How Does the Platform Economy Affect Urban System: Evidence from Business-to-Business (B2B) E-Commerce Enterprises in China

Abstract

In the new paradigm where the digital economy is profoundly reshaping urban spatial organization, how the platform economy transcends traditional geographical constraints to restructure the urban system has become a strategic issue in urban geography and regional economics. This study develops an innovative measurement framework based on Business-to-Business (B2B) e-commerce enterprises to analyze platform-driven urban systems across 337 Chinese cities. Using spatial autocorrelation, rank-size distributions, and urban scaling laws, we reveal spatial differentiation patterns of cities’ B2B platforms. Combining Ordinary Least Squares (OLS) and random forest models with Partial Dependence Plots (PDP), Individual Conditional Expectations (ICE), and Locally Weighted Scatterplot Smoothing (LOWESS), we uncover non-linear mechanisms between platform development and urban attributes. Results indicate that (1) B2B platforms exhibit “superliner agglomeration” and “gradient locking”, reinforcing advantages in top-tier cities; (2) platform effects are non-linear, with Gross Domestic Product (GDP), Information Technology (IT) employment, and service sector shares showing threshold-enhanced marginal effects, while manufacturing bases display saturation effects; and (3) regional divergence exists, with eastern consumer-oriented platforms forming digital synergies, while western manufacturing platforms face path dependence. The findings highlight that platform economy evolution is shaped by a “threshold–adaptation–differentiation” mechanism rather than neutral diffusion. This study provides new insights into urban system restructuring under digital transformation.

1. Introduction

As a critical organizational form of the digital economy, platform-based development is reshaping the spatial organization of contemporary urban systems [1,2,3]. Propelled by mobile Internet, cloud computing, and big data, platform models such as e-commerce, mobility, and service platforms have expanded rapidly, restructuring upstream and downstream industrial linkages and triggering profound transformations in inter-city resource allocation, factor mobility, and functional positioning [4,5,6,7]. Understanding the mechanisms through which platform economies embed into and reconfigure urban systems has thus emerged as a central concern in urban geography and regional studies under digital transition [8]. From the perspective of urban system evolution, conventional research has largely focused on demographic, economic (GDP), and infrastructural factors in shaping the hierarchical structure of cities, emphasizing the stability of the “rank–size” rule and the “center–periphery” logic [9,10]. The ongoing advancements in transportation and communication technologies have progressively dissolved traditional spatial constraints on inter-city linkages.

Recent scholarship explores how network structures, functional connectivity, and inter-urban flows of traffic, information, and capital act as alternative mechanisms influencing urban systems, giving rise to contested notions like “de-hierarchization” and the reconfiguration of functional divisions [11,12,13,14,15,16,17,18,19]. The proliferation of platform economies, accelerated by digital and mobile Internet technologies, has once again reshaped urban systems—particularly visible in platforms like China’s Weibo, Taobao, TikTok (Douyin), and WeChat [20,21,22,23]. Existing evidence suggests that rather than flattening the traditional urban hierarchy, these platforms have reinforced the dominance and resource-attracting power of large cities, exacerbating uneven development across regions. Yet a closer examination reveals that current literature largely overemphasizes consumer-facing (C-end) platforms while neglecting B2B platforms’ structural roles in supply chain integration and cross-regional production networks. Moreover, most existing studies rely on case-based or macro-trend descriptions, lacking systematic quantitative measurement and explanatory modeling. Key questions thus remain unresolved, including how platform economies embed into urban systems, whether they reconfigure hierarchical structures, and how regional and sectoral heterogeneity shapes this process. To address these gaps, this study proposes two core hypotheses: First, do B2B e-commerce enterprises contribute to a more decentralized, multi-center equilibrium in urban systems? Challenging the traditional hierarchical urban model. Second, does the growth of B2B e-commerce follow a non-linear, threshold-dependent trajectory? Where factors such as manufacturing employment, IT employees, and GDP exhibit varying effects across regions.

Based on these hypotheses, this study builds on the traditional urban system theories, particularly Christaller’s central place theory, to explore how B2B e-commerce platforms interact with urban hierarchies. We use firm-level data from the 1688 platform—a leading B2B platform in China—to develop a platform-oriented urban metrics framework and carry out three interrelated analyses. First, we employ spatial autocorrelation, rank-size distributions, and urban scaling laws to reveal the spatial patterns and hierarchical features of B2B platforms. Second, we integrate OLS regression, random forest models, and non-linear effect identification tools (PDP, ICE, and LOWESS) to evaluate marginal effects and threshold mechanisms of urban attributes. Third, we analyze heterogeneity across regions (eastern, central, western China) and sectors (manufacturing, services, consumption, and raw materials).

By embedding B2B platform enterprise data into an urban system framework. In contrast to studies mainly based on service-oriented platforms, the paper highlights China’s unique coupling between platform economies and its manufacturing base, and proposes a new mechanism framework integrating platform dynamics, industrial foundations, regional disparity, and urban hierarchy. The findings not only contribute empirical evidence to the spatial evolution of platform economies but also offer theoretical and practical insights from China for comparative urban development and governance under digital transformation.

2. Literature Review

2.1. Urban System Theory

The concept of the urban system has long been central to urban and regional studies [15,24,25]. Early urban system theories, particularly Christaller’s central place theory (1933), have focused on hierarchical structures and city size distributions, emphasizing the central role of large cities in resource allocation, economic growth, and functional dominance. Christaller’s theory was foundational in understanding the central–peripheral relationships that shape urban systems, which attracted a large number of scholars who enriched the notion of hierarchy by measuring cities’ positions through economic indicators such as GDP, population, and industrial output [26,27,28,29]. These rankings and scales often serve as proxies for cities’ influence and policy relevance at global, national, and regional levels.

However, the rise of globalization and the digital economy necessitates a more nuanced understanding of urban systems. Traditional hierarchical urban systems, as outlined by Christaller, no longer fully capture the complexities of contemporary urban organizations. Instead, urban system studies have progressively embraced a networked perspective, particularly with the increasing influence of the platform economy. Castells’ “space of flows” framework posited that inter-city connections now depend less on geographic proximity than on functional linkages shaped by transportation, information, financial, and human flows [30,31,32,33,34,35,36,37,38,39,40,41,42]. This networked perspective is particularly relevant for understanding the role of B2B e-commerce platforms in urban system restructuring, as they discover new economic nodes in cities that were previously considered peripheral that transcend traditional hierarchical models. Fostering relational, network-based, and even post-hierarchical interpretations, the network turns in urban system theory have not displaced traditional hierarchical structures. Empirical studies demonstrate a strong coupling between nodal connectivity and urban hierarchy. Seemingly decentralized network structures often amplify the agglomeration capacity of core cities, engendering re-polarization and re-stratification phenomena [19,38,39,42,43].

Recently, renewed scholarly attention to the “static structural” dimension has gained traction, which considers that the core–periphery dynamics proposed by Christaller still play a significant role in understanding how cities interact within the platform economy. In particular, urban scaling laws attempt to explain the superliner performance of large cities in platform economies, innovation output, and consumption activity, based on systemic relationships between city size and economic indicators [44,45,46]. While such approaches may not explicitly address network relations, they reveal how structural features persistently shape urban system dynamics—affirming the enduring relevance of hierarchy and scale in contemporary urban theory.

