A methodological scoping review for video analysis in hospitality research Open Access

The digital transformation of the hospitality industry, driven by rapid advancements in information and communication technologies, has reshaped the landscape of customer engagement and operational strategies (Buhalis and Law, 2008; Ali and Frew, 2014; Xiang et al., 2015; Moreno and Tejada, 2019; Law et al., 2020; Rodrigues et al., 2023). Within this evolving digital ecosystem, video content has emerged as a particularly powerful medium for communication, marketing and service delivery in the hospitality sector (Hudson et al., 2012; Coker et al., 2021; Deng et al., 2021; Agrawal and Mittal, 2022).

The growth of video-sharing platforms, such as YouTube and TikTok, has revolutionized how hospitality stakeholders create, share and consume content (Yetimoğlu and Uğurlu, 2020; Zhu et al., 2022; Ercegovac et al., 2023). These platforms have near-supplanted traditional communication channels (e.g. newspapers, radios), offering new opportunities for immersive storytelling, virtual property tours and real-time event showcases (Leung et al., 2017; Tiago et al., 2019; Briliana et al., 2020; Pratisto et al., 2022). The strategic importance of video content is reflected in general business projections, with 89% of businesses expected to use video as a marketing tool in 2025 and 95% of video marketing professionals regarding it as an essential element of their overall marketing strategy (Wyzowl, 2025).

The surge in video content production necessitates an increased focus on video analysis techniques in hospitality research. The potential for extracting meaningful insights from video data spans multiple domains, such as consumer behavior analysis, marketing strategy evaluation, and service quality assessment. However, the volume and complexity of video data may present challenges to researchers employing traditional analytical methods (Núñez et al., 2024). For instance, manual coding approaches in qualitative content analysis, while valuable for in-depth exploration, typically fall short when researchers attempt to process the extensive multimodal data contained in video material (Schreier, 2012). The multifaceted nature of video, encompassing visual, auditory, and textual elements, demands more sophisticated analytical techniques capable of effectively synthesizing these diverse data types (Grzenkowicz and Wildfeuer, 2025).

Despite the growing importance of video analysis in hospitality research, the field lacks a systematic understanding of current methodological practices. Zhu and Cheng (2024) recently conducted a review of automatic video analytics in tourism, providing a technical framework that advocates computational approaches to overcome scalability limitations of manual coding. However, the analytical approaches hospitality researchers are often employing, the types of video content and data components being analyzed, and the methodological patterns and gaps across the broader hospitality field remain unexamined. This absence of empirical evidence limits the field's ability to assess current practices and identify strategic directions for methodological advancement.

To address this gap, we conducted a methodological scoping review of video analysis techniques employed in hospitality research. A scoping review approach allowed us to systematically map the breadth of methodological approaches across this emerging research area (Arksey and O'Malley, 2005; Munn et al., 2018). Specifically, this review aimed to (1) examine the types and characteristics of video content being analyzed in hospitality research, (2) identify methodological trends in video analysis within the field and (3) propose recommendations for future research directions and methodological advancements.

We conducted this methodological scoping review following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines (Tricco et al., 2018). Specifically, we performed the literature search using Web of Science and Scopus, two databases commonly employed by researchers in hospitality studies (e.g. Gomezelj, 2016; Mariani et al., 2018; Mehraliyev et al., 2021). Search terms included combinations of (“video” OR “video on demand” OR “VOD” OR “user generated video” OR “vlog*” OR “video blog” OR “social media video”) AND (“content analysis” OR “thematic analysis” OR “textual analysis” OR “sentiment analysis” OR “video analytics” OR “natural language processing” OR “machine learning” OR “big data analytics” OR “engagement metrics” OR “performance metrics” OR “multimodal analysis”), using Boolean AND logic to require studies to include both video-related terminology and analytical method terms. Video-related terms encompassed various formats (e.g. vlogs, social media videos), while analytical method terms included both traditional qualitative approaches (e.g. content analysis, thematic analysis) and computational techniques (e.g. machine learning, natural language processing, sentiment analysis). We included “multimodal analysis” because videos inherently contain multiple data types (visual, auditory, textual, metadata) that researchers may analyze individually or in combination. We managed and organized all identified studies using Rayyan, a web-based tool designed for scoping reviews (Ouzzani et al., 2016). Database searches were completed on April 18, 2025, with the screening and review process finalized by May 2, 2025.

