Location Criteria for E-Commerce Logistics Facilities: A Scale-Sensitive Analysis

Abstract

The rapid proliferation of e-commerce has reshaped the spatial logic and facility typologies of urban logistics. While the literature on logistics facility location selection is extensive, there is limited understanding of how the relative importance of location criteria varies across facility types shaped by e-commerce. This study addresses this gap by analyzing the location criteria of logistics facilities of different sizes using a multi-criteria decision-making (MCDM) approach. Twenty-five criteria, identified through a literature review and feedback from seven experts in the Istanbul e-commerce logistics sector, were analyzed using the Fuzzy Simple Additive Weighting (SAW) method. The relative weights of criteria were calculated for three facility scales, macro-, meso-, and micro-scales, to reveal how location priorities vary across scales. Proximity to main arteries ranks first across all scales (macro: 0.317, meso: 0.431, micro: 0.409). Land rental values are highly prioritized at both the macro- and meso-scale, while population density ranks prominently at the macro- and micro-scale. At the meso-scale, shopping mall proximity gains notable weight, whereas intermediate arteries stand out as a key factor at the micro scale. These findings advance the understanding of scale-sensitive dynamics in urban logistics and provide a framework for more adaptable and sustainable logistics planning.

1. Introduction

Rapid urbanization has reshaped urban economies, infrastructures, and spatial organization worldwide, with logistics land use emerging as one of the most affected domains. Inefficient siting of logistics facilities generates congestion, emissions, and noise [1,2,3], challenging even advanced cities [4,5]. Since the 2000s, the expansion of e-commerce, exacerbated by the COVID-19 epidemic, has profoundly transformed the spatial dynamics of logistics facilities. Rising demand for fast and flexible delivery has intensified pressure on logistics structures, not only on how goods are stored and delivered but also on the spatial layout within the city, creating a transformation in location selection. Urban renewal, rising land prices, and competition for limited space have made dense areas increasingly unsuitable for large logistics operations [6]. Yet, e-commerce logistics simultaneously requires both large fulfillment centers near consumers and neighborhood-scale last-mile facilities, generating tensions with urban mobility and land scarcity [7]. Some logistics-intensive industries have even begun reoccupying inner-city warehouse spaces [8], raising new challenges for metropolitan planning at multiple scales [9]. These evolving pressures underline the need for analytical frameworks that can explain how economic, spatial, and social factors jointly shape logistics location behavior. While cost, technology, and financial incentives remain central to logistics site selection [10], the literature still lacks an integrated explanation of how these factors interact with spatial and contextual dynamics [11]. Research on e-commerce logistics has mainly focused on last-mile delivery, sustainability, and warehousing [12], yet less attention has been given to how location choices differ by facility type [13] and scale.

Comprehensive studies have addressed logistics placement issues via optimization and MCDM methods; yet, the majority regard logistics networks as homogeneous systems at a singular spatial scale. This disregards the unique spatial dynamics of e-commerce, wherein macro-scale fulfillment centers, meso-scale transfer hubs, and micro-scale delivery units function under varying economic and social restrictions. Fuzzy logic-based multi-criteria decision-making systems have demonstrated efficacy in managing uncertainty in expert assessments but are still underutilized in scale-sensitive logistics analysis.

Recent methodological advances demonstrate their potential: fuzzy-based decision models have advanced considerably in tackling uncertainty within multi-objective and hierarchical problems [14]. Building on this, the present study applies the Fuzzy SAW method within a multi-scale analytical framework to quantify how location criteria priorities vary by facility type. Expert-based qualitative and quantitative assessments are integrated to measure spatial determinants across macro-, meso-, and micro-scales.

Istanbul provides a strong empirical setting as Türkiye’s principal logistics and e-commerce hub. Its dense population, limited land, and dual-continental geography make it ideal for examining spatial, economic, and environmental interactions in logistics design. Findings from the Istanbul case illustrate that this framework can inform planning in other metropolitan areas facing congestion, land scarcity and rapid e-commerce growth. Beyond its spatial implications, e-commerce logistics plays a growing role in promoting sustainable urban development, transforming spatial and environmental dynamics through spatial efficiency, green logistics, and innovative last-mile models that advance SDGs by improving accessibility, lowering emissions, and enhancing urban livability. This study embeds sustainability into the multi-scale framework, linking spatial efficiency with environmental and social objectives to support adaptable and resilient urban logistics planning.

This paper introduces a scale-sensitive approach to logistics facility location using the Fuzzy SAW method, revealing how e-commerce reshapes spatial priorities across macro-, meso-, and micro-levels. The framework offers guidance for planners and policymakers by identifying which criteria remain stable and which vary by context, advancing sustainable and evidence-based urban logistics planning.

The remainder of the paper is structured as follows: Section 2 examines the literature and conceptual framework; Section 3 evaluates Istanbul’s logistics context; Section 4 outlines the methodology; Section 5 presents findings; Section 6 discusses implications for urban policy; and Section 7 concludes the study.