2.2. Platform Economy and the Urban System

The rapid development of the Internet and mobile technologies has catalyzed the emergence of the platform economy, defined by its reliance on data flows, algorithms, and network effects. This new economic form simultaneously transforms traditional production–distribution–consumption chains while reorganizing urban spatial structures and hierarchical systems. At the urban scale, the agglomeration, diffusion, and evolution of platform enterprises now constitute pivotal variables in redefining economic structures and status.

Recent research has predominantly focused on consumer-oriented (C-end) platforms, such as social media and short video applications (e.g., Weibo, Douyin, WeChat). These studies, often based on user mobility and behavioral data, have led to the formulation of digital interaction-based urban systems [7,21,22,47,48,49,50]. Within this framework, cities are re-ranked according to online traffic flows, digital visibility, and platform user engagement—emphasizing a new attention-based spatial hierarchy. While this literature reveals important insights into urban digitalization and the symbolic economy, it often overlooks the role of platforms in transforming the material foundations of urban economies. In contrast, B2B (business-to-business) platforms—which connect firms across industrial, manufacturing, and service sectors—are structurally embedded in the economic core of cities. Unlike C-end platforms that primarily capture the online presence of cities, B2B platforms reflect their industrial strength and digital infrastructure capacity. As one of the three major categories of e-commerce, B2B platforms handle large transaction volumes and serve a vast number of enterprises, especially in manufacturing. Their spatial distribution mirrors the level of industrial digitalization and the potential for economic upgrading within a city [51]. Moreover, B2B services play a crucial role in reducing transaction costs and streamlining supply chains, thereby enabling small and medium-sized enterprises (SMEs) to engage in digital transformation and integrate more effectively into value chains.

This distinction is essential for understanding the spatial impact of platforms on urban systems. While C-end platforms are indicative of digital consumption patterns, B2B platforms offer a more grounded perspective on digital production and economic restructuring. From a planning and policy standpoint, focusing on B2B dynamics provides insight into the deeper mechanisms through which the digital economy reshapes urban hierarchies—not just through visibility or user activity, but through economic restructuring, enterprise transformation, and sectoral upgrading.

However, research on B2B platforms remains underdeveloped. First, most existing studies rely on aggregate firm counts without distinguishing sectoral or functional heterogeneity. Second, many adopt static descriptive methods that fail to model how B2B actively reshape inter-city economic relations and urban hierarchical structures. In China, platforms such as Alibaba’s 1688 bring together B2B firms across diverse sectors—including manufacturing, consumer goods, raw materials, and services—and embed themselves in cities through differentiated spatial logics. These patterns may reflect the formation of a new type of urban system, one fundamentally driven by platform-enabled economic restructuring.

To address these gaps, this study argues that urban system research under the platform economy should retain a focus on hierarchical and scalar dynamics while incorporating the structural transformations induced by digital technologies. It is crucial to understand how platforms cluster, which cities benefit, and whether the resulting structure is becoming more polarized. We propose that B2B platforms—at the intersection of digital and industrial spheres—serve as an effective lens through which to analyze the restructuring of urban systems (Figure 1). Focusing on B2B e-commerce firms with strong ties to the real economy, we develop an analytical framework centered on the platform “Economy–Industry Structure–Urban System Restructuring” triad. Combining urban scaling models and mechanism-based identification approaches, we aim to examine how the platform economy affects industry structure and embeds into urban systems, influences hierarchy and scale, and provides a novel explanatory model for the evolution of urban order in the digital era. This contributes to bridging the current gap in understanding the interrelations between platforms, industries, and urban structures.

3. Materials and Methods

3.1. Research Framework

This study follows a three-stage approach—data acquisition, spatial characterization, and explanatory analysis—to investigate the structure and determinants of China’s urban system in the era of mobile Internet. First, we collect data on B2B e-commerce enterprises using a third-party web crawler to extract information from the 1688 platform (www.1688.com), a leading B2B e-commerce marketplace in China. Business-to-business (B2B) e-commerce refers to digital commerce conducted between businesses through Internet-based platforms. In China, “1688.com” stands out as the largest and most representative B2B e-commerce platform. Through extensive field research and interviews with manufacturers, we found that traditional offline businesses leverage the 1688 online platform to showcase their products to business clients nationwide, effectively expanding their market reach and enhancing sales opportunities, as illustrated in Figure 2. The 1688 platform categorizes B2B enterprises primarily into four sectors: manufacturing, commercial services, consumer goods, and raw materials. Among its millions of listed suppliers, 1688 hosts over 600,000 factories, covering approximately one-tenth of China’s total number of factories and one-third of China’s large-scale manufacturing enterprises. Moreover, the platform supports a professional buyer community exceeding 65 million users, including manufacturers, retailers, and independent online retailers, facilitating a significant volume of commercial transactions and substantially underpinning China’s platform-driven economy. On the 1688 website, the enterprise-level data, including store name, store URL, business model, main product category, and city location, were obtained for 337 prefecture-level cities.

In parallel, socio-economic variables for each city were compiled from the China City Statistical Yearbook 2020. These include gross domestic product (GDP), total population, sectoral output (primary, secondary, tertiary), and employment statistics in key industries affected by platform development. After data cleaning and validation, a total of 287 cities with complete records were retained for analysis. In the second step, spatial visualization techniques were applied to analyze the geographic distribution of B2B enterprises. Using ArcGIS, we mapped the total number of B2B firms as well as the distribution of firms across industrial categories. These include raw materials, consumer goods, industrial goods, and commercial services. The third stage involves quantitative analysis of the factors influencing the development of B2B e-commerce across cities. We explore the relationship between urban characteristics and the spatial structure of B2B activity using both statistical and machine learning methods (Figure 3).

3.2. Data Collection

The primary dataset for B2B enterprises was obtained from the Alibaba-owned 1688 platform. As of 2018, 1688 accounted for 28.4% of China’s B2B e-commerce market share, making it a representative and comprehensive source for enterprise-level platform activity. The platform covers a wide range of industries, which we categorize into four first-level sectors and 46 second-level sectors (

Table 1. Industrial Sector Classification on the 1688 Platform.

First-Level Sector	Second-Level Sectors
Raw Materials	Textiles and leather, packaging, agriculture, home and building materials, pharmaceuticals, metallurgy, rubber and plastics, fine chemicals, steel, chemical raw materials
Consumer Goods	Women’s wear, men’s wear, underwear, children’s clothing, bags, shoes, maternity and baby, home goods, food and beverages, digital and computers, toys, accessories, general merchandise, beauty, gifts and crafts, car accessories, sports and outdoors, pets and gardening, office supplies, paper and printing, home textiles, home appliances, cleaning and care products
Manufacturing Goods	Machinery, electrical equipment, industrial devices, security, electronics, hardware, instruments, lighting
Commercial Services	Advertising, education and training, logistics, design, import/export services

), following the official classification system used on 1688 and previous studies [6]. To ensure the spatial completeness and consistency of the analysis, we selected all prefecture-level administrative units (cities, prefectures, leagues, and autonomous regions) as our spatial units of analysis. Based on the China City Statistical Yearbook 2020, this includes 337 units, excluding Taiwan Province, Hong Kong SAR, and Macau SAR. After removing cities not listed on the 1688 platform and county-level cities not under direct provincial control, a final dataset of 337 spatial units was confirmed.