To be included in this review, articles had to meet the following criteria: (1) published in peer-reviewed journals; (2) written in English; (3) published online up to January 1, 2025, with online first publication dates serving as our reference point; (4) related to the hospitality industry (e.g. tourism, destination marketing, food and beverage, lodging, and/or casino gambling); and (5) included methods for video analysis and/or commentary analysis. We included commentary analysis because user-generated comments on video platforms provide audience response data that many researchers use to understand video reception and impact. Lastly, we excluded review articles, editorials, conference abstracts, theses and dissertations from our analysis.

The search in Scopus returned 638 articles limited to the subject areas of Business, Management and Accounting – a specific filter provided by Scopus. The Web of Science returned 194 articles after filters were applied for Hospitality, Leisure, Sport and Tourism fields. A total of 832 articles were initially imported into Rayyan, from which 69 duplicates were removed, resulting in 763 articles for screening. Following a review of titles, abstracts and full texts, 36 articles met the inclusion criteria. An additional search using the references from the included articles and Google Scholar resulted in the inclusion of one more article, bringing the total to 37 articles included in this scoping review. Figure 1 presents a PRISMA flow diagram that illustrates the different phases of the scoping review process, from initial identification to final inclusion of studies.

This scoping review identified 37 articles that employed video analysis techniques within hospitality research. As illustrated in Figure 2, the chronological distribution of these publications (2017–2025) revealed a notable acceleration in research output beginning in 2023, with this period accounting for over 60% of the reviewed studies (23 out of 37). Table 1 provides a comprehensive overview of the key characteristics of these studies, including their research focus, methodological frameworks and analytical approaches.

The review of 37 studies revealed a diverse range of video content in hospitality research. Vlogs (i.e. video blogs) emerged as the most prevalent category, featuring in 10 studies (27%). This category encompassed travel vlogs (e.g. Dewantara et al., 2025; Li et al., 2024; Mesana et al., 2024), food vlogs (Ma et al., 2024, 2025; Rauf and Pasha, 2024) and lifestyle vlogs (Li et al., 2023). Tourism promotional ads and videos constituted the second most common category, appearing in nine studies (24%) (e.g. Avraham, 2020; Barns, 2024; Georgescu Paquin and Cerdan Schwitzguébel, 2021).

Other content types included travel-related videos (e.g. Chen et al., 2024; Motahar et al., 2024; Yoo et al., 2024a), food and fashion promotion videos (Sharma, 2025; Tavakoli and Ling, 2022), recipe videos (Barrett and Feng, 2021), wine promotion videos (Deng et al., 2022), slot machine videos (Hoebanx and French, 2023), drone vacation videos (Vujičić et al., 2022), TV shows (Warton and Brander, 2017), marketing campaign videos (Yoo et al., 2024b), news reports (Zhang, 2021), documentaries (Wen et al., 2023), digital nomad videos (Yayla et al., 2024), videos featuring humanlike robots (Yu, 2020), museum promotional videos (Lau et al., 2024) and pro-environmental videos (Zhu et al., 2025).

In our analysis of data types, textual data were most prevalent, examined in 33 studies (89%), followed by visual data in 23 studies (62%). Metadata was analyzed in 22 studies (59%), while auditory data appeared in seven studies (19%). The analysis of data-type combinations revealed that four studies (11%) examined all four data types, eight studies (22%) analyzed three types, 20 studies (54%) focused on two types and five studies (14%) examined a single data type.

Regarding analytical scope, we identified three distinct approaches: 21 studies (57%) concentrated exclusively on primary video content analysis, four studies (11%) focused solely on user-generated comments and 12 studies (32%) adopted an integrated approach examining both video content and audience responses through comments.