2. Towards a Multi-Scalar Logic: Revisiting Facility Location Theories in the Age of E-Commerce

The location of logistics facilities has been a central theme in the literature, with early contributions by Owen and Daskin [15] and later by Farahani and Hekmatfar [16], reflecting both its strategic importance and the inherent difficulties of balancing costs, accessibility, and land use conflicts. Fundamentally, it embodies a decision-making process that reconciles stakeholder interests with spatial and environmental constraints. The study of facility location has a longstanding history in logistics and operations research; yet, the dynamics of urban environments and the advent of new economic and technological factors continue to reshape its significance and complexity. The transformation of the logistics real estate sector, increasingly dominated by global firms operating through multi-scalar distribution networks, has been emphasized by Hesse and Rodrigue [17]. Building on this perspective, numerous studies have examined the location characteristics of logistics facilities, highlighting factors such as access to larger and more affordable suburban parcels and proximity to highway networks and airports [17,18] as well as the expansion of the logistics industry driven by globalization and new production and distribution dynamics [11,19,20,21]. The literature about the locational challenges of logistics facilities in urban areas predominantly revolves around the concept of logistics sprawl, examining the site selection of logistics warehouses and their evolution over time. The location selection of such warehouses is based on a complex supply chain cost structure and numerous factors, including transportation and accessibility, distribution activities, regional economic conditions, facility equipment, land and real estate constraints, and the organization of logistics flows such as last-mile operations [1].

In recent years, logistics warehouses have been shaped not only by these classic cost and access factors but also by structural transformations occurring on a global scale. The globalization of the supply chain, the rise in integrated global companies, and the rapid growth of e-commerce courier services are changing the spatial patterns of warehouses, differentiating their relocation dynamics according to the facility’s scale, type, and function [8]. In this context, new studies examining the impact of last-mile facilities on logistics sprawl have also been added to the literature [2]. E-commerce’s unique operational structure and its impact on global supply and production chains are leading to the further decentralization of distribution networks. This network, which combines large fulfillment and sortation centers with smaller facilities such as delivery stations, urban centers, and consolidation centers, aims to reduce delivery times [22]. Xiao et al. [10] demonstrate that the swift expansion of e-commerce in Shenzhen has resulted in a structural incongruity between conventional warehouse frameworks and contemporary e-retail demands, prompting the emergence of multi-story facilities, the conversion of obsolete industrial zones, and the regional relocation of significant logistics hubs.

Similarly, the location preferences of logistics facilities can be defined by some key characteristics. These include lower local market dependence compared to retail businesses and lower capital intensity compared to manufacturing [23], shorter planning horizons thanks to third-party logistics services, and a shift from ownership to leasing. Furthermore, low land costs, highway and airport connections, labor access, and supportive regulatory conditions are also determining factors in location selection [23,24]. Nonetheless, empirical information regarding the variation in these preferences by facility type or their temporal variations remains scarce [13].

Schorung et al. [8] expand this viewpoint by examining Amazon’s logistics network across various facility types and sizes, uncovering a dual spatial strategy: substantial suburban and exurban fulfillment or cross-dock centers clustered around significant transport infrastructures, alongside a dense network of smaller urban delivery stations and Prime hubs addressing last-mile demand. Their findings emphasize how facility size influences location patterns and requirements, indicating that large warehouses prefer peripheral land availability, whereas tiny urban locations are concentrated in dense places to minimize delivery times and attain economies of scale.

Researchers have employed several approaches and decision-making frameworks to address the issues of selecting sites for logistics hubs. Multi-criteria weighting allows for the comparison of alternatives by determining the importance of different criteria in the decision-making process. The numerical weight assigned to each criterion reflects its relative importance in the decision-making process, enabling a multifaceted evaluation. A key element of MCDM approaches is uncovering stakeholder preferences through scoring and weighting processes.

Various researchers have employed different approaches and analytical frameworks to investigate logistics center location selection.

Table 1. Summary of previous research on logistics facility location selection methods.

Year	Authors	Facility Type/Scope	Case Study/Scale	Method Used
2010	Awasthi et al. [25]	Urban Distribution Center	Single city–urban scenario	Fuzzy TOPSIS (MCDM)
2016	Cristea et al. [26]	Logistics center	Romania	ELECTRE III (MCDM)
2019	Büyüközkan & Mukul [27]	Smart city logistics solutions	Istanbul, Türkiye	Fuzzy Weighted Sum Method
2019	Özmen & Aydoğan [28]	Logistics center	Kayseri, Türkiye	BWM–EDAS (hybrid MCDM)
2020	Novotná et al. [29]	Micro-hub (last-mile logistics)	Pardubice, Czechia	BWM–CRITIC–WASPAS (hybrid MCDM)
2021	Xiao et al. [10]	E-commerce logistics facilities	Shenzhen, China	GIS-based spatial statistics
2023	Feng et al. [30]	Emergency logistics center	Xi’an, China	Entropy + CRITIC + VIKOR (MCDM)
2023	Wang et al. [5]	Urban logistics center	Beijing, China	Multi-objective shortest path model + MCDM
2023	Gonzalez et al. [31]	Sustainable last-mile logistics solutions	6 European cities	STAR Logistics Methodology (REMBRANDT + Delphi, MCDM)
2024	Pajić et al. [32]	Strategic warehouse	Belgrade, Serbia	IMF–SWARA + MARCOS (MCDM)
2024	Boggio-Marzet et al. [33]	Urban last-mile logistics (policy)	European cities	Multi-actor Multi-criteria Evaluation (MAMCA)
2024	Sakai et al. [7]	E-commerce logistics facilities	Tokyo region	Spatial distribution analysis; location preference model
2024	Schorung et. al. [8]	E-commerce logistics facilities	United States	GIS-based spatial analysis
2025	Chen et al. [34]	Urban logistics center	Seoul, South Korea	AHP + TOPSIS (MCDM)

presents a summary of pertinent studies, detailing authors, publication years, methodologies, and facility types to elucidate the contemporary research landscape.