The web-crawling process was conducted between July and December 2020. After removing duplicates, erroneous coordinates, and records with missing values, we obtained 1,389,730 valid enterprise records. Among them, 250,063 firms were in the raw materials sector, 772,582 in consumer goods, 359,718 in industrial goods, and 7367 in commercial services. For cartographic analysis of spatial distribution, we used a 1:32 million base map of China from the official standard map service (http://bzdt.ch.mnr.gov.cn/), map approval number GS (2019)1822.

Statistical data on urban economic and social development were obtained from the China City Statistical Yearbook 2021. Due to missing values in some indicators, the final analysis included 287 prefecture-level cities with complete data on population, GDP, gross value added by industry (primary, secondary, tertiary), employment in manufacturing, wholesale, and retail, and information transmission, computer services, and software sectors. Cities with missing values were excluded from statistical modeling and are clearly documented.

3.3. Methodology

3.3.1. Rank–Size and Urban Scaling Law

Zipf’s rank–size rule, proposed by G.K. Zipf in 1949, has been widely applied in urban geography research [21,24,29]. We first employ the rank–size rule to analyze the data for 337 prefecture-level and higher cities, including the number of B2B e-commerce firms per city (overall and by major types such as commercial services, manufacturing, and raw materials). The rank–size rule can be expressed by the formula:

$$In{P}_r=a-qI{n}^r$$

(1)

Urban scaling law provides a theoretical framework to quantify the relationship between urban indicators (e.g., built-up area, GDP, innovation output) and city population size. Its core assumption is that such relationships reflect the systemic properties of the urban system rather than the characteristics of individual cities. The general form of the scaling relationship follows a power law:

$$Y=Y_0{N}^{\beta }$$

(2)

3.3.2. Global Moran’s I

To analyze the spatial clustering of the B2B e-commerce industry across cities, we calculate the Global Moran’s I index. Moran’s I is used to measure the degree of spatial autocorrelation for a variable across a geographical space. It is given by:

$$I={\displaystyle \frac{n}{{\sum }_i{\sum }_j{w}_{ij}}}·{\displaystyle \frac{{\sum }_i{\sum }_j{w}_{ij}\left(B2B_i-B\overline{2}B\right)\left(B2B_j-B\overline{2}B\right)}{{\sum }_i{\left(B2B_i-B\overline{2}B\right)}^2}}$$

(3)

3.3.3. Ordinary Least Squares (OLS) Regression

To identify the key determinants of B2B e-commerce development at the city level, we construct an OLS regression model using city-level socio-economic indicators. Drawing on insights from the urban scaling framework and platform economy literature, we focus on variables that capture population size, economic scale (GDP), industrial structure (value added of the primary, secondary, and tertiary sectors), and employment in key industries. These include the number of employees in manufacturing, wholesale and retail trade, and the information technology sector.

$$B2B_i={\beta }_0+{\beta }_{1}GD{P}_i+{\beta }_{2}Po{p}_i+{\beta }_{3}Industr{y}_i+{\beta }_{4}Worker{s}_i+{\epsilon }_i$$

(4)

The OLS model serves as a baseline to evaluate the linear relationships between these variables and the scale of B2B platform development across cities. However, given the potential for non-linearity and threshold effects in urban and industrial systems, linear models alone may be insufficient for capturing the full complexity of platform-based urban dynamics.

3.3.4. Random Forest Regression and Non-Linear Mechanism Identification

To address the limitations of linear models, we employ Random Forest Regression (RFR), a non-parametric ensemble learning algorithm based on decision trees. RFR is particularly suitable for spatial and economic datasets, offering strong robustness, no distributional assumptions, and the ability to handle multicollinearity and complex non-linear interactions.

We evaluate the relative importance of each input variable using the Mean Decrease in Impurity (MDI) criterion, calculated as:

$$\widehat{f}\left(x\right)={\displaystyle \frac{1}{T}}\sum _{t=1}^T{f}_t\left(x\right)$$

(5)

where $S_t^j$ is the set of nodes in tree t where variable j is used for splitting, and ${∆I}_{s,t}$ denotes the reduction in impurity (e.g., variance), and due to the split at the node, this metric allows us to rank explanatory variables by their predictive relevance in explaining B2B platform expansion.

To further interpret the influence pathways of key predictors, we introduce the following tools:

• Partial Dependence Plots (PDP): to visualize the average marginal effect of a variable on the predicted outcome, holding other variables constant.
• Individual Conditional Expectation (ICE) plots: to assess heterogeneity in response curves across different cities.
• Segmented regression and LOWESS (Locally Weighted Scatterplot Smoothing): to identify non-linear patterns and critical thresholds in variables such as GDP, tertiary sector output, and employment in the information sector.

4. Results

4.1. Spatial Patterns and Distribution Analysis

4.1.1. Overall Spatial Distribution of B2B Enterprises

The number of B2B e-commerce enterprises varies significantly across Chinese cities (Figure 4). Spatially, B2B enterprises exhibit a clear east-to-west decreasing gradient, aligning with the long-standing industrial spatial structure of coastal–central–western China. Coastal cities demonstrate particularly high B2B activity and constitute the densest areas of platform economy development. The Pearl River Delta region—centered on Guangzhou, Foshan, Zhongshan, Dongguan, Huizhou, and Shenzhen—hosts 322,479 B2B firms, a total comparable to that of the entire Yangtze River Delta.

In parallel with established industrial hubs like Guangzhou, Shenzhen, Shanghai, and Suzhou, several emerging e-commerce cities in eastern China have achieved notable prominence. The cases of Jinhua (notably through Yiwu’s development), Dongguan, and Shantou are particularly striking, with each hosting more than 40,000 registered B2B enterprises—a clear indicator of robust growth momentum. Similar dynamic development patterns are observable in the Beijing-Tianjin-Hebei region, where B2B e-commerce has experienced accelerated expansion. Interestingly, even some traditionally less economically developed cities have managed to carve out regional competitive advantages through digital commerce initiatives. This phenomenon is exemplified by Cangzhou (Hebei Province) and Linyi (Shandong Province), both of which now boast over 30,000 B2B firms while simultaneously generating measurable spillover effects across their respective surrounding regions.

4.1.2. Sectoral Distribution Patterns of B2B Enterprises

Four primary B2B sectors reveal evident spatial concentration patterns. Of the 250,063 raw materials enterprises, 86% are located in ten eastern provinces and municipalities, with Guangdong alone accounting for 29.7%—equivalent to 5.3% of all B2B enterprises nationwide. Among the 772,582 consumer goods firms, 84.8% are situated in the eastern region. Guangdong, Zhejiang, and Fujian collectively account for 64.3% of these, representing 35.8% of total B2B firms. Industrial goods enterprises (359,718) follow a similar pattern, with 84.8% in the east. In contrast, commercial service B2B firms are fewer in number (7367), with a lower eastern concentration (71%) and a relatively higher share (22%) in central provinces.