YouTube served as the predominant data source, with 25 out of 37 studies (68%) relying on it exclusively or primarily. Alternative platforms were utilized less frequently: Bilibili, a popular video-sharing platform in China, appeared in four studies (11%), while Douyin/TikTok appeared in three studies (8%). The remaining studies drew from diverse sources: Facebook (Tavakoli and Ling, 2022), television show episodes (Warton and Brander, 2017), travel crowdfunding projects (Chen et al., 2024) and government project videos (Bernal et al., 2024). Additionally, one study did not specify its data source (Yıldırım and Kaya, 2024).

The scale of video samples varied across studies, ranging from a focused examination of two videos (Yu, 2020) to a comprehensive analysis of 5,008 videos (Yoo et al., 2024a). Similarly, studies examining associated comments exhibited diverse sample sizes, spanning from 186 comments (Hoebanx and French, 2023) to an extensive collection of 133,680 comments (Ma et al., 2024).

Qualitative methodological approaches were most frequently employed in the reviewed studies, appearing in 17 out of 37 studies (46%). Mixed methods represented the second most common approach, utilized in 14 studies (38%), while quantitative methods were applied in six studies (16%).

Qualitative content analysis emerged as the dominant analytical method, appearing in 20 studies (54%). Statistical analysis represented the second most common analytical approach, appearing in 11 studies (30%). These studies (e.g. Sharma, 2025; Vujičić et al., 2022; Zhu et al., 2025) employed various statistical techniques ranging from descriptive statistics to more complex inferential methods, including both parametric tests (e.g. t-tests, ANOVA, chi-square) and nonparametric alternatives (e.g. Mann–Whitney U tests, Kruskal–Wallis H). Thematic analysis, which enables systematic identification and analysis of patterns within qualitative data (Braun and Clarke, 2006; Nowell et al., 2017), ranked as the third most common method, utilized in 10 studies (27%) (e.g. Hoebanx and French, 2023; Nazir, 2023; Wen et al., 2023).

Our review also identified an emerging trend toward computational and machine learning-based analytical techniques. Computational sentiment analysis, which quantifies emotional content in text (Liu, 2012), was applied in five studies (14%) (e.g. Ma et al., 2024, 2025; Rauf and Pasha, 2024), while Latent Dirichlet Allocation topic modeling, a statistical approach for identifying latent thematic structures within text corpora (Blei, 2012), was employed in 3 studies (8%) (Deng et al., 2022; Ma et al., 2025; Zhang, 2021). Additionally, several studies implemented specialized analytical approaches designed for specific research objectives, including acoustic analytics for voice and music characteristics analysis (Barnes, 2024) and spatial analysis using landmark recognition (Li et al., 2024).

Of the 37 studies reviewed, 27 (73%) explicitly specified the tools or software used in their research methodologies. Python emerged as the most frequently employed programming language, utilized in 10 studies (27%) through various specialized libraries. NVivo, a specialized qualitative data analysis software, was the second most frequently used tool, appearing in six studies (16%). Statistical Package for the Social Sciences (SPSS), a comprehensive statistical software package, ranked third, used in five studies (14%), followed by R, an open-source programming language for statistical computing and graphics, employed in three studies (8%), and Data Miner, a data extraction tool, also employed in three studies (8%).

The review also identified several specialized analytical tools deployed for specific research purposes. These included acoustic analysis software, such as Soundgen (Barnes, 2024), social media analytics tools like FanpageKarma (Huertas et al., 2017) and extraction software including QDA Miner Lite (Mesana et al., 2024) and Webometric Analyst (Vujičić et al., 2022). Other specialized tools included SentiStrength for sentiment analysis (Zhang, 2021), Microsoft Azure AI Video Indexer for scene detection (Zhu et al., 2025), Ucinet 6 and NetDraw for network analysis (Yoo et al., 2024b) and Baidu AI for geospatial recognition (Li et al., 2024).

