Introduction

Heritage Tourism Destinations (HTDs), characterized by their concentration of cultural heritage resources, provide immersive experiences that convey historical narratives and cultural identity1,2. Among these, urban historic districts, preserving early urban layouts, architectural styles, and social structures, serve as crucial components of HTDs, embodying deep historical memories and cultural narratives3,4,5,6. As typical representations of “living heritage,” urban historic districts are vital carriers for the integrated development of culture and tourism7,8. On one hand, they feature unique street scales, vibrant social dynamics, and a strong sense of place identity, offering spatial experiences that contrast sharply with the functional homogeneity of modern urban environments, thus shaping a distinctive Destination Image (DI)9,10. On the other hand, their open accessibility, everyday use, and mixed-function design allow tourists to deeply immerse themselves in the cultural context11. Through the development of cultural tourism, these historical heritage can be utilized during conservation and inheritance, strengthening cultural dissemination while providing sustainable income to support heritage preservation, thereby contributing to the sustainability of urban cultural heritage conservation7,8. However, with the rise of social media, this process of transmitting and protecting heritage value is facing profound challenges and transformations.

Destination Image (DI), defined as an individual’s overall impression of a tourist destination formed through the integration of perception, cognition, and affect12,13, is key to influencing tourist behavior, and a unique destination image constitutes a core advantage for heritage tourism1,2,6. In the digital era, DI has bifurcated into two distinct forms: the On-site Perception Image (OPI), derived from tourists’ direct physical experiences, and the Online Destination Image (ODI), predominantly shaped by User-Generated Content (UGC) circulating online14,15,16,17,18. When a historic district rapidly gains popularity (“goes viral”) on social media—a process defined by the large-volume, high-speed dissemination of information through social networks19,20—due to its unique cultural symbols, its ODI often becomes highly idealized and symbolized. While this attracts substantial traffic, it also risks creating excessively high pre-trip expectations among tourists21,22,23,24. This cognitive gap between online portrayal and offline reality, known as the Destination Image Gap (DIG)25,26,27,28, has become an increasingly complex challenge in managing heritage tourism in the “influencer” era. Adapted from the Service Quality Model in marketing research29, DIG describes the discrepancy between pre-trip expectations and actual on-site experiences30, primarily reflecting the difference between ODI and OPI in the context of digital destination marketing. Indeed, combining UGC analysis to capture ODI with on-site surveys to measure OPI has become an established methodological approach for investigating these gaps31. This disparity is particularly pronounced in historic districts where authentic cultural value (reliant on immersive experience and slow-paced exploration) is often obscured by fragmented and hyper-curated digital representations32,33,34. This not only threatens tourist satisfaction but may also erode the authenticity and cultural ecology of the heritage site.

Consequently, for the sustainability of heritage tourism in the digital era, coordinating the management of ODI and OPI has emerged as a new challenge against the backdrop of the digital transformation of destination marketing. The significance of the Destination Image Gap (DIG) lies in its direct impact on tourist satisfaction and revisit intentions35,36. Scholars predominantly employ Expectation Disconfirmation Theory (EDT) for theoretical explanation, positing that when on-site perception (OPI) fails to meet expectations formed online (ODI), the resulting negative gap triggers cognitive dissonance, leading to dissatisfaction, negative word-of-mouth, and reduced revisit intentions35,37,38,39,40. Furthermore, research has confirmed a significant “asymmetry” in this effect: the detrimental impact of negative gaps (disappointment) on satisfaction far outweighs the positive effect of positive gaps (pleasant surprises)35,41. This finding reveals a hidden practical problem for “viral” historic districts: if management is inadequate, the traffic generated by heavy reliance on online marketing might be merely ephemeral (“a flash in the pan”). Once tourists’ idealized “filter” shatters, negative feedback can rapidly backlash, undermining the long-term sustainable development of the destination42,43.

Therefore, managers of historic districts must not only focus on traditional OPI aspects (such as the facilities and environment of the on-site tourism experience) but also manage ODI and DIG, adding complexity to governance. Academia has extensively explored the relationship between DIG and satisfaction36,42,43, recognized the heterogeneity in the impact of different tourism management elements (e.g., accommodation, dining, cultural activities) on satisfaction40,41, and explored and validated relevant patterns across various types of tourist destinations41,44,45. These valuable contributions successfully provide managers with a clear diagnosis of what the problems are (e.g., “negative gaps exist in environmental quality”). The challenge, however, is that not all perception gaps are created equal. They differ significantly in their characteristics (e.g., magnitude and direction) and, more importantly, in their weighted and asymmetric impact on overall satisfaction. However, this diagnosis naturally leads to a subsequent, highly practical management challenge that remains less explored: Considering the limited finite resources and multiple identified perception gaps, which problem should a manager fix first, and what is the appropriate strategy to manage that gap? (e.g., should they focus on making online promotions more authentic, or strive to enhance the actual offline tourism experience?).

To address the critical management questions of what to prioritize and which strategy to apply, this study’s primary contribution is methodological, introducing a novel Integrated Diagnostic and Prioritization (IDP) Framework. This central innovation is a dual-logic diagnostic tool designed to translate complex perceptual data into clear strategic direction. This framework’s first component is the Satisfaction-Gap Strategy (SGS) Matrix, a categorical tool that cross-analyzes on-site satisfaction (performance) against the perception net gap (disconfirmation) to classify all management elements into distinct strategic quadrants. Complementing this classification, the Optimization Index for Variables (OIV) is an innovative grading tool that systematically prioritizes the items by integrating three distinct metrics: (a) the normalized perception net gap, (b) the normalized on-site satisfaction, and (c) the asymmetric impact weights. Theoretically, this study contributes by empirically testing and validating the asymmetric impact of perception gaps on satisfaction within the highly relevant and contemporary context of a ‘viral’ heritage destination. By focusing on a classic, high-traffic site experiencing intense social media attention, this research supplements existing destination image theory, confirming its applicability and exploring the heterogeneous nature of these impacts across different management dimensions in such settings. Practically, this research provides a significant practical contribution by demonstrating how the IDP framework enables precision management. The framework offers managers a tangible and replicable tool that moves beyond a generic list of problems to a prioritized action plan. By combining the strategic guidance of the SGS Matrix (providing the means and strategy) with the data-driven ranking of the OIV Index (providing the priority), managers can formulate more precise, resource-efficient, and effective interventions.