MCDM methodologies, including the Analytical Hierarchy Process (AHP), the Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS), the Preference Ranking Organization Method for Enrichment Evaluations (PROMETHEE), and the Best-Worst Method (BWM), have been extensively utilized in logistics location studies. Each of these methodologies possesses distinct advantages: AHP facilitates the hierarchical organization of intricate issues and is effective for pairwise comparisons; TOPSIS evaluates alternatives based on their proximity to the optimal solution; and BWM reduces discrepancies in expert evaluations. Nonetheless, these models typically presume exact and deterministic inputs, hence constraining their relevance in contexts where expert opinions encompass ambiguity and subjective assessments [5,7,29]. The Fuzzy SAW method provides a more adaptable framework for assessing uncertain and imprecise data, enabling the representation of expert opinions via linguistic variables and membership functions [34]. In comparison to other fuzzy-based MCDM methodologies (e.g., Fuzzy TOPSIS or Fuzzy AHP), the Fuzzy SAW method is computationally more straightforward, necessitates fewer steps for consistency verification, and possesses a transparent weighting framework that facilitates result interpretation. Due to these characteristics, the Fuzzy SAW approach serves as an exceptionally appropriate instrument for assessing logistics facilities across many scales, particularly when several experts assess criteria with varying degrees of uncertainty. This study employs the Fuzzy SAW method to assess the relative significance of spatial, infrastructural, and socio-environmental elements across various facility scales.

Building on these methodological insights, important conceptual gaps remain regarding how spatial and scale-dependent dynamics are incorporated into logistics facility location analysis. Despite the expanding body of research, the majority of studies continue to approach the location problem from a singular scale viewpoint and do not systematically illustrate how the priority of locational criteria changes across different facility types and sizes. While the advantages of flexibility under uncertainty and multi-criteria decision-making are emphasized, the integration of multi-scale realities and spatially explicit frameworks has not been sufficiently developed. This underscores the necessity to scientifically test how location criterion priority alter for facilities of different scales and purposes. This methodological rationale also provides a foundation for integrating sustainability perspectives into spatial analyses of logistics systems.

Beyond operational improvements, research highlights the increasing need of incorporating sustainability concepts into the design of urban and e-commerce logistics. The research concentrates on spatial efficiency, advancements in last-mile delivery, and the mitigation of environmental impacts [35,36,37,38,39]. Sustainable last-mile solutions, including cargo bikes, electric fleets, delivery lockers, and crowdshipping, have demonstrated a substantial reduction in road congestion and CO₂ emissions, alongside enhanced service efficiency [40,41,42]. These advancements underscore the necessity for logistics networks that incorporate accessibility, land use optimization, and ecological sustainability. In addition to operational enhancements, scholars highlight that sustainable urban logistics and facility site design are essential for promoting the Sustainable Development Goals (SDGs) through the integration of spatial efficiency, environmental performance, and social equality. Spatial efficiency in logistics facility planning is attained by optimizing facility locations to improve accessibility and network performance while reducing congestion and energy use [43,44]. The enhancement of environmental performance is achieved by the adoption of eco-friendly transportation methods and the implementation of route optimization tactics that mitigate emissions and noise pollution [45,46,47]. The frequently neglected aspect of social equality necessitates the assurance of equitable access to logistical services and the mitigation of spatial disparities [48,49]. Collectively, these insights provide a conceptual basis for the proposed framework, demonstrating how sustainable logistics facility planning integrates economic efficiency with environmental stewardship and social equity principles that correspond directly with SDG 8 (Decent Work and Economic Growth), SDG 10 (Reduced Inequalities), SDG 11 (Sustainable Cities and Communities), and SDG 13 (Climate Action).

Building on this understanding, the proposed framework embeds sustainability into a multi-scalar analytical structure, linking spatial efficiency with environmental and social goals to foster adaptive and equitable urban logistics planning.

Scale-sensitive analysis is crucial in Türkiye, where swift urbanization and growing e-commerce exert heightened demand on logistics systems, particularly in Istanbul, a vital trade and logistics hub of the nation. Official sources underscore Istanbul’s pivotal position in the national e-commerce ecosystem; yet, no scholarly research has investigated the city’s logistics dynamics based on facility scale and type. As one of Türkiye’s most important logistics centers, Istanbul combines high e-commerce demand, a dense population, and limited land supply, making it an exemplary case of the planning challenges faced by large metropolitan areas and the reason it was chosen as the case study. This research provides scale-sensitive and analytical insights to decision-makers in urban logistics planning and contributes to global discussions of e-commerce-focused spatial restructuring of logistic facilities.

3. Overview of Istanbul’s E-Commerce Logistic Market

This study examines Istanbul, the largest metropolitan area in Türkiye, where e-commerce and logistics industries are concentrated. Istanbul has a critical importance in national and international logistics networks due to its population exceeding 15 million and its geographical location. According to the e-commerce reports published by the Republic of Türkiye Ministry of Trade, Türkiye’s e-commerce volume in 2023 reached 1.85 trillion Turkish liras (≈56 billion USD), an increase of 115.15 percent compared to the previous year, and the same report stated that Istanbul became the leading province in terms of e-commerce logistics activities in 2023, accounting for 55.27 percent of all cargo shipments in Türkiye. It also became the largest recipient of cargo deliveries, receiving 29.74 percent of all incoming shipments.

The city, which is a complex urban settlement with high population density, land use diversity, natural and historical areas, industrial and production infrastructure, and multi-layered transportation connections, presents both challenges and opportunities for e-commerce logistics. At the same time, the presence of many e-commerce marketplaces and third-party logistics providers makes Istanbul an ideal study area for examining spatial location selection criteria at different planning scales.

