which in turn should influence visit intention (H5b). A conceptual model integrates source credibility, match-up congruence, and dual mediation pathways.

Design and case selection: The study used a case-based survey centered on FangQi Kiki, a prominent TikTok (Douyin) short travel video creator with ~20 million followers and 456 posts. A representative 1:06 video, “Big Fish and Begonia,” filmed at Fujian Tulou (Yongding and Nanjing), with 1.97 million likes and 89,000 comments, was embedded in the questionnaire. Measurement: The questionnaire had four parts: (1) demographics (gender, age, education, social platforms used, daily short video viewing frequency, familiarity with FangQi Kiki); (2) attitudes toward the short video (ATV), attitude toward the destination (ATD), and visit intention (VI); (3) internet celebrity’s trustworthiness (ICT), expertise (ICE), and attractiveness (ICA); (4) internet celebrity–destination congruence (C/D-C). All constructs (parts 2–4) used 7-point Likert scales (1=strongly disagree to 7=strongly agree). Items were adapted from prior literature: ATV (Choi and Rifon, 2012), ATD (Xu and Pratt, 2018), VI (single item), ICT (Erdogan, 1999), ICE (Halima et al., 2017), ICA (Zhang et al., 2022), C/D-C (Xu and Pratt, 2018). Sampling and data collection: Initial snowball pretest informed refinements. The formal survey used a mix of random and stratified sampling on SoJump. Respondents watched the embedded short video and then completed the survey. Data collection periods: March 26–April 8, 2023 and October 28–November 5, 2023. Of 783 responses, 134 invalid (missing values, straightlining, residing in Fujian Province) were removed, yielding 649 valid cases (effective rate 82.89%). Sample distribution spanned 116 cities in 31 provinces/regions across China; distance to Fujian Tulou ranged ~360–4600 km. Demographics: 48.4% male, 51.6% female; age: 18–26 (53.6%), 27–35 (25.9%), 36–45 (11.1%); education: below college (18.5%), junior college (21.4%), undergraduate (52.2%), postgraduate+ (7.9%); platforms used: WeChat (93.8%), TikTok (77.3%); daily short video viewing: occasionally (33.3%), often (52.5%); familiarity with FangQi Kiki: unfamiliar (46.2%), know something (37.0%), often watch (12.6%), fan (4.2%). Analysis: Reliability and validity assessments showed strong internal consistency: Cronbach’s alpha 0.865–0.958; AVE 0.719–0.857; CR 0.911–0.968. KMO 0.748–0.904; Bartlett’s test p<0.001; discriminant validity supported (square roots of AVE exceeded inter-construct correlations). Structural equation modeling (SEM) using Amos 26.0 tested the hypothesized model. Model fit indices indicated good fit: χ²/df=2.348, CN=319, RMSEA=0.046, RMR=0.044, GFI=0.933, AGFI=0.909, CFI=0.980, NFI=0.966, TLI=0.975. Mediation was tested with bias-corrected bootstrapping (2000 resamples; 95% CI). Multiple-group analyses examined differences across familiarity levels; a one-way ANOVA also tested familiarity effects on visit intention.