The research follows a four-stage analytical process: (1) ODI Analysis: Using NLP to extract dimensional constructs of ODI from UGC; (2) DIG Survey and Assessment: Measuring perception gaps by aligning ODI-derived themes with field-surveyed OPI; (3) PLS-SEM Modeling: Applying PLS-SEM to examine the relationship between DIG and tourist satisfaction, assessing dimensional heterogeneity and asymmetric effects; (4) IDP Framework Application: Employing the Integrated Diagnostic and Prioritization (IDP) Framework to identify intervention priorities.

Methods

Study area

Harbin, the capital of Heilongjiang Province, serves as the research area for this study (Fig. 1). It is recognized both as a National Famous Historic and Cultural City and as a UNESCO “City of Music.” Since late 2023, Harbin has attracted significant digital attention due to its distinctive European-style architectural heritage and rich ice-and-snow cultural resources. These features have contributed to the city’s emergence as a major tourism hub, further amplified by the 2025 Asian Winter Games. During the 2024 Spring Festival, Harbin received 10.09 million visitors (81.7% year-on-year increase) with tourism revenue reaching 16.42 billion RMB (235.4% year-on-year growth)46.

Fig. 1: Study area.
figure 1

(Map data: © Mapbox; Ministry of Natural Resources of China, GS (2023) 2797. Photographs: Authors).

Full size image

This study specifically focuses on the Zhongyang Street Historic District—Harbin’s most iconic cultural tourism precinct (Fig. 1). Anchored by the pedestrianized Zhongyang Street, the district links several major urban landmarks, including Saint Sophia Cathedral, Sophia Square, the Flood-Control Monument, and the Songhua River. Designated as one of China’s first “Famous Historic and Cultural Streets” in 2009, the district holds the national heritage significance through its unique fusion of Eastern and Western architectural traditions47.

Since 2023, both Harbin and Zhongyang Street have garnered widespread attention on digital platforms, making them especially relevant for investigating perception-experience gaps. During the 2024 Spring Festival, posts mentioning “Harbin” on Rednote surged by 254%, Douyin (TikTok) accumulated over 1.26 million new videos48. On Meituan, searches for “Zhongyang Street” increased eightfold, while user reviews and travel guides rose ninefold and tenfold, respectively49. These metrics underscore Zhongyang Street’s status as one of China’s most-viewed historic districts, and highlight the growing role of social media in shaping the destination image. This digital prominence makes the site an ideal context for examining the disconnect between online perceptions and on-site experiences.

Research framework

This study develops a four-stage research framework (Fig. 2) to address the previously proposed research questions.

Fig. 2
figure 2

Research framework.

Full size image

Stage 1: ODI analysis

To objectively decode the ODI from the vast and unstructured corpus of UGC collected, this study employed a Natural Language Processing (NLP) pipeline. This computational approach was chosen over traditional manual content analysis, which can be subjective and labor-intensive. By contrast, NLP allows for a data-driven, bottom-up identification of thematic structures and affective orientations directly from the tourists’ own language, minimizing researcher-imposed biases and ensuring the extracted ODI dimensions authentically reflect the collective online perception essential for the subsequent gap analysis50,51. Indeed, NLP techniques are widely recognized and increasingly utilized in tourism research for analyzing large volumes of online text data and understanding destination image formation from UGC52,53,54.

The implemented pipeline consisted of three core, sequential components executed using Python. First, keyword extraction and vectorization were performed utilizing the Term Frequency-Inverse Document Frequency (TF-IDF) algorithm to transform the textual data into numerical vectors and identify salient terms within the processed UGC corpus. The specific configuration of the “TfidfVectorizer” (from Python’s scikit-learn library) is detailed in Table 1. These settings aim to focus on terms that are both reasonably frequent and discriminative across the corpus.

Table 1 Configuration parameters for NLP models
Full size table

Following vectorization, sentiment analysis was conducted to evaluate the emotional polarity of the texts and specific keywords using the SnowNLP library for Python. SnowNLP was specifically chosen due to its design for processing Chinese text and its proven effectiveness in analyzing sentiment polarity within Chinese UGC, leveraging its built-in model trained on relevant datasets55,56. This makes it particularly suitable for capturing the affective nuances expressed by Chinese tourists in our data compared to generic, non-language-specific tools.

Finally, topic modeling was employed using Latent Dirichlet Allocation (LDA), a widely used unsupervised machine learning technique for discovering latent semantic structures (“topics”) in text data52,53, to extract the underlying thematic dimensions of the ODI from the TF-IDF matrix. Determining the optimal number of topics (K) is crucial. We tested K values ranging from 3 to 10 and evaluated the resulting models based on coherence scores (e.g., C_v coherence) and perplexity metrics, alongside qualitative assessment of topic interpretability. Based on these evaluations, K = 5 was selected as it provided the most distinct and meaningful thematic clusters. The detailed configuration of the LDA model, implemented using the “LatentDirichletAllocation” class from Python’s scikit-learn library, is presented in Table 1. The use of default priors allows the topics to be primarily driven by the data itself57.

This systematic NLP process yielded a multidimensional ODI framework, comprising the extracted themes (topics), associated keywords, and their sentiment orientations. This data-driven ODI framework directly informed the design of the subsequent field survey questionnaire (Stage 2) and provided the baseline for the DIG analysis.