Figure 5 illustrates distinct spatial patterns across sectors. Raw materials and industrial goods B2B enterprises are highly concentrated in coastal cities and a few regional hubs in central and western China, reflecting the influence of industrial bases. Consumer goods B2B firms show widespread clustering along the southeastern coast and the North China Plain, with particularly high densities in the Pearl River Delta, Chaoshan region, southeastern Fujian, and Yangtze River Delta. Unexpectedly, certain northern cities also exhibit strong consumer goods clustering. Linyi (Shandong) has become a key node with high concentrations in general merchandise, home goods, appliances, cleaning products, beauty, and car accessories. Baoding (Hebei), focused on maternal and infant products, car accessories, toys, and sports goods, hosts 21,232 B2B enterprises—surpassing Beijing’s 18,622. The spatial layout of commercial service B2B firms differs from that of traditional services, with no significant concentration in top-tier cities. Instead, they are distributed across cities such as Wuhan, Shenzhen, Zhengzhou, Shanghai, Jinhua, Wenzhou, Xiamen, and Qingdao. Logistics service firms are mainly concentrated in transport hubs or cities with active e-commerce sectors and high logistics demand, such as Shenzhen, Guangzhou, Changsha, Ganzhou, and Tianjin. Import–export agents exhibit a coastal and port-oriented distribution, clustering in Shenzhen, Guangzhou, Jinhua, Shanghai, Qingdao, Ningbo, and Xiamen.

4.1.3. Regional Spatial Characteristics of the B2B E-Commerce Sector

Regionally, eastern cities lead across all four B2B categories, with enterprise numbers consistently in the upper quartile. Cities such as Guangzhou, Yiwu, and Linyi function as “polarized centers” that drive regional platform economies (Figure 5). Central cities fall within the mid-range, exhibiting stable means but notable variation, reflecting a phase of exploration and spatial imbalance in B2B development. Western cities show overall low enterprise counts and a “long-tail drag” phenomenon, with lagging performance due to constraints in digital infrastructure and industrial support.

Sector-specific regional differences are also apparent. Manufacturing-oriented platforms are most concentrated in the east, with boxplot distributions indicating high and tightly clustered values—evidence of “industry–platform coupling”. Service-oriented platforms exhibit lower regional variation but show outlier volatility, reflecting a “floating” pattern detached from local industry bases (Figure 6). Consumer goods platforms are more active in the east and central regions, following a “distribution-driven” path guided by end-market demand. Raw materials platforms are weakest in the west, indicating the digital lag of midstream industrial links and the high threshold for transformation.

4.1.4. Spatial Autocorrelation and Interpolation Analysis

Spatial autocorrelation analysis reveals significant clustering of B2B e-commerce firms across Chinese cities, closely aligned with the existing urban system structure [52,53]. Local spatial autocorrelation analysis using GeoDa identifies prominent “high–high” clusters in coastal regions such as the Pearl River Delta, Yangtze River Delta, and Shandong Peninsula, as well as Cangzhou and Langfang in Hebei, with adjacent “low–high” transition zones.

Further analysis shows a strong correspondence between the spatial distribution of B2B firms and the layout of national urban agglomerations. Using ArcGIS for spatial interpolation at the prefecture-level (Figure 7), we find high congruence between B2B enterprise density and the six major urban clusters: Yangtze River Delta, Pearl River Delta, Beijing–Tianjin–Hebei, Chengdu–Chongqing, Middle Yangtze River, and Central Plains (

Table 2. Number of B2B e-commerce companies in typical urban clusters in China.

City Cluster	The Number of Cities	The Number of B2B Companies
Pearl River Delta	9	348,780
Yangtze River Delta	26	278,821
Beijing-Tianjin-Hebei	13	131,060
Central Plains	29	83,439
Shandong Peninsula	8	51,844
Chengdu-Chongqing	16	14,094
Sum	101	908,038

).

The Yangtze River Delta cluster (including Shanghai, Hangzhou, and Nanjing among 26 cities) hosts 278,821 B2B firms. The Pearl River Delta (including Guangzhou, Shenzhen, and Dongguan among 9 cities) contains 348,780, more than half of which are consumer goods platforms—making it the most densely concentrated and influential B2B core zone nationwide [54]. In contrast, the Beijing–Tianjin–Hebei region hosts 131,060 firms, but its overall concentration is lower. This region’s industrial composition is more balanced between consumer and industrial goods sectors, forming a multi-sectoral embedded pattern.

Overall, eastern coastal provinces dominate B2B enterprise distribution, with industrial weight shifting further eastward and southward. Central regions exhibit a pattern of “centralized and contiguous” development, with strong performance in core cities and emerging continuous growth in surrounding areas. Western regions demonstrate relatively weak clustering and lower development levels.

4.2. Empirical Analysis

4.2.1. Analysis of Urban Hierarchical Structure

(1) 

Rank–Size Distribution

To examine hierarchical disparities in the distribution of B2B e-commerce firms across China, we conducted a rank–size analysis based on the total number of B2B enterprises in 337 prefecture-level and above cities, as well as their breakdown across four primary sectors: commercial services, consumer goods, industrial goods, and raw materials. The results reveal a pronounced primacy effect in all cases, indicating intensified polarization within the urban system, using population (based on the 2020 census) and GDP as comparative indicators. According to log–log scatter plots and fitted curves (Figure 8;

Table 3. Results of city Rank-Size based on population, GDP, and B2B e-commerce.

Equation			$\mathit{I}\mathit{n}{\mathit{P}}_{\mathit{r}}=\mathit{a}-\mathit{q}\mathit{I}{\mathit{n}}^{\mathit{r}}$
	Population	GDP	B2B Sum	Business Services	Manufacturing	Raw Materials	Consumer
Weight	Unweighted
Intercept ($a$)	9.4977 ± 0.7914	10.215 ± 0.1279	17.688 ± 0.3452	9.3621 ± 0.1568	17.8900 ± 0.6523	15.3390 ± 0.1356	17.24 ± 0.2450
Slope ($q$)	−0.7914 ± 0.0275	−1.0543 ± 0.0259	−2.4545 ± 0.0356	−1.6805 ± 0.0195	−2.5592 ± 0.02688	−2.2992 ± 0.0275	−2.7208 ± 0.01488
Residual sum of squares	55.492	71.15324	62.855	23.954	71.690	55.244	68.103
Pearson’s r	0.908 **	−0.908 **	−0.895 **	−0.961 **	−0.879 **	−0.908 **	−0.885 **
$R^2$ (COD)	0.7112	0.82772	0.8004	0.9240	0.7724	0.8246	0.7838

Note: ** p < 0.05.

), six variables—population, GDP, total B2B enterprises, and the number of firms in each B2B sector—conform to the rank–size distribution model. all with R² values above 0.7, suggesting good fit and statistical significance. However, Zipf exponents vary markedly across indicators. The Zipf exponent for population is approximately 0.8, significantly less than 1, indicating a relatively weak primacy structure—mid-ranked cities also maintain considerable population sizes. The exponent for GDP is around 1.05, suggesting moderate primacy and stronger economic concentration. In contrast, Zipf exponents for B2B enterprise totals and their subtypes all exceed 1.5; for manufacturing, industrial, and raw materials B2B sectors, the exponents exceed 2.2, representing a classic Pareto-type distribution.

Compared to population and GDP, the development of B2B enterprises shows more pronounced hierarchical disparities, with higher-ranked cities demonstrating far greater B2B activity. This is particularly evident in manufacturing, industrial, and raw materials sectors, which tend to concentrate in a few major cities. Contrary to the notion of digital decentralization, the platform economy under mobile Internet does not diffuse evenly. Instead, cities with superior resource endowments and industrial bases stand out more rapidly, intensifying hierarchical differentiation and reinforcing the Matthew effect.