Typologies and Operational Flows of E-Commerce Logistic Facilities

In order to base the location selection criteria on a case study and to perceive the spatial structure and operational flow of logistics, interviews were conducted with seven logistic experts from three major e-marketplaces and four cargo/delivery service providers in Istanbul. In-depth interviews provided important insights into the operational flow, facility typologies, spatial distributions and decision-making logic of logistics for e-commerce in Istanbul.

These findings clarified the functioning of e-commerce logistics as a separate subsystem within urban logistics, defined by its unique facility hierarchy and operational dynamics. This article defines “e-commerce logistics” as the specialized networks developed by e-marketplaces through the establishment of their own facilities (such as fulfillment and transfer centers) and the expanding networks of cargo and postal companies, which are increasingly accessible to the public due to the surge in e-commerce volume. Although most small and medium-sized sellers in Türkiye continue to utilize ordinary shipping carriers for delivery, larger entities as e-marketplaces are establishing their own logistics hubs at both macro- and meso-levels. This study examines the geographical effects of e-commerce in Istanbul by evaluating the co-evolution of both structures.

E-marketplaces operate their own distribution centers and large-scale warehouses, these areas are located in transportation corridors with wide land availability close to the provincial border of Istanbul and close to intercity roads. These macro-level facilities are responsible for storage, order picking and preparation. The operational typology of cargo companies begins at the medium scale. Similarly, e-marketplaces conduct their medium-scale operations through cargo companies, resulting in similar operational patterns. Companies generally choose points where they can find dense and suitable space between the city center and its peripheries. These centers function as sorting, transfer, distribution and collection centers between regional distribution and last mile delivery. At the micro-scale, there are mini hubs and cargo branches, which function as sorting, transfer, distribution and collection centers, but at the neighborhood scale, they are spread to all points of the city. The synthesis of these operational flows obtained from experts, has provided important insights into understanding the types and locations of facilities in Istanbul’s urban logistics ecosystem.

Table 2. Operational characteristics and facility types of firms represented by the experts.

Experts	Type of Experts’ Firms	Operations	Operational Facilities
Expert 1	E-Marketplace	Fulfillment, Sorting	Fulfillment Center, Large-Scale Warehouse
Expert 2	E-Marketplace and Cargo Transportation	Fulfillment, Sorting, Transfer, Distribution, Collection	Fulfillment Center, Large-Scale Warehouse, Transfer Center, Cargo Branches, Delivery Points
Expert 3	E-Marketplace and Cargo Transportation	Fulfillment, Sorting, Transfer, Distribution, Collection	Fulfillment Center, Large-Scale Warehouse, Transfer Center, Cargo Branches, Delivery Points
Experts 4, 5, 6, 7	Cargo Transportation	Sorting, Transfer, Distribution, Collection	Transfer Center, Mini Hub, Cargo Branches

presents the operational activities conducted by the interviewed e-marketplaces and cargo companies, along with the types of logistics facility units where these activities occur.

There are three ways of product flow in e-marketplaces; the first is the products that the e-marketplace keeps in its own warehouses and purchases, the second is the products that contracted product sellers put in the warehouses of the e-marketplaces, the operational flow of these products is as in Figure 1.

Shipments may be sorted by machines in fulfillment centers and subsequently loaded into vehicles, contingent upon their destination. Conversely, unsorted goods are dispatched immediately to exit transfer centers using cargo company trucks, where they are sorted and subsequently moved to arrival centers by smaller vans. Here, they are sorted again and transferred to the cargo branches closest to the delivery point, and then distribution is made from the cargo branches to the delivery points.

Thirdly, product sellers are only present in the e-marketplace as a virtual store and they send the products to the buyer themselves via cargo companies, as in Figure 2.

Cargo companies collect deliveries from businesses of all sizes, come together at the nearest cargo company, where they are sorted according to the exit transfer center they will go to, products arriving at the transfer center are sorted again according to the exit transfer center, products arriving at the exit transfer center are sorted again here and then delivered to the cargo branch, and from there they are distributed to the delivery point and the shipment is delivered.

The scopes that the logistics units determined in Figure 3 provide services in the city according to their type and area sizes are visualized by scaling. Accordingly, while fulfillment centers are large warehouses on the macro-scale, transfer/sorting centers are located on the meso-scale, and units that are mini hubs and cargo branches are located on the micro-scale.

This basic structure is also preserved in the operational flow specific to Istanbul. Due to the e-commerce shopping volume of Istanbul; large-scale fulfillment centers, warehouses of e-marketplaces and mostly production or storage areas of large manufacturing companies are located in Istanbul and its borders.

These operational diagrams not only reflects the current logistics structure of Istanbul, but also provides an empirical basis for evaluating the level of relevance and weight of location selection criteria at different spatial scales.

Figure 4 includes all fulfillment centers, large-scale warehouses, and transfer centers located in Istanbul for the major e-marketplace companies interviewed for the study; it also includes the distribution of cargo branches, mini-hubs, and transfer centers for the two major cargo companies interviewed. Cargo companies do not have fulfillment centers due to their operational structure. A comprehensive database of facilities for all cargo companies operating in Istanbul is unavailable; however, data gathered from expert interviews with representatives of e-marketplaces and cargo companies, along with Google Maps bookmarks, offers a qualitative framework that illustrates the spatial organization based on facility typologies.