Stage 2: DIG survey and assessment

This stage focused on collecting field data to assess tourists’ OPI, satisfaction levels, and the DIG. The process involved three sequential steps. First, semi-structured interviews were conducted with visitors to gather initial qualitative insights into their perceptions of the street’s spatial characteristics, cultural atmosphere, and service facilities. Subsequently, insights derived from these interviews, alongside the literature review and the ODI dimensions extracted in Stage 1, were integrated to develop the structured questionnaire, ensuring semantic alignment between online and offline indicators (the detailed scale development process is described in previous section). Finally, this questionnaire was administered on-site to gather quantitative survey data. Responses were collected using five-point Likert scales designed to measure both tourist satisfaction with various destination elements and the perception gaps, utilizing a dual-channel scoring system to capture gap directionality (positive gap: offline > online; negative gap: offline < online), as further elaborated in previous section.

Stage 3: PLS-SEM modeling

To analyze the complex relationships between multidimensional perception gaps (both positive and negative) and tourist satisfaction, this study employed Partial Least Squares Structural Equation Modeling (PLS-SEM) using SmartPLS 4.1.0 software58,59. PLS-SEM is a well-established second-generation multivariate analysis technique widely applied in tourism and hospitality research, particularly suitable for developing theories in exploratory research and when prediction is the main objective60,61,62. The choice of PLS-SEM over traditional Covariance-Based SEM (CB-SEM) is deliberate and grounded in several key rationales pertinent to this research context. First, the primary objective of this study is predictive and explanatory—to identify the key drivers of satisfaction that possess the greatest managerial relevance for intervention. PLS-SEM, being a variance-based method focused on maximizing the explained variance of dependent constructs, is better suited for such prediction-oriented research goals compared to the theory-testing focus of CB-SEM60,61,62. Second, our proposed model exhibits considerable complexity, incorporating multiple first-order latent variables (representing dimensional positive and negative gaps) influencing dimension-specific satisfaction constructs, which in turn form a second-order latent variable for overall satisfaction. PLS-SEM demonstrates greater robustness in handling such complex models, including hierarchical constructs, without stringent sample size requirements63. Third, survey data in tourism research often deviates from multivariate normality assumptions required by CB-SEM. PLS-SEM is non-parametric and thus less sensitive to violations of these distributional assumptions, making it a more appropriate technique for analyzing our collected data58,61. Therefore, PLS-SEM is deemed the most suitable method for validating the proposed theoretical pathways and identifying practical leverage points in our specific research context.

As illustrated in Fig. 3, the measurement model maps perception gap indicators and dimension-level satisfaction indicators (both positive and negative) onto their respective latent constructs. Satisfaction at the dimension level acts as a mediating variable, contributing to overall satisfaction (SAT), which is conceptualized as a second-order construct. The structural model follows the pathway: Dimension DIG→Dimension Satisfaction→Overall Satisfaction.

Fig. 3
figure 3

Impact path modeling.

Full size image

Stage 4: IDP framework application

Building upon the previous three stages, this stage details the application of the Integrated Diagnostic and Prioritization (IDP) Framework, a dual-logic tool developed to identify and prioritize critical factors shaping the destination image of Zhongyang Street. The objective of this final stage is to translate analytical results into practical optimization strategies by integrating insights from DIG analysis, satisfaction ratings, and path coefficients derived from the PLS-SEM model. Two complementary analytical logics are employed.

First, a strategic classification is conducted using the Satisfaction-Gap Strategy (SGS) Matrix. This tool addresses the question of what type of intervention is appropriate by constructing a two-dimensional, four-quadrant matrix that cross-analyzes perception gaps and satisfaction levels. For each tourism factor i, the Perception Net Gap (\({PN}{G}_{i}\)) is calculated as the difference between the average Negative Gap (\(N{G}_{i}\)) and the average Positive Gap (\(P{G}_{i}\)):

$${PN}{G}_{i}=N{G}_{i}-P{G}_{i}$$
(1)

Satisfaction is evaluated by comparing each factor’s average satisfaction score to the overall sample mean. Based on whether a factor’s score is above or below this threshold, it is classified as having high or low satisfaction. As shown in Table 2, this yields four strategic categories.

Table 2 The satisfaction-gap strategy (SGS) matrix and management implications
Full size table

Second, to further prioritize perception factors for intervention (addressing the question of which problem to fix first), a grading logic is applied using the Optimization Index for Variables (OIV). The index is calculated based on three key indicators: Negative Gap Effect strength (NGE), Positive Gap Effect strength (PGE) and Satisfaction level (SAT). The index serves to quantify each factor’s strategic importance, enabling evidence-based prioritization in destination image optimization. The formula is as follows:

$${NG}{E}_{i}=\left|{{\rm{\beta }}}_{{NG},i}\right|\cdot N{G}_{i}$$
(2)
$${PG}{E}_{i}={{\rm{\beta }}}_{{PG},i}\cdot P{G}_{i}$$
(3)
$${PN}{{GE}}_{i}={NG}{E}_{i}-{PG}{E}_{i}$$
(4)
$${PNGE\_nor}{m}_{i}=\frac{{PN}{{GE}}_{i}-\min \left({PNGE}\right)}{\max \left({PNGE}\right)-\min \left({PNGE}\right)}$$
(5)
$${SAT\_nor}{m}_{i}=\frac{{{SAT}}_{i}-\min \left({SAT}\right)}{\max \left({SAT}\right)-\min \left({SAT}\right)}$$
(6)
$$O{{IV}}_{i}={PNGE\_nor}{m}_{i}\,-{SAT\_nor}{m}_{i}\,+\,1$$
(7)

\({\beta }_{{PG},i}\), \({\beta }_{{NG},i}\) are the path coefficients corresponding to the influence of Positive Gap (PG) and Negative Gap (NG) of the specific item i on the dimensional satisfaction; \({SA}{T}_{i}\) is the mean satisfaction score.

Survey design

Perceptions of destination image are highly context-dependent, with influencing factors varying across environmental and cultural settings44. To address this complexity, this study adopted a sequential mixed-method approach comprising qualitative scale development followed by quantitative validation (Fig. 4).

Fig. 4
figure 4

Survey design.