• Hypothesis testing (standardized paths):
  • Trustworthiness (ICT):
    • ICT → ATV: β=0.233, p=0.002 (supported, H1a)
    • ICT → ATD: β=0.139, p=0.021 (supported, H1b)
    • ICT → VI: β=0.167, p=0.023 (supported, H1c)
  • Expertise (ICE):
    • ICE → ATV: β=0.144, p=0.158 (not supported, H2a)
    • ICE → ATD: β=-0.012, p=0.884 (not supported, H2b)
    • ICE → VI: β=0.316, p=0.001 (supported, H2c)
  • Attractiveness (ICA):
    • ICA → ATV: β=0.071, p=0.415 (not supported, H3a)
    • ICA → ATD: β=0.047, p=0.478 (not supported, H3b)
    • ICA → VI: β=-0.076, p=0.349 (not supported, H3c)
  • Celebrity–destination congruence (C/D-C):
    • C/D-C → ATV: β=0.141, p=0.031 (supported, H4a)
    • C/D-C → ATD: β=-0.010, p=0.843 (not supported, H4b)
    • C/D-C → VI: β=0.009, p=0.883 (not supported, H4c)
  • Dual mediation links:
    • ATV → ATD: β=0.683, p<0.001 (supported, H5a)
    • ATD → VI: β=0.216, p<0.001 (supported, H5b)
• Mediation (95% CI):
  • ICT → ATV → ATD → VI: CI [0.010, 0.104] (significant)
  • ICT → ATD → VI: CI [0.001, 0.091] (significant)
  • ATV → ATD → VI: CI [0.079, 0.237] (significant)
  • C/D-C → ATV → ATD → VI: CI [0.000, 0.069] (not significant; includes 0)
• Model fit: χ²/df=2.348; RMSEA=0.046; RMR=0.044; GFI=0.933; AGFI=0.909; CFI=0.980; NFI=0.966; TLI=0.975; CN=319.
• ANOVA on familiarity and visit intention: F=15.067, p<0.001; highest VI in the “often watch her short videos” group.
• Multiple-group SEM by familiarity: Path coefficients varied in magnitude but not in significance across groups; effect of expertise on VI (ICE → VI) decreased with increasing familiarity (estimates approximately 0.293 to 0.208 across groups).

Findings indicate that for internet celebrity short travel videos, credibility-related attributes matter most for driving visit intentions. Trustworthiness significantly influenced attitudes toward the video and destination, and directly increased visit intention, underscoring the importance of authenticity, objectivity, and sincerity in a context where visual enhancements are common and tourism decisions are risky. Expertise did not shape attitudes toward the video or destination but directly boosted visit intention, suggesting audiences leverage perceived knowledge and experience to reduce uncertainty and inform travel choices; however, its effect weakens with increasing familiarity with the influencer. Physical attractiveness did not affect attitudes or intentions, contradicting the “beauty is power” heuristic; in short video contexts with ubiquitous filters, audiences prioritize content and narratives over looks. Celebrity–destination congruence improved attitudes toward the short video but did not translate into destination attitudes or visit intention, suggesting a divergence from traditional advertisement-based match-up effects; while congruence may increase online popularity, travel decisions hinge more on destination merits. The dual mediation path from video attitude to destination attitude to visit intention was strong, reaffirming DAR and dual mediation mechanisms: liking the short video elevates destination attitude, which increases visit intention. Overall, content quality and source credibility outweigh physical attractiveness and congruence in this influencer-driven, content-marketing environment, distinguishing it from conventional celebrity endorsements where endorser image can dominate.

The study develops and validates a conceptual model integrating source credibility theory, the match-up hypothesis, and the DAR framework to explain how internet celebrities’ short travel videos influence visit intentions. Using a large sample exposed to a representative TikTok travel video, results show trustworthiness and expertise significantly drive visit intentions, while physical attractiveness and celebrity–destination congruence do not. Attitudes toward the short video strongly shape destination attitudes and intentions, evidencing dual mediation processes. The work differentiates influencer-based short video marketing from traditional celebrity endorsements, highlighting an “Appearance Fallacy” in online contexts and emphasizing that credible, expert-led, authentic, and narrative-rich content is more impactful than endorser looks. Future research should extend beyond a single-influencer case, test moderating roles of beauty/filter intensity, and examine narrative and co-creation dynamics to deepen understanding of how short video content shapes pre-travel experiences and actual visitation.

• Single case study centered on one internet celebrity (FangQi Kiki) may limit generalizability; future multicase, mixed-method designs (including content analysis, web text, interviews) are recommended.
• Desensitization to physical attractiveness due to beauty/filter use was not directly modeled; future experiments should test beauty/filter degree as a moderator on perceptions and travel intentions.
• The role of personal narratives and story-sharing warrants further investigation to assess effects on video experience and on-site travel experiences.
• Experience co-creation between influencers and destinations remains underexplored and merits in-depth study.