This scoping review analyzed methodological approaches in video analysis within hospitality research, examining 37 studies published between 2017 and 2025. Our investigation revealed several distinct patterns and challenges, characterizing this dynamic field.

The distribution of data types analyzed across studies reflected the multifaceted nature of video content, with textual elements being examined most frequently, followed by visual components, metadata and auditory features. When examining the sources of this multimodal content, YouTube dominated as the data source platform, with alternative platforms like Bilibili, TikTok and Facebook having more limited representation. In terms of methodological approaches, the analytical methods showed a preference for qualitative methods and mixed methods, with qualitative content analysis being the most widely applied technique, followed by statistical analysis and thematic analysis. The tools employed across studies revealed the multidisciplinary nature of video analysis, with programming languages like Python and R used alongside specialized software such as Data Miner and NVivo. These patterns are examined in detail below.

Textual data appeared most frequently in the reviewed studies, with visual data following closely behind. Metadata analysis appeared in more than half of the studies, while auditory data remained notably underutilized. This distribution highlighted both the multimodal character of video content and specific methodological gaps, particularly the limited integration of auditory elements such as voice tone, music and ambient sounds that could provide valuable contextual information about hospitality experiences (e.g. Kemp et al., 2019; Liu et al., 2024).

The results also demonstrated a clear preference for multidimensional analytical frameworks, with most studies analyzing two or more data types. Only a small percentage of studies limited their analysis to a single data type, suggesting recognition among researchers that comprehensive video analysis requires examination of multiple elements. This methodological choice facilitated a more holistic view of video content, as illustrated by Salangsang et al. (2022), who integrated various elements to provide a richer understanding of travel promotion videos.

The prevalence of YouTube as the primary data source across the reviewed studies highlighted its central role in hospitality research. This platform’s prominence can be attributed to its extensive global reach (Statista, 2024) and researcher-friendly accessibility (YouTube, 2025). Additionally, the platform hosts diverse content types, from amateur vlogs to professional marketing videos, allowing researchers to examine various aspects of the hospitality industry through a single data source (Arthurs et al., 2018).

However, this heavy reliance on YouTube presented certain limitations. The platform’s content moderation policies and algorithmic recommendations may create an environment that does not fully represent the broader landscape of hospitality-related videos (Rieder et al., 2018). For example, prior research found YouTube’s algorithmic preference for mainstream professional sources over independent creators in news content (Nechushtai et al., 2024). This pattern suggests that hospitality researchers searching for industry-related news or destination updates may similarly encounter algorithmically curated results dominated by established media outlets, potentially overlooking authentic local perspectives and small business voices that remain buried in search results. Additionally, while YouTube has a diverse user base, it may not capture the full spectrum of demographics or cultural contexts relevant to global hospitality research, potentially leading to underrepresentation of certain perspectives (Arthurs et al., 2018). For instance, YouTube’s inaccessibility in China (Reuters, 2009) significantly reduces content creation from Chinese nationals, who are less likely to share their travel experiences and hospitality perspectives on a platform they cannot easily access, thereby underrepresenting voices from the world’s largest outbound tourism market (UN Tourism, 2025).

The emergence of alternative platforms like Bilibili and TikTok in several reviewed studies (e.g. Li et al., 2024; Zhu et al., 2022) indicated a growing recognition of the need for more diverse data sources. These platforms provide access to different geographical regions, user demographics and content formats, potentially enriching the field’s understanding of video content in hospitality. For instance, TikTok’s short-form videos represent a rapidly growing trend in content consumption with potential implications for hospitality marketing (Tham et al., 2024). Similarly, Bilibili’s popularity in China offers researchers opportunities for conducting cross-cultural comparisons and region-specific trend analysis (Li et al., 2024). However, incorporating these alternative platforms introduces new challenges that researchers must address, such as language barriers, cultural nuances and platform-specific features that may require developing new analytical approaches.