Full size image

Qualitative questionnaire development was initiated by integrating three evidence sources to create the destination image perception scale: (a) ODI dimensions derived from UGC analysis; (b) systematic literature review of heritage tourism studies; (c) semi-structured interviews with 26 tourists (gender-balanced, aged 18-50), recruited via random sampling at Zhongyang Street. Interviews followed a structured protocol, adapted from previous studies64,65, and explored perceptions of spatial environments, cultural characteristics, and service facilities. This process yielded 20 preliminary measurement items, which were then refined through three rounds of Delphi review conducted by a panel of three professors. The final destination image perception scale retained 13 items across three dimensions (Table 3).

Table 3 Items for the Destination Image Scale and its resources
Full size table

Following scale development, quantitative data collection was conducted using a structured questionnaire based on the refined scale administered through on-site surveys. To ensure sample validity, strict screening criteria were applied: respondents had to be first-time visitors to Harbin, and their primary information sources had to be social media or online platforms. The questionnaire comprised three components of demographic profile, satisfaction assessment and perception gap measurement. Perception gaps were measured using a novel dual-channel scoring system designed to capture gap directionality through mutually exclusive scoring: (a) Positive Gap (PG: 1–4) was recorded when the on-site experience exceeded online expectations (NG = 0); (b) Negative Gap (NG: 1–4) was recorded when the experience fell below expectations (PG = 0); (c) PG = 0 and NG = 0 indicated expectation-experience alignment. This design enabled the precise quantification of directional perception gaps, supporting more nuanced analysis of expectation–experience mismatches.

Data collection

UGC related to Zhongyang Street was collected from two distinct, highly influential Chinese online platforms: Trip.com Group (Ctrip), a leading online travel agency representing specialized travel review websites, and RedNote (Xiaohongshu), a major lifestyle-sharing social media platform known for significantly influencing trends, including “viral” tourism phenomena among younger demographics66. This dual-platform approach aims to capture both dedicated travel feedback and broader social buzz surrounding the destination.

The web scraping process targeted posts and reviews published between January 2023 and February 2025 containing the specific keywords “Zhongyang Street” or “Harbin Zhongyang Street”. To focus on content with higher visibility and potential influence, an initial filtering was applied: only posts with more than 50 likes on RedNote and reviews with a rating of 4 stars or higher on Trip.com were retrieved. This yielded an initial dataset of 212 posts from RedNote and 435 reviews from Trip.com.

Following collection, the raw text data underwent a rigorous pre-processing phase. This included removing irrelevant content (e.g., advertisements, non-textual elements) and standardizing text format. Crucially, a de-duplication mechanism was applied to eliminate identical or highly similar entries. Specifically, we calculated the cosine similarity between text vectors (e.g., TF-IDF) of the collected content and removed duplicates exceeding a high similarity threshold (e.g., 0.95), ensuring the uniqueness of each analyzed text. After cleaning and de-duplication, the final dataset used for analysis comprised 176 posts from RedNote and 428 reviews from Trip.com, totaling approximately 176,200 Chinese characters.

It is acknowledged that UGC data inherently possesses limitations regarding sample representativeness, potentially reflecting the views of more digitally active or specific tourist segments. However, the selection of Trip.com and RedNote aims to mitigate this by covering two different but dominant channels for information acquisition and sharing. Trip.com captures perspectives from users actively seeking and evaluating travel-specific information on a dedicated platform, while RedNote reflects image formation within a broader social and lifestyle context, particularly potent in shaping trends and pre-trip expectations related to “checking-in” (daka) at popular sites. As both platforms are leading players in their respective domains within the Chinese market, the collected UGC is considered highly relevant for capturing the prevailing ODI from diverse online sources pertinent to this study’s objectives.

The questionnaire survey was conducted between February 13 and February 20, 2025. A total of 300 questionnaires were distributed, resulting in 220 completed responses, yielding a response rate of 73.3%. After data cleaning—which involved excluding responses with uniform answers, missing data, or completion times deemed too short for validity—a total of 208 valid questionnaires were retained, corresponding to an effective response rate of 94.5%. Table 4 summarizes the demographic characteristics of the respondents. The gender distribution was relatively balanced, with female respondents accounting for 55.3% of the sample. The majority of participants were young to middle-aged adults (18–35 years old), comprising 77.9% of the total. A similarly high proportion (87.0%) had attained higher education. In terms of geographic origin, most respondents came from Northeast China (59.1%), followed by North China (7.7%). Regarding travel characteristics, 85.1% of participants reported a travel budget below 4000 RMB, and 58.6% indicated a trip duration of no more than 7 days. Taken together, the sample exhibited a balanced and representative distribution, supporting the validity of subsequent quantitative analysis.

Table 4 The demographic characteristics of the participants
Full size table

Results

Results of ODI analysis

Following the standardized NLP workflow detailed in previous section, the analysis of the collected UGC revealed the core thematic structure of Zhongyang Street’s ODI. Latent Dirichlet Allocation (LDA) topic modeling extracted five distinct thematic categories: (1) Russian-style Cultural Experience, (2) Architectural Landmarks and Heritage, (3) Visitor Leisure Activities, (4) Seasonal Atmosphere and Environment, and (5) Old Street Consumption Experience. A subsequent combined analysis of keyword frequency (derived from TF-IDF) and sentiment scores (calculated via SnowNLP) was conducted to assess each theme’s salience and affective orientation within the online discourse, as visualized in Fig. 5.

Fig. 5
figure 5

ODI analysis results.

Full size image

The most dominant theme was “Russian-style Cultural Experience”, characterized by the highest keyword frequency and the most favorable sentiment values. Representative keywords included “architecture” (813 occurrences), “pedestrian street” (470), “Russia” (344), and “charming style” (293), all of which are closely associated with the street’s Russian heritage. Notably, sentiment scores for terms such as “Russian style” (0.94), “charming style” (0.95), and “Russia” (0.93) consistently exceeded 0.92, indicating strong emotional resonance with the district’s historical and cultural ambiance.