The existing land use structure in Istanbul is a key determinant of the spatial distribution of logistics facility types throughout the city. The city is situated between two continents and has limited residential areas due to its topography divided by the Bosphorus straits, high slopes, and water basins and forests in the north. In the southern part of the city, a significant portion of industrial and logistical areas are located along the main east–west routes. One of the fulfillment centers is situated along the main intercity artery at the eastern periphery of the city; the other two centers are positioned along the same corridor in accessible locations outside the city, near the Istanbul province boundary. The large-scale warehouses indicated on the map represent only the facilities of the interviewed e-marketplace companies and are located near or adjacent to main arteries and industrial areas. Transfer centers are often in highly accessible locations and linked to main arteries; these centers are predominantly found at the western edge of the city and along the eastern provincial boundary. Cargo branches are increasingly prevalent in central regions characterized by significant residential, commercial, and population density.

Experts have highlighted multilayered problems such as the displacement of logistics activities to the city’s peripheries due to rising rents and land pressures, the conversion of existing logistics areas to residential functions, the uncontrolled increase in branching, parking shortages, accessibility challenges to labor, and social unrest at the neighborhood level. This demonstrates that the rapid growth of e-commerce is creating new spatial pressures and planning challenges within the urban fabric. The Istanbul example demonstrates that logistics is no longer merely an economic activity but a multi-scale planning issue affecting the organization of urban space. This context forms the fundamental basis of this study and reinforces the necessity of a scale-sensitive analytical approach to understand the location trends of logistics facilities at different scale.

In conclusion, Istanbul’s e-commerce-based logistics system exhibits a hierarchical structure between macro-, meso-, and micro-scales. Fulfillment centers and large warehouses located on the city’s peripheries are supported by transfer centers near main arteries; this network is complemented by mini-hubs and cargo branches located within the urban fabric. The resulting spatial structure thus concretely reflects the scale-sensitive approach developed in this study, spatially revealing the functional relationships between different facility typologies within the city and the urban periphery. Evaluating this structure from a sustainable urban planning perspective is critical for both increasing logistics efficiency and developing a balanced urban logistics strategy aligned with environmental protection and quality of life objectives.

4. The Weighting Framework: Location Priorities Across Facility Scales in E-Commerce Logistics

This section outlines the methodological approach employed in the investigation. The research elucidates the integration of expert perspectives, literature findings, and fuzzy multi-criteria assessment to evaluate the spatial priorities of logistical facilities across various scales. Figure 5 illustrates the logical progression of procedures involved in ascertaining, assessing, and interpreting the site priorities of e-commerce logistics facilities.

This section also provides a detailed explanation of the expert-based assessment process, criteria determination and weighting methods, and adaptations to the Fuzzy SAW approach.

4.1. Expert Profile and Data Collection

Expert selection was conducted to reflect comprehensive knowledge across various scales and typologies of e-commerce logistics operations. Seven logistics experts from three prominent e-commerce marketplaces and four cargo/distribution service providers in Istanbul participated in the study. The main selection criterion required each expert to possess a minimum of ten years of professional expertise in logistics planning, operational management, or strategic decision-making relevant to urban and e-commerce logistics. Experts were chosen to represent all tiers of logistics operations. Prior to commencing the evaluation procedure, all experts were informed about the study’s objectives, the assessment scale, and the application of fuzzy linguistic variables. A pilot study was executed with two experts to evaluate the intelligibility of the survey instrument and the applicability of the fuzzy scale. Following the pilot, essential modifications were implemented, and comprehensive in-person interviews were carried out to ascertain the primary determinants and contextual factors affecting the placement of logistical facilities. The preliminary findings, along with a synthesis of the literature, were utilized to establish a definitive set of criteria and incorporated into the fuzzy assessment process. Data collection was conducted between January and May 2024 through a combination of face-to-face interviews and online evaluation sessions with the experts.

The survey was conducted with each expert through online sessions. Experts evaluated the significance of each criterion using a five-point linguistic scale from “very low” to “very high.” All expert evaluations were assigned equal significance, as the study aimed to compare location priorities across scales rather than to examine individual opinion variances.

Although seven experts may seem limited in number, this panel size is methodologically substantiated by the literature on expert-based and Fuzzy MCDM applications. Prior research highlights the absence of a universal sample size for expert panels, asserting that methodological validity is contingent upon the diversity, expertise, and consistency of members. In expert-based MCDM and fuzzy MCDM research, a minimum of three experts is deemed adequate, although the prevalent practice typically involves five to ten experts [50,51,52]. The use of fuzzy set theory adeptly captures the uncertainty and subjectivity inherent in expert evaluations, hence diminishing the necessity for extensive samples [52,53]

To mitigate the possible homogenization of viewpoints identified in prior study, the expert group was deliberately diversified to encompass individuals from organizations functioning at various spatial and operational sizes. Although there was considerable commonality among the experts regarding breadth, their positions, organizational hierarchies, and decision-making contexts varied considerably. This enhanced the variety of perspectives and the dependability of the study. Each expert offered insights relevant to their operating scale, enhancing the comparative analysis of facility types.

4.2. Selection of Criteria

The selection of suitable locations for e-commerce logistics facilities requires a multidimensional evaluation process because of the complex interplay between infrastructure conditions, spatial accessibility, socio-environmental factors, and stakeholder considerations. A hierarchical evaluation structure is adopted to analyze location-selection criteria for e-commerce logistics facilities in Istanbul, as summarized in

Table 3. Literature-based location selection criteria for e-commerce logistics facilities.