Among the 37 studies reviewed, qualitative approaches were most frequently employed. Nearly half of the reviewed studies employed exclusively qualitative methods, while mixed methods studies similarly relied heavily on qualitative frameworks, typically supplementing them with only basic quantitative components such as descriptive statistics. Within these qualitative methodological frameworks, qualitative content analysis emerged as the most frequently employed analytical technique. The field’s concentrated reliance on this specific method, however, appears to have created several methodological limitations. The widespread adoption of qualitative content analysis has occurred without establishing standardized protocols specifically designed for video data analysis, creating potential difficulties for methodological consistency and cross-study comparison (Elo et al., 2014). Additionally, the resource-intensive nature of manual coding may become increasingly problematic as video content volume expands exponentially, raising questions about the scalability and representativeness of current analytical practices. Furthermore, the method’s inherent vulnerability to subjective interpretation remains an ongoing challenge for researchers seeking analytical rigor (Schreier, 2012).

The recent emergence of quantitative-only studies in the field may represent a response to these methodological limitations. This shift toward advanced computational approaches suggests that researchers are beginning to explore alternative frameworks that could address the scalability and standardization challenges inherent in traditional qualitative methods. Developing these computational capabilities could benefit from engaging with computer vision research, which has established sophisticated techniques (e.g. Chen et al., 2018; Redmon et al., 2016) for automated scene recognition and object detection that could be adapted for hospitality video analysis. However, rather than completely abandoning qualitative insights, the field may benefit most from developing integrative approaches that combine computational efficiency with the contextual sensitivity that characterizes hospitality research. Such hybrid methods could leverage machine learning’s capacity for pattern recognition across large datasets while preserving interpretive depth through targeted qualitative analysis (Ma et al., 2024, 2025). For instance, researchers examining food vlogs could use computational sentiment analysis to process thousands of viewer comments identifying overall emotional responses to specific cuisines or dining experiences, while employing qualitative analysis to understand the personal narratives that explain why certain foods evoke particular emotional reactions. Similarly, studies of destination marketing in travel videos could apply computer vision algorithms to automatically identify and categorize visual elements across hundreds of videos, such as natural landscapes, urban scenes or cultural activities, and then use qualitative interpretation to analyze how these visual patterns construct destination identities and influence tourist expectations.

The review revealed diverse analytics tools employed in hospitality video analysis, with Python, NVivo and SPSS emerging as the most commonly used. These tools presented researchers with specific advantages and implementation challenges. Python, the most frequently utilized programming language across the reviewed research, offers flexibility through its specialized libraries but necessitates coding skills that often fall outside the expertise of hospitality researchers (Guttag, 2016). NVivo, a qualitative data analysis software used in multiple studies, provides comprehensive functionality but involves licensing fees and requires training for effective use (Dhakal, 2022). SPSS, a comprehensive statistical software package, provides a user-friendly interface but has limitations in handling unstructured video data and involves licensing costs that may constrain researchers with limited budgets (Pallant, 2020).

These technical barriers point to a notable challenge in hospitality video analysis: the gap between available analytical capabilities and researcher expertise. Addressing this gap may require both educational and technological approaches. From an educational perspective, hospitality programs might consider integrating coding and analytics training into their curricula to broaden researchers’ technical skills and facilitate adoption of more sophisticated methodological approaches.

From a technological perspective, advances in analytical tools themselves may help address these barriers. AI-assisted software that integrates natural language processing and computer vision algorithms might automate routine aspects of video coding while preserving opportunities for nuanced human interpretation. Such developments could democratize access to advanced analytical methods by reducing both technical prerequisites and associated costs. Additionally, collaborative, cloud-based platforms would facilitate larger-scale video analysis projects, allowing researchers to pool both computational resources and methodological expertise across institutions.