The second theme, “Architectural Landmarks and Heritage” reflected visitors’ aesthetic engagement with both Western architectural styles and Chinese historical monuments. Frequent terms included “style” (299), “European-style” (236), “Renaissance” (105), “Flood Control Monument” (194), and “Baroque” (191). Sentiment scores generally surpassed 0.91, indicating high levels of aesthetic appreciation.

The third theme, “Visitor Leisure Activities” focused on casual tourism and social interaction behaviors. Keywords such as “daka” (289), “photo-taking” (177), and “delicious” (160) highlighted Zhongyang Street’s role as a social media hotspot. However, this theme exhibited the lowest average sentiment score (approximately 0.83), with negative reviews citing overcrowding and mediocre service quality as common concerns.

The fourth theme, “Seasonal Atmosphere and Environment” captured perceptions related to winter festivities and climatic aesthetics. Common terms included “ice sculptures” (102), “winter” (116), and “lively” (134). The average sentiment score reached 0.86, indicating generally positive evaluations of the seasonal atmosphere.

The fifth theme, “Old Street Consumption Experience” encapsulated perceptions of commercial activities intertwined with historical authenticity. Key terms included “century-old” (225), “old street” (128), “culture” (114), “smoked sausage” (96), and “cost-effective” (110). Crucially, “cost-effective” not only appeared frequently but also registered the highest sentiment score (0.99), reflecting strong approval of affordable yet culturally immersive consumption experiences.

In summary, the themes “Russian-style Cultural Experience” and “Old Street Consumption Experience” emerged as the most prominent and positively perceived, reinforcing Zhongyang Street’s dual identity as both a cultural landmark and a commercial destination. In contrast, “Visitor Leisure Activities” received more mixed affective responses, highlighting potential shortcomings in crowd management and service delivery that merit managerial attention.

Results of DIG survey and assessment

Tourist satisfaction surveys revealed overall positive evaluations of Zhongyang Street, with all 13 indicators averaging above 3.9. Among the three dimensions, Tourist Attractions (TA) scored the highest, Tourist Facilities (TF) showed moderate consistency, while Tourist Environment (TE) underperformed (Fig. 6).

Fig. 6
figure 6

DIG and satisfaction survey results.

Full size image

Within the TA dimension, TA2 “distinctive architectural styles” (4.41) and TA4 “symbolic historical landmarks” (4.25) emerged as top performers, highlighting the street’s strengths in architectural aesthetics and historical identity. The TF dimension exhibited internal variation: although TF3 “shopping experience” received the highest rating (4.07), TF2 “public service facilities” scored lowest (3.95), suggesting that while the commercial atmosphere is well-received, basic amenities (e.g., restrooms and seating) require improvement. The TE dimension consistently underperformed, with TE1 “street and pedestrian environment” recording the lowest score (3.91), and TE3 “seasonal landscapes and events” also falling below average (4.03). Although TE5 “wayfinding system and integrated services” received a relatively favorable rating (4.17), most indicators within this dimension reflected dissatisfaction with environmental conditions.

DIG analysis revealed a clear predominance of negative perception gaps, with positive gaps concentrated in a few high-performing indicators. Across all 13 items, over 90 respondents on average reported negative gaps, nearly twice as many as those reporting positive gaps, indicating a widespread perception that on-site experiences did not meet online-induced expectations. Negative gaps were particularly pronounced in the TE dimension, where TE2 “historical and cultural atmosphere”, TE3 “seasonal landscapes and events” and TE4 “resident interaction and tourism ambience” exhibited mean differences of 1.622, 1.598 and 1.596 points, respectively, suggesting major shortcomings in the actual delivery of seasonal, cultural and native experiences compared to their online portrayal. Conversely, notable positive gaps appeared in TF2 “public service facilities” (1.900), TE2 “historical and cultural atmosphere” (1.609), and TF4 “cultural and creative product design” (1.593), indicating that certain cultural expressions exceeded expectations. Crucially, TF2: “public service facilities” achieved the largest positive gap (1.900), demonstrating that high-quality basic services can deliver unexpected satisfaction. Notably, TE2 (historical and cultural atmosphere) recorded high mean values for both the positive gap (1.609) and the negative gap (1.622), indicating a significant perceptual heterogeneity among tourists regarding the authenticity and transmission of the cultural experience.

Comparative analysis of satisfaction scores and DIG data revealed distinct structural patterns across the three dimensions. TA emerged as the core strength, characterized by consistently high satisfaction, underscoring the critical role of visual and cultural assets in shaping positive destination images. TF displayed “functional stability with internal divergence”. While overall satisfaction remained steady, the contrast between TF4’s negative gap and TF2’s positive gap revealed a tension between infrastructural shortcomings and strengths in creative experience design. TE constituted the primary area of dissatisfaction, marked by the highest frequency and intensity of negative gaps. This identifies it as the critical weak link requiring urgent improvements in environmental maintenance, cultural interpretation, and service integration.

In summary, Zhongyang Street exhibits strong appeal through its heritage attractions and cultural consumption opportunities. However, deficiencies in environmental management and service infrastructure remain major constraints. This imbalance highlights the need for holistic strategies to align digital image expectations with on-site experiences more effectively.

Results of PLS-SEM modeling

PLS-SEM modeling was applied to examine how positive and negative perception gaps affect tourist satisfaction. Positive gaps (PG) and negative gaps (NG) were modeled as three first-order latent variables each (PG_TF, PG_TA, PG_TE and NG_TF, NG_TA, NG_TE), derived from 13 observed indicators. These gap variables directly influence dimensional-specific satisfaction (SAT_TF, SAT_TA, SAT_TE), which together form the second-order construct of overall satisfaction (SAT) (Fig. 7).

Fig. 7
figure 7

PLS-SEM model results.