Criteria	References
C1. Proximity to Main Arteries	[18,21,23,25,54,55,56,57,58,59,60]
C2. Proximity to Intermediate Arteries
C3. Proximity to Railway Routes
C4. Proximity to Airports
C5. Proximity to Ports
C6. Proximity to Transfer and Node Points
C7. Proximity to Residential Areas	[23,25,57,58,61,62,63,64]
C8. Proximity to Commercial Areas
C9. Proximity to Logistics Areas (Logistics Facility + Storage Areas)
C10. Proximity to Industrial Zones
C11. Proximity to Truck, Lorry, Garage Areas
C12. Proximity to Shopping Malls
C13. Proximity to Customs Areas
C14. Proximity to Free Zones
C15. Proximity to Empty Areas (Land Expansion Potential)	[23,58,63,64]
C16. Population Density	[65]
C17. Educational Status of the Population
C18. Income Level of the Population
C19. Age Status of the Population
C.20 Labor Potential	[66,67]
C.21 Land Rental Values	[3,19,21,23,25,54,57,58,59,68]
C.22 Land Purchase Values
C.23 Physical suitability of the settlement area (slope, ground, etc.)	-
C.24 Distance to forest, water basin, agricultural areas
C.25 Resistance to natural disasters risks

. This study established a thorough set of 25 criteria for location selection to assess the spatial rationale of e-commerce logistics facilities in Istanbul. These evaluation criteria were first derived from an extensive review of the relevant literature and then refined through expert consultations to ensure contextual relevance and practical applicability.

In the setting of Istanbul, environmental aspects were used to represent the city’s distinct physical and ecological limitations. Istanbul contains vast forest and water-basin regions necessitating conservation, is situated in active seismic zones, and features high gradients and elevation variations that may impede settlement viability. Consequently, three environment-related criteria physical suitability of the settlement area, distance to forest, water basin, and agricultural areas, as well as resilience to natural catastrophe risks were integrated as author-defined characteristics pertinent to the case study.

4.3. Fuzzy SAW Method

The simple additive weighting method is one of the most widely used MCDM techniques. Its basic approach is based on calculating a weighted sum of the performance scores of each alternative across all criteria. An evaluation score is obtained for each alternative. To achieve this, a normalized decision matrix must be prepared. This normalization process creates a common scale that allows the scores of all alternatives to be compared. The advantage of this method is that the raw data are subjected to a proportional linear transformation, which means that the relative order of magnitudes is preserved in the standardized values [69].

Chou et al. [70] proposed the fuzzy SAW method to solve problems under fuzzy environment. Triangular Fuzzy Numbers (TFNs) were preferred due to their simplicity and effectiveness in representing linguistic uncertainty in expert evaluations. The membership functions employed in this investigation were linear and specified over the interval [0, 1] for each linguistic variable. The steps of the fuzzy SAW method are as follows:

• Step 1. Choosing the criteria to be used as reference in decision-making, namely C_j (j = 1, 2,…, m) and forming a committee of experts E_i (i = 1, 2,…, n) for evaluation.
• Step 2. Experts evaluate the criteria using linguistic terms as defined in

  Table 4. Linguistic scale for Fuzzy SAW.

  S No	Linguistic Variable	Code	Fuzzy Number
  1	Very low	VL	(0, 0, 0.25)
  2	Low	L	(0, 0.25, 0.5)
  3	Medium	M	(0.25, 0.5, 0.75)
  4	High	H	(0.5, 0.75, 1)
  5	Very High	VH	(0.75, 1, 1)

  .
• Step 3. Construct the fuzzy decision matrix DM = [x_ij], where x_ij represents the fuzzy evaluation of criterion C_j by expert E_i.
• Step 4. Determine the aggregated fuzzy score for each criterion using the arithmetic mean of expert evaluations:

  $$A_j = {\displaystyle \frac{\mathrm{x}{1}_j+\mathrm{x}{2}_j+\dots +\mathrm{x}{n}_j}{n}}$$

  (1)

  where j = 1, 2, …, m; n is the number of experts.
• Step 5. Defuzzify each aggregated fuzzy scores by applying the centroid method:

  $$e_j = {\displaystyle \frac{a_j+b_j+c_j}{3}}$$

  (2)

  where ($a_j+b_j+c_j$) are the lower, middle and upper bounds of the triangular fuzzy number $A_j.$
• Step 6. Calculate the normalized weight of each criterion W_j by dividing its defuzzified score by the total of all defuzzified scores:

  $$W_j = {\displaystyle \frac{e_j}{{\sum }_{j = 1}^m{e}_j}}$$

  (3)

  where $\sum _{j = 1}^m{W}_j = 1$ and the weight vector W = (W₁, W₂, …, W_m) is constructed.

In this study, the classical Fuzzy SAW approach was adapted by integrating expert-based scale differentiation. The weighting method was performed independently at macro-, meso-, and micro-scales to account for scale-dependent differences in criterion significance.

This study utilized an expert-based fuzzy weighting framework rather than a regression-based model, so statistical multicollinearity tests, such as VIF, were not conducted. Rather, conceptual overlaps and any dependencies among the criteria were managed through iterative expert interviews and literature assessment. Such an approach is widely adopted in the fuzzy MCDM literature, in which the complexities of multiple interconnections are conceptually mitigated through hierarchical structuring and carefully expert selection, rather than through direct statistical methods [71,72,73,74,75].

As no statistical model fitting was used in the study, the correlations between criteria do not parametrically affect the results. The dependability of the weighing process is guaranteed by the consistency of expert opinions, iterative validations, and consensus, which constitute the essential mechanism for robustness in fuzzy MCDM systems [73,74].