This review advances methodological knowledge in hospitality research by providing the first systematic documentation of video analysis practices across the field. While previous work examined automatic analytics in tourism (Zhu and Cheng, 2024), our comprehensive mapping reveals a critical pattern: qualitative content analysis remains the dominant analytical technique, while computational and machine learning-based analytical approaches are emerging but remain underutilized. This gap between available capabilities and current practices contributes to understanding methodological adoption patterns in emerging research areas. By systematically documenting patterns in data-type utilization, platform concentration, analytical approaches and tool selection, this review provides an empirical foundation for methodological advancement in hospitality video analysis. Our findings enable researchers to situate their methodological choices within the broader landscape, identify underexplored approaches and make informed decisions about methodological innovation.

Based on the patterns observed in current methodological practices, we identify three strategic priorities for advancing video analysis in hospitality research: (1) platform diversification represents a critical need, given YouTube’s dominance in 68% of reviewed studies. Researchers should examine alternative platforms like Bilibili and TikTok to determine whether current findings are platform-specific or generalizable across different video environments. For example, food vlog engagement metrics on YouTube’s long-form content may significantly differ from TikTok’s algorithm-driven short videos, potentially affecting how hospitality brands should approach different platforms. (2) Investigating underutilized auditory components could enhance analytical depth, as our analysis found that auditory data appeared in only 19% of studies. Voice tone, background music and ambient sounds could provide rich contextual information about hospitality experiences that visual and textual analysis alone cannot capture. Researchers could integrate these components by analyzing vocal characteristics to assess emotional responses in customer review videos, examining background music choices in destination promotional videos or investigating ambient environmental sounds in travel vlogs to determine authenticity cues. For instance, studies examining restaurant review videos could investigate whether background ambient noise influences perceived authenticity beyond what visual food presentation alone conveys. (3) Hybrid analytical approaches offer solutions to current scalability limitations by integrating automated computational techniques with manual qualitative interpretation on the same dataset. These approaches use computational methods to achieve breadth of coverage across large datasets while preserving qualitative depth for interpretive insights. This methodological integration differs from multimodal analysis, which examines multiple data types (e.g. visual, auditory, textual) within videos; hybrid approaches instead focus on combining different analytical methods to address both scalability and interpretive depth. For example, researchers analyzing travel vlogs could employ automated scene detection and visual recognition to identify patterns in destination imagery across hundreds of videos and then use qualitative narrative analysis on selected examples to examine how vloggers construct destination identities through specific storytelling techniques and visual sequencing. Figure 3 summarizes these current practices, observed methodological patterns and strategic priorities for advancing video analysis in hospitality research.

This scoping review had several limitations that warrant consideration when interpreting findings. First, restricting our search to English-language publications excluded substantial relevant research published in other languages, particularly given the global nature of video content and platform usage. This limitation became especially apparent when considering regions like China, where extensive video analysis research likely existed in Chinese-language publications but remained inaccessible through English-focused academic searches. Future reviews could incorporate multilingual literature searches and foster international collaboration to capture diverse methodological approaches and cultural perspectives.

Second, the rapid evolution of video production and analysis technologies created inherent challenges for academic literature to reflect current industry developments. The time lag between research conduct and publication means that recent advancements in video analytics might not be fully reflected in scholarly literature. Our exclusive focus on peer-reviewed publications further limited access to innovative developments typically reported first in industry publications, conference proceedings and technical white papers (e.g. AXIS Communications, 2021; Pensees Singapore, 2023).

Third, our database selection strategy created potential coverage gaps. While Web of Science and Scopus are widely used in hospitality research (e.g. Gomezelj, 2016; Mariani et al., 2018; Mehraliyev et al., 2021), they may not encompass all relevant publications, particularly those focusing on technological innovation (Gomezelj, 2016). The interdisciplinary character of video content analysis means that relevant studies could appear in computer science, media studies or engineering journals that fall outside traditional hospitality research databases. Future reviews should expand database coverage (e.g. Science Citation Index Expanded, Social Sciences Citation Index) and disciplinary scope to achieve more comprehensive literature identification.

Fourth, we did not calculate inter-rater reliability measures during the screening process. While not required by PRISMA-ScR guidelines (Tricco et al., 2018), such measures would enhance methodological transparency and rigor in future scoping reviews.