(2) 

Urban Scaling Law

The urban scaling analysis indicates a superliner agglomeration trend for B2B enterprises, with power-law exponents greater than 1. Larger cities are associated with disproportionately more B2B firms. Notably, the scaling exponent for GDP reaches 1.58, indicating exponential growth of B2B activity in more economically advanced cities and reinforcing existing economic hierarchies (Figure 9).

To systematically analyze the determinants of B2B firm distribution, we adapt the conventional urban scaling model to a multivariate regression framework. Our empirical results reveal distinct sectoral variations: the consumer goods sector demonstrates the strongest association with B2B growth (β = 0.589, p < 0.01), followed by raw materials (β = 0.260, p < 0.01). These findings imply that platform economic development exhibits stronger correlations with sector-specific supply-demand dynamics.

$$\log B2B={\beta }_1\times \log Pop+{\beta }_2\times \log GDP+\sum _k{\gamma }_k{Z}_k+\epsilon$$

(6)

When controlling for industrial structure, the effects of population and GDP become statistically insignificant, implying that traditional urban hierarchy indicators are not decisive in the platform economy. In addition, employment in IT and manufacturing sectors shows no significant impact, suggesting that B2B platforms are characterized by “light-asset” models with less reliance on traditional labor inputs and more dependence on digital infrastructure and network connectivity.

In sum, the B2B platform economy exhibits highly hierarchical, superliner, and structurally polarized patterns across China’s urban system. Its spatial logic aligns more closely with a “platform–industry coupling” model of resource reconcentration, rather than technology-driven, uniform diffusion—highlighting critical implications for urban policy and platform governance.

4.2.2. Mechanism Identification of B2B Platform Development

(1) 

OLS Regression Analysis

Which urban factors influence the development of B2B e-commerce? To address this question, we calculated correlations among key indicators. All variables were log-transformed to reduce skewness and heteroscedasticity, thereby improving model robustness. The log-transformed correlation heatmap shows strong positive relationships between B2B firm counts and population size (r = 0.70), GDP (r = 0.40), and tertiary sector share (r = 0.27). Manufacturing employment is particularly strongly correlated (r = 0.77), highlighting the role of industrial agglomeration in supporting B2B expansion. In contrast, the primary sector share is negatively correlated with B2B development, reflecting weak ties between B2B e-commerce and agriculture (Figure 10).

OLS regression models were constructed using the total number of B2B firms and sector-specific counts (industrial, commercial services, raw materials, consumer goods) as dependent variables. Due to high collinearity (r > 0.9) between mobile phone users and population, and between IT employees and commercial service employees, these variables were excluded to avoid multicollinearity. The independent variables include population, per capita GDP, industrial structure (primary, secondary, tertiary), employment structure (manufacturing and IT employment), and regional dummies (east, west). Key coefficients, standard errors, and significance levels are presented in

Table 4. Regression model based on urban scaling law.

	Constant	Pop	GDP	Manf	Comp	Raw	Serv	IT-e	Manf-e
$\mathit{\beta }$	1.023 ***	−0.014	0.016	0.155 ***	0.589 ***	0.260 ***	0.008	−0.02	0.035
T	11.695	0.779	0.695	7.92	29.001	23.69	1.079	−1.68	0.158
${\mathit{R}}^2$	0.897

Note: *** Correlation is significant at the 0.01 level (two-tailed).

. R² values are approximately 0.8 for all sectors except commercial services (R² = 0.562), indicating that the models explain over 80% (or at least 50%) of the variation in B2B firm distribution. F-statistics are significant at the 0.001 level, and variance inflation factors (VIFs) are below 6, confirming limited multicollinearity [55]. Population, secondary and tertiary sector shares, and manufacturing employment have significant positive effects on B2B development (p < 0.01), indicating strong ties between urban economic activity and B2B platform growth.

The manufacturing employment base emerges as a particularly critical determinant of B2B expansion. Industrial powerhouses like Dongguan and Guangzhou have demonstrated a remarkable capacity to scale their B2B sectors in the mobile Internet era, leveraging their established industrial ecosystems. Our analysis reveals pronounced regional disparities: eastern cities show robust positive associations with B2B development (p < 0.001), whereas western cities display consistently negative coefficients (p < 0.001), highlighting fundamental divergences in regional economic maturity and digital market preparedness.

Three key patterns emerge from sectoral analysis:

• Population size exerts uniformly positive effects across all B2B categories (p < 0.001).
• Industrial composition shows differential impacts—primary sector development negatively correlates with commercial services platforms, while secondary and tertiary sector growth significantly accelerates industrial, raw materials, and consumer goods platforms.

The manufacturing workforce effect proves especially salient, with cities boasting larger industrial employment bases supporting significantly greater concentrations of both industrial and consumer-oriented B2B firms (p < 0.001). Regional analysis confirms eastern China’s comprehensive advantage across all platform types (p < 0.001), while western regions generally constrain development—with the notable exception of commercial services platforms, which show regional neutrality (

Table 5. Results of OLS Regression Analysis.

	B2B	Manufacturing	Service	Raw	Consumer
ln Population	(1.419)	(1.497)	(0.704)	(1.458)	(1.343)
	0.000 ***	0.000 ***	0.001 ***	0.000 ***	0.000 ***
ln GDP	(0.465)	(0.445)	(−0.397)	(0.511)	(0.382)
	0.066 *	0.116	0.160	0.034 **	0.153
ln primary	(−0.080)	(−0.020)	(−0.418)	(−0.128)	(−0.114)
	0.512	0.883	0.003 **	0.274	0.378
ln secondary	(0.686)	(0.830)	(−0.097)	(0.550)	(0.663)
	0.013 **	0.007 **	0.751	0.036 **	0.023 **
ln tertiary	(1.189)	(1.202)	(0.531)	(1.601)	(0.982)
	0.003 **	0.007 **	0.233	0.000 ***	0.020 **
ln manufacturing	(0.400)	(0.445)	(0.283)	(0.347)	(0.406)
	0.001 ***	0.001 ***	0.034 *	0.002 **	0.001 ***
ln IT employees	(−0.289)	(−0.257)	(0.019)	(−0.300)	(−0.275)
	0.029 **	0.082 *	0.899	0.017 **	0.048 **
ln east	(1.168)	(1.348)	(0.922)	(1.111)	(1.183)
	0.000 ***	0.000 ***	0.000 ***	0.000 ***	0.000 ***
ln west	(−1.536)	(−1.782)	(−0.196)	(−1.139)	(−1.589)
	0.000 ***	0.000 ***	0.272	0.000 ***	0.000 ***
const	(−2.237)	(−4.046)	(−3.443)	(−4.215)	(−2.452)
	0.092 *	0.007 *	0.021 **	0.001 ***	0.080 *
N	289	289	289	289	289
R2	0.811	0.808	0.562	0.807	0.791
Adj. R2	0.805	0.802	0.547	0.801	0.784
F	0.000	0.000	0.000	0.000	0.000

Note: *** p < 0.001, ** p < 0.05, * p < 0.1.

).