To further assess the robustness of the fuzzy weighting results, a qualitative sensitivity analysis was conducted. Minor adjustments (±10%) to the criteria weights did not substantially affect the ranking of the criteria or their relative significance across scales. This illustrates the robustness of the fuzzy weighing outcomes and their insensitivity to minor variations in expert assessments.

5. Application and Results

This section outlines the application procedure utilized to investigate the varying significance of logistics site criteria across distinct spatial scales (macro, meso, and micro) within Istanbul’s e-commerce logistics sector.

Initially, in-person, open-ended interviews were performed with senior executives from logistics firms engaged in the e-commerce industry. The discussions concentrated on comprehending the logistics flows, operational frameworks, and site selection preferences that influence e-commerce distribution systems in Istanbul. The participants were all high-level decision-makers with comprehensive knowledge of operational and strategic processes within major e-commerce marketplaces and cargo firms in Türkiye.

Based on insights from these interviews and an extensive review of the related literature, a list of 25 location criteria influencing the spatial organization of e-commerce logistics activities was established. The definitions, scope, and scale-based relevance (macro, meso, micro) of each criterion were explained to the experts. Afterwards, each expert was asked to evaluate the importance of the criteria using a five-point linguistic scale. This procedure generated the data required to analyze how sectoral knowledge and professional judgments align in determining the priorities of location criteria across scales.

The collected linguistic assessments were converted into triangular fuzzy numbers, and the analysis was carried out using the Fuzzy SAW method. The aggregated fuzzy values were subsequently defuzzified to obtain crisp scores, which were then normalized to calculate the final weights of each criterion for the three spatial scales.

• Step 1. A set of 25 criteria was identified through a detailed literature review and expert opinions.
• Step 2. Experts evaluated the importance of each criterion on a five-point linguistic scale in

  Table A1. Expert evaluations of location selection criteria across scales.

  	Macro							Meso							Micro
  	E1	E2	E3	E4	E5	E6	E7	E1	E2	E3	E4	E5	E6	E7	E1	E2	E3	E4	E5	E6	E7
  C1	VH	VH	L	VH	VH	H	H	VH	VH	VH	VH	VH	VH	VH	VH	VH	VH	H	VH	VH	VH
  C2	H	VH	VL	VL	M	VH	VH	H	VH	VL	VL	VH	H	H	H	VH	VL	VH	VH	H	VH
  C3	VL	VL	VL	VL	M	H	H	VL	VL	VL	VL	H	M	M	VL	VL	VL	VL	VH	M	L
  C4	H	VL	VL	VL	M	H	H	H	VL	VL	VL	M	M	M	H	VL	VL	VL	VH	M	L
  C5	VH	VL	VL	VL	M	M	M	H	VL	VL	VL	M	M	M	H	VL	VL	VL	VH	M	L
  C6	H	VL	M	VL	L	M	M	VH	VL	VH	VL	M	H	H	H	VL	VL	VL	VH	H	H
  C7	H	H	VL	VL	M	H	H	H	H	VL	L	H	VH	VH	H	H	VL	L	VH	VH	VH
  C8	VL	H	M	VL	M	M	H	VL	H	M	VL	H	H	H	VL	H	M	L	VH	H	H
  C9	H	H	M	VL	M	M	M	VL	H	M	L	H	H	H	VL	VL	M	L	VH	H	H
  C10	L	VL	VL	VL	M	H	H	VL	VL	VL	VL	H	M	M	VL	VL	VL	VL	VH	M	L
  C11	H	VL	VL	VL	M	VH	VH	H	VL	VL	VL	M	M	M	H	VL	VL	VL	VH	M	L
  C12	M	VH	VH	L	H	VH	VH	H	H	VH	L	H	H	H	H	M	VH	VL	VH	M	L
  C13	VL	VL	VH	L	L	H	VH	VL	VL	VH	L	M	M	H	L	VL	VH	L	VH	M	H
  C14	H	H	M	M	H	M	M	H	H	VL	L	H	M	M	H	M	VL	VL	VH	M	L
  C15	H	M	VL	M	M	H	H	VH	H	VL	M	M	H	H	H	M	VL	VL	VH	M	M
  C16	H	VH	VL	H	VH	VH	VH	M	VH	VL	H	H	VH	VH	H	M	VL	H	VH	VH	VH
  C17	H	VH	VL	H	VH	H	H	M	VH	VL	H	H	VH	VH	H	M	VL	H	VH	VH	VH
  C18	H	VH	VL	M	H	M	M	L	VH	VL	M	H	H	H	H	M	VL	H	VH	H	M
  C19	VL	M	VL	VH	H	M	M	VL	M	VL	M	H	M	M	H	M	VL	M	VH	M	M
  C20	VH	H	VL	H	H	M	M	VH	VH	VL	M	H	VH	H	VH	M	VL	H	VH	VH	H
  C21	VH	H	M	VH	H	H	H	VH	VH	M	M	H	VH	H	VH	M	VL	L	VH	VH	VH
  C22	H	H	VL	M	H	M	VL	VH	M	VL	M	H	VH	H	H	M	VL	L	VH	VH	VH
  C23	H	VL	VL	M	H	H	M	H	M	VL	M	H	M	M	H	M	VL	M	VH	M	M
  C24	H	VL	VL	M	M	L	M	H	M	VL	M	M	M	M	H	M	VL	M	VH	M	M
  C25	H	L	VL	M	M	M	M	H	M	VL	M	M	M	M	H	M	VL	M	VH	M	M

  . These linguistic assessments were subsequently converted into triangular fuzzy numbers to construct the fuzzy decision matrix.
• Step 3. Equations (1)–(3) are applied, and the aggregated values (Agg_a, Agg_b, Agg_c), defuzzified weights (DW), and normalized weights (NW) are calculated for each criterion within each scale, as shown in

  Table A2. Aggregated and weighted values of location selection criteria across scales.