Full size image

The measurement model demonstrated satisfactory reliability and validity (Table 5). All outer loadings exceeded 0.70, and both Cronbach’s α and Composite Reliability (CR) values were above 0.70, indicating strong internal consistency62. Convergent validity was established with Average Variance Extracted (AVE) values exceeding 0.5067, and discriminant validity was confirmed by Heterotrait-Monotrait (HTMT) ratios below 0.9068 (Table 6). In the structural model, explanatory power varied across dimensions, SAT_TA (R² = 0.643) and SAT_TE (R² = 0.598) exhibited higher explanatory power, whereas SAT_TF (R² = 0.483) was comparatively lower. The second-order SAT construct achieved R² = 1.000 as it is fully derived from dimensional satisfactions (Table 7).

Table 5 PLS-SEM outer model results
Full size table
Table 6 PLS-SEM model HTMT ratio values
Full size table
Table 7 PLS-SEM inner model R2 and Q² results
Full size table

Path coefficient analysis confirmed the asymmetric effects between perception and negative gaps on satisfaction (Table 8). The p-value of all paths is less than 0.05, indicating a significant impact relationship between each path. Positive gaps enhanced satisfaction, with strongest impacts from PG_TF → SAT_TF (β = 0.339) and PG_TE → SAT_TE (β = 0.322), while PG_TA → SAT_TA was weaker (β = 0.199). This indicates that, in the Zhongyang Street context, pleasant surprises in tourist facilities and environment contribute more to satisfaction than visual or cultural attractions alone. Negative gaps severely reduced satisfaction, particularly NG_TA → SAT_TA (β = -0.703), followed by NG_TE → SAT_TE (β = -0.574) and NG_TF → SAT_TF (β = -0.485). This suggests that unmet expectations in attractions (e.g., aesthetics, symbolic landmarks, or historical experience) are the most damaging to satisfaction.

Table 8 PLS-SEM inner model results
Full size table

Interestingly, the impact rankings form a reverse symmetry: the stronger a positive effect in one dimension, the weaker its corresponding negative effect (e.g., TF), and vice versa. This reflects tourist’ differing psychological responses to surprises and disappointments, and the disappointments are more damaging. The coefficient magnitudes suggest that negative perception gaps have greater influence than positive ones, highlighting the asymmetric effect emphasized by Expectation Disconfirmation Theory (EDT).

Indirect path analysis (Table 9) further revealed the heterogeneity of dimensions. Among positive paths, PG_TF (0.123) and PG_TE (0.120) most significantly enhanced overall SAT, while PG_TA (0.076) trailed. Among negative paths, NG_TA (−0.267) exerted the strongest dampening effect, exceeding NG_TE (−0.214) and NG_TF (−0.176). These results reinforce a dimension-based priority: facilities and environments generate positive surprises, whereas attractions, when failing expectations, severely diminish satisfaction. Managing attraction-related gaps, especially symbolic or aesthetic ones, should thus be a strategic focus.

Table 9 PLS-SEM inner model indirect effects results
Full size table

Analysis at the factor-level (Table 10) reveals three key mechanisms. Firstly, attractions function as high-impact levers. TA2 (architectural style, β = −0.2355) and TA3 (resource integration, β = −0.2390) showed strong negative effects, reflecting high-expectation, high-disappointment dynamics. Notably, TA3’s negative impact quadrupled its positive effect (0.0667), highlighting severe aesthetic disillusionment when expectations fail. Conversely, TA4 (symbolic landmarks, β = 0.0676) reinforced its role in exceeding expectations. In this sense, attractions operate as sensitive amplifiers: they are capable of generating high value, but equally prone to deep backlash when perception gaps widen. For culturally-driven destinations, managing expectation accuracy and curating authentic, symbolically potent experiences is therefore critical.

Table 10 The indirect effect of 13 individual image perception items
Full size table

Secondly, the environment serves as a stabilizing foundation. On the positive path, TE4 (seasonal atmosphere, β = 0.1074) and TE5 (wayfinding and integrated services, β = 0.1068) both contribute strongly among all the indicators, indicating their role in enhancing baseline satisfaction. On the negative side, TE5 (wayfinding and integrated services, β = −0.1841) and TE1 (street and pedestrian environment, β = −0.1741) emerge as significant pain points, underscoring the need for systemic improvement in basic services. Unlike the attraction dimension, environmental elements set the conditions for a stable and coherent experience improvement, their surprise enhance more on satisfaction.

Finally, facilities function as elastic, low-volatility buffers. TF2 (public service facilities, β = 0.1041) exhibits the highest positive effect, suggesting it offers a “quick win” for enhancing satisfaction. Meanwhile, TF1 (accommodation and dining, β = −0.1554) shows mild negative effects, functioning as a predictable, low-volatility support dimension. Overall, facility-related indicators show more moderate, symmetric effects, suggesting they are perception buffers rather than central levers. Their influence lies in supporting or sustaining satisfaction, rather than directly shaping emotional highs or lows. This elasticity and manageability make them ideal targets for incremental upgrades in urban tourism planning.

In summary, most indicators exhibit stronger negative than positive effects, forming a pronounced asymmetric impact structure. Particularly, TA items, TE5, and TE1 are critical negative leverage points. These findings confirm that in the context of Zhongyang Street, mitigating expectation–experience discrepancies, especially in cultural and environmental dimensions, is more crucial than merely providing delight. Strategic improvements should thus prioritize closing negative gaps, especially in areas where symbolic, functional, or experiential quality is failing to meet expectations.

Results of the IDP Framework

Applying the IDP Framework revealed distinct patterns. The SGS Matrix (Fig. 8) categorized the factors, while the OIV Index (Table 11) provided a clear prioritization.

Fig. 8
figure 8

Results of the IDP Framework: SGS Matrix Diagnosis.