(2) 

Variable Importance Ranking from Random Forest

To further validate the explanatory power of variables identified in the OLS model under a more complex non-linear framework—and to uncover hidden mechanisms behind the spatial agglomeration of B2B e-commerce—this study applies a random forest regression model. This approach ranks variable importance and enables deeper exploration of marginal effects and non-linear patterns using Partial Dependence Plots (PDP), Individual Conditional Expectation (ICE) curves, and segmented regression techniques.

After excluding the four sector-specific B2B subtype variables, we developed a random forest model using total B2B firm counts as the dependent variable. The variable importance rankings remained largely consistent with the OLS regression results, with traditional structural factors including GDP, population size, and manufacturing employment retaining high importance scores—reaffirming their fundamental role in shaping the spatial distribution of platform enterprises. Notably, the number of IT employees and tertiary sector size demonstrated significantly greater importance in the random forest model compared to the OLS results. This divergence suggests that in non-linear modeling frameworks accounting for interaction effects, these variables representing digital economy capacity and service-oriented ecosystems exert stronger marginal impacts on B2B development. The findings collectively indicate that while platform economy growth remains grounded in traditional economic foundations, its expansion dynamics and competitive potential increasingly depend on digital infrastructure and service sector coordination capabilities. These results highlight the complementary value of employing both linear and non-linear modeling approaches to fully capture the multidimensional drivers of B2B platform development across different regional contexts (Figure 11).

(3) 

Marginal Effects and Non-linear Mechanism Identification

Building on the variable importance ranking and OLS analysis, we selected six core variables—GDP, population, manufacturing employees, IT employees, east (regional dummy), and secondary sector share—to construct PDP and ICE plots, thereby analyzing marginal contributions and inter-city heterogeneity.

• The development of the B2B e-commerce platform economy is more dependent on structural and industrial foundations rather than linear expansion driven by aggregate indicators. The Partial Dependence Plot (PDP) results indicate that the marginal effect of GDP remains relatively stable overall, without exhibiting a significant acceleration trend. Individual Conditional Expectation (ICE) curves reveal pronounced urban heterogeneity: high-GDP cities, such as top-tier metropolitan areas, show declining responses, while most mid- and low-GDP cities display increasing trajectories.
• Population serves as a fundamental variable for the development of B2B platforms, with a stable positive effect but limited marginal contribution. The PDP curve demonstrates a consistent upward effect of population on the number of B2B firms, particularly when log (Population) exceeds 6.0 (approximately corresponding to a resident population over four million). The ICE curves exhibit strong alignment, suggesting a similar response logic across most cities. Overall, population acts as a foundational support in terms of both consumption and labor supply, providing a “floor effect” in the spatial expansion of B2B platforms, rather than serving as a driver of explosive growth.
• The number of manufacturing employees exhibits a clear “threshold effect” and serves as a dominant support factor for the platform economy. The PDP curve rises sharply between log values of 1.5 to 3.0, indicating that the growth of B2B platforms relies on reaching a critical mass in manufacturing employment, which then activates synergistic effects. This non-linear relationship suggests that the platform economy is not a “detachment from the real economy,” but rather one that effectively embeds itself only upon a substantial industrial foundation. ICE curves generally trend upwards with minimal variation across cities, indicating that manufacturing labor provides a strong, broadly applicable, and stable incentive for platform development.
• IT employment demonstrates a robust positive marginal effect, with the partial dependence plot (PDP) revealing a particularly pronounced slope in the mid-to-high value range. Individual conditional expectation (ICE) curves maintain a steady upward trajectory across observations, underscoring the growing criticality of digital workforce capabilities for platform ecosystem development. These patterns capture an ongoing transition in economic organization—from traditional manufacturing-centric models toward information-driven platform coordination mechanisms.
• Regional analysis through the eastern region dummy variable (east) uncovers substantial structural advantages. The PDP illustrates systematically elevated B2B concentrations in eastern cities, while ICE curve clustering demonstrates consistent group-level benefits. These findings empirically validate three key advantages of eastern regions: (1) superior institutional environments, (2) more developed digital and physical infrastructure, and (3) more mature industrial ecosystems that collectively foster platform economy growth. The results align with existing literature on regional innovation systems while providing new quantitative evidence of their impact on digital platform diffusion [21,22,51].
• The share of the secondary sector has a moderate negative effect on B2B platform development and shows signs of marginal decline. The secondary industry variable demonstrates a weak positive correlation with the number of B2B firms, with a flattening slope in the higher range. ICE curves are more dispersed, reflecting inter-city differences in dependency on the industrial sector. While the secondary industry still functions as an “industrial anchor” for the platform economy, its driving force diminishes with structural upgrading, gradually giving way to digital factors and enhanced industrial–informational integration.

To further identify threshold effects and critical inflection points, we introduced LOWESS smoothing and segmented linear regression models. The results show the following (Figure 12 and Figure 13):

• The GDP variable exhibits a marginal shift at a log value of approximately 1.70, showing a rapid growth trend in the early stages. However, the growth rate of B2B platform firms gradually slows down, and in the extremely high GDP range (log > 2.5), the marginal effect weakens and even turns negative. This indicates that the development of the platform economy is not solely driven by aggregate economic volume but rather relies on the combined support of industrial structure, demographic scale, and other complementary factors.
• Population maintains a smooth linear growth trend. LOWESS and segmented regression results are consistent, indicating that population primarily provides market size and user base support, but with stable marginal returns and no clear non-linear jump.
• Manufacturing employment exhibits a “structural transition” effect. In cities with a weak manufacturing base, the number of B2B firms increases rapidly as employment rises. However, once the log-transformed value reaches approximately 1.72, the growth rate slows, indicating a saturation point. This suggests that while labor-intensive industries facilitate the initial expansion of platform economies, they may impede spatial coordination during the upgrade phase, as the reliance on traditional manufacturing becomes less conducive to sustaining platform-based growth.
• IT employment exhibits a highly non-linear positive effect, characterized by a “technological take-off point” at a log value around −1.05. Before this point, the increase in IT employment leads to a rapid rise in B2B firm numbers, indicating that even a small enhancement in IT capacity significantly boosts platform economy growth in cities with weaker digital foundations. However, once this threshold is crossed, the growth rate becomes more stable and linear, suggesting that in cities where IT employment has already reached a substantial level, further increases do not proportionally enhance B2B development. This confirms the central role of IT talent in the initial stage of platform expansion but also indicates that after achieving a certain digital maturity, B2B growth becomes less dependent on continued IT workforce expansion.
• Secondary sector share exerts a positive effect at lower levels but shows a turning point at −0.94, after which marginal returns become negative. The LOWESS curve also declines, indicating a possible “industrial mismatch” problem. When industrial structure misaligns with platform logic, it may hinder further development.

In summary, B2B platform growth is shaped by non-linear dynamics where GDP and manufacturing employment show early-stage boosts followed by saturation, then IT employment triggers rapid initial growth, then stabilizes, population offers consistent linear support, and a high secondary sector share may lead to industrial mismatch.

(4) 

Heterogeneity Analysis

Building upon the spatial and mechanism-based analyses, B2B e-commerce development in China exhibits marked regional and sectoral heterogeneity. To examine whether the observed mechanisms remain stable across different regions and types of B2B platforms—or whether structural divergence exists—this study constructs sub-sample random forest models based on regional divisions (eastern vs. western cities) and B2B sector categories. Partial Dependence Plots (PDP) and variable importance heatmaps are used to systematically identify heterogeneous expansion mechanisms of the B2B platform economy. According to the variable importance rankings from the sub-sample random forest models, eastern and western cities demonstrate clear divergence in the key drivers of B2B development. In eastern cities, variables such as GDP, IT employees, and tertiary sector share exhibit high importance scores, suggesting that platform growth has entered a “high-foundation–high-elasticity” stage (Figure 14). In this phase, digital talent and service capacity become central to enhancing platform competitiveness.

In contrast, B2B development in western cities remains more reliant on traditional factor endowments, with population, manufacturing employment, and secondary sector share emerging as the most influential variables. This reflects a “low-path-dependence” trajectory, whereby platform expansion depends on conventional industrial foundations. Although the platform economy is extending inland, these results confirm the existence of stage-based regional divergence in its development mechanisms.

PDP and ICE plots for both sub-samples further illustrate the spatial differences in marginal effects (Figure 15 and Figure 16):

• The marginal effect of GDP exhibits a declining trend in eastern cities, with the PDP curve showing a slight downward slope and ICE curves reflecting a saturation in platform firm growth in the mid-to-high GDP range. This suggests that in economically advanced eastern cities, platform development may have moved beyond the foundational stage into a phase of structural integration, where GDP alone is no longer a decisive driver of platform expansion. In contrast, GDP still plays a positive role in western cities, where the PDP curve shows a gradual upward trend and ICE curves are more concentrated, indicating that the platform economy is still in a stage of potential being gradually unlocked through economic growth.
• With respect to population, eastern cities display a clear “leap response” structure. The PDP curve rises steeply in the range of log (Population) ≈ 5.5 to 6.0, and ICE curves are sharply and consistently upward, indicating that B2B platform development is dependent on reaching a critical urban population threshold, producing a non-linear amplification effect through population density. In western cities, however, the slope of the population PDP is relatively mild, and ICE curves are more dispersed, suggesting that population has not yet emerged as a defining factor in triggering platform expansion.
• The marginal effect of IT employment is relatively stable in eastern cities but increasing in western regions. In the east, the PDP curve is nearly flat, and ICE curves remain at high levels with limited upward movement, indicating that digital talent has already been embedded in the organizational logic of platforms and provides consistent but stable support. In contrast, the western PDP curve has a visibly positive slope, and ICE curves show broader dispersion and steeper ascents, implying that digital elements are still in a rapid accumulation and transformation phase, now becoming a key emerging driver of platform growth.
• The spatial divergence of manufacturing employment is even more pronounced. In western cities, the PDP curve rises steadily, and ICE curves trend consistently upward, indicating that manufacturing remains the primary foundation for B2B platform development, with the platform economy deeply embedded in traditional industrial structures. In eastern cities, however, the PDP curve is flatter, and ICE curves are closely clustered but lack an upward trend, suggesting that platform growth is increasingly decoupled from industrial dependence. However, when the number of people employed in the manufacturing industry reached a higher level (about four, logarithmic scale), the number of B2B e-commerce companies still increased significantly.

To identify sectoral-level heterogeneity, this study constructs industry structure variables based on four dominant types of B2B platforms: manufacturing, services, raw materials, and consumer goods. The results show distinct mechanisms across different platform types:

• For manufacturing-oriented B2B platform firms, GDP exhibits a relatively flat response, and ICE curves show no major fluctuation, indicating that aggregate economic size plays a limited role in this type of platform development. In contrast, population shows significantly stronger marginal effects, with the PDP curve rising sharply above log(Population) = 5.5, and ICE curves trending consistently upward. This suggests that manufacturing platforms remain embedded within traditional industrial foundations and spatial market logic, primarily driven by transitions in population scale.
• The growth logic of service-oriented B2B platforms exhibits stronger structural responsiveness. PDP and ICE curves indicate heightened marginal sensitivity to population, with “jump-like” increases in platform numbers in high-population cities, forming a clear population threshold effect. The share of the tertiary sector also begins to rise beyond log values of −0.8, showing that industrial upgrading is emerging as a supportive force for service platform expansion. Meanwhile, GDP and IT employment show weaker marginal effects, suggesting that service-oriented platforms are still transitioning from a “population–consumption” model toward a “structure–digital” model.
• Consumer-oriented B2B platforms demonstrate a multi-driver growth mechanism. PDP and ICE curves show consistent positive relationships with GDP, population size, IT workforce levels, and tertiary sector development, though the strength of these relationships varies considerably between cities. These patterns suggest consumer platforms now develop through integrated structural drivers rather than isolated factors. These platforms embody a “population–industry–digital” fusion model, aligning most closely with the ideal trajectory for high-quality platform economy development.
• The growth mechanism of raw material-oriented platforms displays stronger path dependency. Their responses to GDP, IT employment, and the tertiary sector are all relatively flat, with the PDP curve remaining nearly level and ICE curves showing low-level clustering. This reflects that digital transformation in this segment remains in its early stage.

Further PDP and ICE plots reinforce these findings (due to the limited length of the article, see Appendix A). In consumer-oriented cities, marginal effects are more stable, and returns on information-based variables are higher. In manufacturing-oriented cities, many variables exhibit a “rise-then-fall” non-linear structure, reflecting risks of marginal saturation or contraction.

Overall, the heterogeneity analysis reveals that B2B platform expansion cannot be assumed to follow a “technologically neutral” trajectory. Instead, adaptive mechanisms must be designed according to sectoral types and urban ecosystems. These findings highlight the necessity of place-specific and industry-sensitive strategies for effective digital spatial governance.

5. Discussion

The platform economy is increasingly shaping urban economic development and driving the restructuring of urban systems [56,57,58]. As information technology rapidly advances, traditional industries are increasingly integrated with the digital economy, fostering B2B e-commerce platforms that not only intensify polarization and differentiation within urban systems but also offer traditional industrial cities opportunities for leapfrogging, thereby reshaping spatial organization and structural configurations [6,51,59,60,61]. This study emphasizes the necessity of incorporating the platform economy into urban system research, moving beyond the conventional focus on consumer-oriented (C-end) platforms [62,63]. Instead, it returns to the foundational logic of urban system studies: examining how changes in urban economies fundamentally reshape urban structures [64,65]. Figure 17 shows that the platform economy embedding itself into various urban industries promotes industrial digitalization and enhances economic efficiency, particularly by facilitating the digital and informational transformation of traditional sectors, such as manufacturing, consumer, and services [66,67]. This transformation promotes urban economic growth and functional upgrading. As an essential carrier of the platform economy, cities undergo significant spatial and economic restructuring under the dual forces of digital transformation and industrial upgrading. This results in the formation of new spatial layouts and economic nodes, further altering the hierarchical structure of urban systems. This evolution not only strengthens the dominance of first-tier and megacities but also diversifies the functional roles of small and medium-sized cities, creating a platform economy-driven urban network system (Figure 17).