Full size image
Table 11 Results of the IDP Framework: OIV Index Prioritization
Full size table

Tourist Facilities (TF) exhibit suboptimal satisfaction scores yet demonstrate positive perception gaps (Fig. 8 and Table 11). All four variables reside in the “Core Experience Failure” quadrant. This indicates that although the on-site experience (OPI) was better than the low online expectations (ODI), the actual experience itself remains unsatisfactory (low satisfaction), signaling a fundamental shortage in service delivery. Future efforts should focus on dramatically enhancing the OPI (offline experience). Among them, TF1 (ranked 8th), TF3 (11th), and TF4 (6th) show relatively higher satisfaction. Conversely, TF2 (ranked 12th) combines positive gaps with extremely low satisfaction, implying severe, ongoing deficiencies in fundamental infrastructure, convenience services, and informational guidance.

Tourist Attractions (TA) demonstrate robust performance, with all elements exceeding satisfaction means. However, they are categorized as “Online Hype Risk”, forming the destination’s online perceptual foundation. TA1 (5th), TA2 (7th), and TA4 (3rd) reinforce spatial-cultural stability, but the actual experiences (OPI) slightly lag the even higher online expectations (ODI). This suggests a risk of backlash from over-promotion and calls for managing online authenticity (ODI) alongside improving narrative expression. Notably, TA3 (ranked 1st in priority) also falls into this quadrant, revealing critical gaps in attractions’ connectivity and intergration despite its high satisfaction, highlighting it as the most severe case of “Online Hype Risk”.

Tourist Environment (TE) displays acute structural weaknesses with polarized performance. High-priority variables demand urgent intervention. TE1 (ranked 2nd), with low satisfaction, falls into the “Dual Intervention Need” category. This reflects persistent issues with crowding and sanitation that are failing on-site (OPI) and being over-promised online (ODI), undermining the visitor experience. Similarly, TE3 (4th) and TE2 (9th) fall in this same high-risk quadrant, highlighting a need for enriched cultural programming. Conversely, TE5 (10th) and TE4 (13th), both within the “Online Hype Risk” group, received above-average satisfaction, yet still reveal misalignment between advertised imagery and lived experience, especially regarding humanistic and atmospheric elements.

In summary, Tourist Attractions (TA) provide a stable cultural image base but are at risk of online hype. Tourist Facilities (TF) represent a core failure of the on-site experience. Tourist Environment (TE) is the key deficiency, suffering from both experience failures and hype risks. A dual-pronged strategy combining spatial governance and cultural programming is essential for enhancing both environmental and cultural dimensions. These integrated efforts will help bridge the online-offline perception gap, promote synergy between functional and symbolic qualities, and ultimately contribute to a high-quality, immersive tourism image for historic districts like Zhongyang Street.

Discussion

This study offers actionable, evidence-based strategies derived directly from the analysis of perception gaps and their impact on tourist satisfaction at Harbin’s Zhongyang Street, providing valuable insights for managers of historic districts navigating the complexities of the digital media landscape. To provide the most actionable recommendations, this discussion is restructured to directly follow the diagnostic logic of the Integrated Diagnostic and Prioritization (IDP) Framework.

The primary value of the IDP Framework is its ability to move beyond a simple list of problems to a prioritized action plan. The OIV Index (Table 11) provides this clear prioritization, revealing what truly drives satisfaction in a heritage context. The ‘Top-Priority’ intervention points are TA3 (integration of natural and cultural eesources, OIV rank 1) and TE1 (street and pedestrian environment, OIV rank 2). Following them, the ‘High-Priority’ cluster includes TA4 (cultural memory and historical landmarks, rank 3), TE3 (seasonal landscapes and events, rank 4), TA1 (exotic street spaces, rank 5), and TF4 (cultural and creative product design, rank 6). These top six factors are not ancillary services: they collectively define the core heritage experience and “sense of place” of the historic district. They encompass the holistic landscape integration (TA3), the physical comfort of the streetscape (TE1), the symbolic landmarks (TA4), the unique seasonal atmosphere (TE3), the street’s aesthetic charm (TA1), and the heritage production design (TF4). Their high OIV ranking strongly suggests that for heritage tourism, experiential and atmospheric authenticity are the most critical drivers of tourist satisfaction.

Conversely, the OIV index places most Tourist Facilities (TF) items (rank 8, 11, and 12) in the lower-priority half, compared with the other two dimensions. This finding is significant: it does not imply these basic services are unimportant (as their “Core Experience Failure” diagnosis in the SGS matrix has shown), but it precisely illustrates a crucial distinction. For “viral” heritage sites, getting the core heritage experience (defined by TA, TE, and experiential products like TF4) right is far more critical than optimizing basic functional services.

After identifying these top six priorities, the IDP Framework’s second layer, the SGS Matrix (Table 2, Fig. 8), provides the specific “how-to” strategy for each. We focus on these top-ranked items grouped by their management dimension.

The largest high-priority group is the Tourist Attractions (TA) dimension, which includes TA3 (integration, rank 1), TA4 (landmarks, rank 3), and TA1 (street spaces, rank 5). The SGS matrix diagnoses all three as ‘Online Hype Risk’. This diagnosis is crucial because the on-site experience (OPI) for these attractions is already good (high satisfaction), but the online “hype” (ODI) is promising even more (OPI < ODI). The strategy here is not just to “maintain” but must be twofold. The first is to manage ODI authenticity to prevent backlash. This involves shifting the promotional focus from user-generated content that is often “over-filtered” or “intentionally beautified” to maximize engagement, towards promoting official, high-quality content (e.g., cultural documentaries, historical archives) that sets realistic and authentic expectations. The second priority is to elevate the OPI to truly match the hype. For the top-ranked TA3, this means addressing the “fragmented experience” by implementing curated thematic routes and using digital tools like Augmented Reality (AR) guides to deliver the real integrated experience promised online. For TA1 and TA4, this involves enhancing “cultural storytelling continuity” through tangible means, such as discreet interpretative signage, QR-code-based historical narratives, or AR overlays that bring the static architecture to life.