Analyzing the distribution of B2B E-Commerce Enterprises demonstrates that B2B platforms significantly reinforce existing urban hierarchies rather than promote balanced, decentralized urban growth. While much research suggests that digital platforms gradually shift urban spatial organization from hierarchical, distance-based “central places” to network-based “central flows”, our findings highlight an apparent “superlinear agglomeration” phenomenon within B2B e-commerce enterprises. On the one hand, platforms amplify the competitive advantages of established economic centers, notably within China’s Yangtze River Delta and Pearl River Delta regions. Here, the interplay between traditional industrial bases and platform-driven digital economies has resulted in a pronounced “winner-take-all” scenario, further consolidating these regions’ dominance in the national urban system [39,68,69]. Conversely, cities such as Linyi, Cangzhou, and Yiwu, characterized by specialized industrial advantages, have successfully leveraged B2B e-commerce platforms to emerge as significant new nodes within regional networks. These findings underscore how the dual agglomeration and dispersion capacities of the Internet and mobile Internet reshape new economic spatial patterns, manifesting as central polarization coupled with peripheral emergence. This dual dynamic results in a platform economy-driven urban system characterized by strengthened core cities, the rise of new emerging nodes, and enhanced opportunities for peripheral urban areas.

A central contribution of this research is its elucidation of the non-linear and heterogeneous mechanisms underlying the relationships between urban attributes (GDP, population, manufacturing employment, digital workforce) and platform growth [70,71,72]. Cities exhibit clear thresholds and saturation points, highlighting structural mismatches associated with excessive reliance on traditional manufacturing employment or GDP expansion alone. The diminishing marginal returns of GDP at higher development levels and the plateauing effects of manufacturing employment suggest that future urban digital strategies must prioritize structural coordination rather than merely economic scale expansion. Moreover, pronounced regional disparities are highlighted through the contrasting developmental paths observed in eastern and western Chinese cities. Eastern cities benefit substantially from digital synergy and advanced industrial structures, whereas western cities often face constraints due to weaker digital infrastructures and path-dependent industrial embedding. Consequently, tailored policy approaches are essential. Western regions should prioritize foundational digital infrastructure investments and industrial upgrading, while eastern regions—already digitally mature—should enhance service-oriented platform ecosystems and digital talent capacities.

Based on the findings of this study, several adaptive mechanisms for urban planning and policy interventions are proposed to support the growth of B2B platforms in various urban contexts. The planners and policymakers could be tailored to the unique industrial and digital profiles of cities, addressing the non-linear growth dynamics identified in the analysis. For cities with strong manufacturing bases but limited digital infrastructure, the focus should be on enhancing the digital integration of traditional industries, particularly through the adoption of B2B platforms by traditional industries. Investments in digital technologies can significantly improve supply chain efficiencies and market access for local industries. Additionally, fostering digital innovation hubs and investing in digital workforce training are critical to support ongoing platform growth in these cities. In cities with high GDP, where platform growth has started to show diminishing returns, the emphasis should shift toward service-oriented digital platforms and the development of digital talent. Policies should encourage cross-sector synergies between traditional industries, such as finance, logistics, and technology, to further expand the platform economy in these more mature urban areas. For cities with high population growth, the focus should be on ensuring that digital infrastructures leverage the large consumption and labor supply to foster B2B platform development. By facilitating greater digital access, these cities can better integrate their population into the platform economy, thereby supporting inclusive growth. Finally, significant regional disparities exist in the development of B2B platforms. Eastern cities, with better-established digital infrastructures and service sectors, should continue to build on their service-driven platforms while investing in digital talent pools. In contrast, western cities, which remain more reliant on traditional manufacturing, should prioritize investments in digital infrastructure to enable their integration into the platform economy and support industrial upgrading.

From an international perspective, similar phenomena of digital-induced urban polarization are evident globally [73,74,75]. Sohn, Kim, and Hewings compared the impacts of information technology on the urban spatial structures of Chicago and Seoul, demonstrating that while Chicago exhibited clear spatial structure reinforcement, Seoul experienced a more dispersed spatial pattern induced by information technology [76]. Ioannides et al. investigated the impact of fixed telephony on urban structures between 1980 and 2000, indicating that an increase in telephone lines per capita encouraged spatial decentralization of populations but led to greater concentration in urban size distribution [77]. Thus, insights from China’s experience with B2B-driven urban transformation can provide valuable comparative lessons, particularly for developing economies seeking effective digital urban development strategies. Furthermore, this study addresses a significant gap in current academic discourse, which tends to focus predominantly on consumer-oriented (C-end) platforms while overlooking the structural transformative impacts of B2B e-commerce platforms. Given the profound integration of B2B platforms within manufacturing and logistics chains, future research should expand investigations into their impacts on urban industrial organization, labor markets, and spatial planning practices.

Despite the contributions of this study, there are several limitations that must be acknowledged, which could impact the interpretation and generalizability of the findings. One primary limitation is the cross-sectional nature of the data, which are derived from a single platform, 1688, over a limited time period. This dataset may not fully capture the broader scope and the dynamics of the digital economy to the urban economy. As such, the findings may be constrained in terms of their ability to reflect the broader trends across multiple platforms or over time. Additionally, while 1688 is a significant and representative B2B e-commerce platform in China, relying solely on this platform may introduce biases, as it does not account for the variability in platform types or the diversity of industries that other platforms, such as Alibaba’s international sites or Amazon Business, represent. This limitation may influence the validity of the results, particularly when examining sectoral differences or comparing the findings with those in other geographical contexts. Another limitation lies in the regional scope of the study, primarily focused on China, and this may limit the broader applicability of the findings to other countries or regions with different economic conditions or digital infrastructures. Future research should incorporate longitudinal data and data from multiple platforms—such as Alibaba’s international sites or Amazon Business—to improve the generalizability of the results. Expanding the study to include other regions would also help enhance the external validity of the findings.

6. Conclusions

In conclusion, this study highlights that the platform economy does not lead to a uniform flattening of urban hierarchies, but rather drives a dynamic and differentiated restructuring of urban systems. By embedding into various industrial structures, the platform economy fosters industrial digitalization and urban economic upgrading. This dual process transforms traditional economic patterns and reshapes the spatial organization of cities, reinforcing the dominance of developed regions while creating growth opportunities for emerging cities through platform integration. The restructuring of China’s urban system under the platform economy is characterized by concentrated agglomeration, marginal thresholds, and distinct regional trajectories, rather than balanced diffusion. B2B platforms, as key agents in this transformation, play a critical role in reshaping urban systems. Theories such as Christaller’s central place theory, which emphasize center–periphery hierarchies, remain relevant for understanding the continued dominance of core cities in the platform economy. However, this study also reveals that the integration of B2B platforms in peripheral cities challenges traditional hierarchical structures, uncovering the potential for new emerging nodes within urban systems. This shift goes beyond the classic center–periphery dynamic, showing how the platform economy is enabling previously peripheral cities to better integrate into the global digital economy.

To facilitate the effective development of B2B platforms, urban planners and policymakers must recognize the non-linear dynamics identified in this study, which are shaped by key factors such as GDP, manufacturing employment, IT employment, and population size. Policymakers should implement adaptive mechanisms that cater to the unique characteristics of cities, particularly their industrial bases and digital capacities. In cities with strong manufacturing foundations, investments in platform integration and digital innovation hubs should be prioritized, while in more developed cities, policies should focus on the growth of service-oriented platforms and digital talent development. Regional disparities must also be addressed through targeted interventions, ensuring that western cities invest in digital infrastructure while eastern cities expand their service-based digital economies. These findings not only contribute to our understanding of how digital platforms are reshaping urban hierarchies but also provide actionable insights for policymakers to design effective and context-specific strategies to manage urban transformations in the digital era.