Next, the Tourist Environment (TE) dimension includes the other two high-priority items, TE1 (street environment, rank 2) and TE3 (seasonal events, rank 4). Both fall into the “Dual Intervention Need” quadrant. This diagnosis indicates a critical failure, as the offline experience is poor (low satisfaction) while the online hype is simultaneously “over-promising”. The required strategy is therefore an urgent and simultaneous intervention. For TE1, this necessitates immediately fixing the OPI by addressing “crowding, sanitation, and traffic conflict”. And this must be coupled with managing ODI expectations by avoiding exaggerated promotions (e.g., “most comfortable,” “most relaxing”) that are unrepresentative of the typical crowded on-site experience. Similarly, for TE3, the “seasonal spatial design” and “cultural program” must be substantially enriched to match the idealized winter-themed images promoted online, and reduce the exaggerated promotions at the same time.

Finally, the TF4 (cultural products, rank 6) item from the Tourist Facilities (TF) dimension presents a unique challenge. It is diagnosed as “Core Experience Failure”. The SGS matrix reveals that the positive gap is misleading. It merely signifies that the online expectation (ODI) was even lower than the poor offline experience (OPI). The low satisfaction score is the decisive indicator of this failure. Therefore, the strategy must be an exclusive focus on improving the OPI. The “cultural and creative product design”is failing on-site. Management must move beyond “generic souvenirs” and incentivize “artisanal shops focusing on authentic local crafts,” ensuring products genuinely reflect the unique “European-style” and “Baroque” heritage of the street, as this authenticity is precisely what the OIV index identified as being more critical to tourists than basic services.

In summary, this discussion has shown, through the OIV index, that prioritizing the core heritage experience (the top six items) yields the highest leverage over tourist satisfaction. Subsequently, the SGS matrix provides the differentiated, actionable strategies required for these priorities. The framework determines that “Online Hype Risk” (e.g., for TA3) must be managed by curating ODI authenticity first, and then elevating OPI. “Dual Intervention Need” (e.g., for TE1) requires an urgent dual OPI-ODI response. Furthermore, “Core Experience Failure” (e.g., for TF4) demands an exclusive focus on improving OPI. By using this framework to first prioritize (OIV) and then apply the correct strategy (SGS), managers can most effectively align the compelling online image with a satisfying and authentic on-site reality, moving “beyond the hype” towards long-term resilience.

To conclude, this study’s systematic investigation of the alignment between ODI and OPI at Harbin’s Zhongyang Street leads to several critical conclusions for managing “viral” heritage destinations. First, the online destination image (ODI) is a highly idealized construct. It predominantly emphasizes cultural symbolism (e.g., “russian-style cultural experience”) and atmospheric elements, which sets high, and often unrealistic, pre-trip expectations. Second, the study demonstrates a significant misalignment between this online “hype” and the on-site “experience”. Despite generally positive OPI (especially for Tourist Attractions), the analysis confirms a prevalence of negative perception gaps. These gaps are most severe not in the core attractions, but in the Tourist Environment (TE) dimension (e.g., cultural communication and wayfinding) and basic Public Service Facilities (TF2), indicating widespread mismatches in experiential and functional expectations. Third, the PLS-SEM modeling quantitatively confirms the theoretical foundation for our diagnostic framework: the impact of these gaps is both asymmetric and dimensionally heterogeneous. The finding that negative gaps in the Tourist Attractions (TA) dimension (e.g., NG_TA → SAT_TA = −0.703) are the most detrimental to satisfaction provides the critical insight that managing “disappointment” in core heritage attributes is paramount. Finally, based on these preceding conclusions, the IDP framework diagnosis confirms that the top six OIV-ranked factors (TA1, TA3, TA4, TE1, TE3, and TF4) collectively define the “core heritage experience” and are the most critical drivers of tourist satisfaction. The framework also provides the clear, differentiated strategies required for these priorities. These strategies include managing ODI authenticity for the “Online Hype Risk” found in the TA cluster, demanding an urgent OPI-ODI dual intervention for the “Dual Intervention Need” of the TE cluster, and requiring a 100% focus on improving the OPI for the “Core Experience Failure” of TF4.

This research offers several contributions to the literature on destination image, perception gaps, and heritage tourism management in the digital age. By focusing explicitly on urban historic districts undergoing viral social media attention, this study expands the research context beyond conventional settings, addressing the pressing need to understand how “hype” impacts heritage sites’ authenticity and sustainability. This research’s primary contribution is the development and application of the Integrated Diagnostic and Prioritization (IDP) Framework, a novel tool tailored for sustainable management practice within this context. This framework’s innovation is its dual-logic approach that moves beyond simple gap identification to actionable strategy. It systematically links online image (ODI) with on-site experience (OPI) and PLS-SEM modeling to strategically classify issues using the SGS Matrix and quantitatively prioritize interventions using the OIV Index. Through its application, this study also empirically demonstrates the asymmetric and dimensionally heterogeneous effects of perception gaps on tourist satisfaction in this “viral” context. These theoretical findings are crucial as they form the empirical basis for the weighted-impact calculations within our OIV Index. This highlights the distinct ways satisfaction responds to positive versus negative gaps under digital influence, thereby offering both empirical insights for Expectation Disconfirmation Theory and a practical, data-driven foundation for destination management.

While this study offers meaningful insights, several limitations highlight avenues for future research. First, the focus on a single case study, Harbin’s Zhongyang Street, limits the generalizability of findings. Comparative studies across diverse heritage contexts are needed to enhance external validity. Second, inherent limitations exist in the data sources and methodology; UGC may reflect biased segments, the cross-sectional survey captures only a snapshot, and the linear SEM model simplifies complex relationships. Future research could benefit from longitudinal or geo-tagged data, incorporate multimodal UGC analysis (images, videos), and explore non-linear dynamics or the moderating role of tourist characteristics (e.g., first-time vs. repeat visitors). Finally, extending the model to include behavioral outcomes beyond satisfaction, such as destination loyalty, revisit intention, and spending patterns, would provide a more comprehensive understanding of the long-term consequences of managing online-offline image gaps for sustainable heritage tourism development.