  	Macro					Meso					Micro
  	Agg_a	Agg_b	Agg_c	DW	NW	Agg_a	Agg_b	Agg_c	DW	NW	Agg_a	Agg_b	Agg_c	DW	NW
  C1	4.00	5.75	6.50	5.42	0.058	5.25	7.00	7.00	6.42	0.067	5.00	6.75	7.00	6.25	0.066
  C2	3.00	4.25	5.25	4.17	0.045	3.00	4.25	5.50	4.25	0.045	4.00	5.50	6.25	5.25	0.055
  C3	1.25	2.00	3.75	2.33	0.025	1.00	1.75	3.50	2.08	0.022	1.00	1.75	3.25	2.00	0.021
  C4	1.75	2.75	4.50	3.00	0.032	1.25	2.25	4.00	2.50	0.026	1.50	2.50	4.00	2.67	0.028
  C5	1.50	2.50	4.00	2.67	0.029	1.25	2.25	4.00	2.50	0.026	1.50	2.50	4.00	2.67	0.028
  C6	1.25	2.50	4.25	2.67	0.029	2.75	4.00	5.25	4.00	0.042	2.25	3.25	4.75	3.42	0.036
  C7	2.25	3.50	5.25	3.67	0.040	3.00	4.50	5.75	4.42	0.046	3.25	4.75	5.75	4.58	0.048
  C8	1.75	3.00	4.75	3.17	0.034	2.25	3.50	5.25	3.67	0.038	2.50	4.00	5.50	4.00	0.042
  C9	2.00	3.50	5.25	3.58	0.039	2.25	3.75	5.50	3.83	0.040	2.00	3.25	4.75	3.33	0.035
  C10	1.25	2.25	4.00	2.50	0.027	1.00	1.75	3.50	2.08	0.022	1.00	1.75	3.25	2.00	0.021
  C11	2.25	3.25	4.50	3.33	0.036	1.25	2.25	4.00	2.50	0.026	1.50	2.50	4.00	2.67	0.028
  C12	3.75	5.50	6.25	5.17	0.056	3.25	5.00	6.50	4.92	0.052	2.50	4.00	5.25	3.92	0.041
  C13	2.00	3.25	4.50	3.25	0.035	1.75	3.00	4.50	3.08	0.032	2.25	3.75	5.00	3.67	0.038
  C14	2.50	4.25	6.00	4.25	0.046	2.00	3.50	5.25	3.58	0.038	1.75	3.00	4.50	3.08	0.032
  C15	2.25	3.75	5.50	3.83	0.041	2.75	4.25	5.75	4.25	0.045	2.00	3.25	4.75	3.33	0.035
  C16	4.00	5.50	6.25	5.25	0.057	3.50	5.00	6.00	4.83	0.051	3.50	5.00	6.00	4.83	0.051
  C17	3.50	5.00	6.25	4.92	0.053	3.50	5.00	6.00	4.83	0.051	3.50	5.00	6.00	4.83	0.051
  C18	2.50	4.00	5.50	4.00	0.043	2.50	4.00	5.50	4.00	0.042	2.75	4.25	5.75	4.25	0.045
  C19	2.00	3.25	4.75	3.33	0.036	1.50	2.75	4.50	2.92	0.031	2.25	3.75	5.25	3.75	0.039
  C20	2.75	4.25	5.75	4.25	0.046	3.50	5.00	6.00	4.83	0.051	3.50	5.00	6.00	4.83	0.051
  C21	3.75	5.50	6.75	5.33	0.058	3.75	5.50	6.50	5.25	0.055	3.25	4.75	5.50	4.50	0.047
  C22	2.00	3.25	5.00	3.42	0.037	3.00	4.50	5.75	4.42	0.046	3.00	4.50	5.50	4.33	0.045
  C23	2.00	3.25	5.00	3.42	0.037	2.00	3.50	5.25	3.58	0.038	2.25	3.75	5.25	3.75	0.039
  C24	1.25	2.50	4.25	2.67	0.029	1.75	3.25	5.00	3.33	0.035	2.25	3.75	5.25	3.75	0.039
  C25	1.50	3.00	4.75	3.08	0.033	1.75	3.25	5.00	3.33	0.035	2.25	3.75	5.25	3.75	0.039

  .
• Step 4. In order to obtain the final ranking values, the DW and NW of each criterion were multiplied, as follows: F_i = DW_i × NW_i

This allows the model to account for both the experts’ fuzzy-based importance assessments and the relative significance of each criterion within the overall evaluation framework.

The rank ordering of all criteria within each scale (from 1 to 25) is provided in

Table 5. Final ranking of location selection criteria across scales.

Rank/ Scale	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19	20	21	22	23	24	25
Macro	C1	C21	C16	C12	C17	C14	C20	C2	C18	C15	C7	C9	C22	C23	C11	C19	C13	C8	C25	C4	C5	C6	C24	C10	C3
	0.317	0.307	0.297	0.288	0.261	0.195	0.195	0.187	0.173	0.159	0.145	0.139	0.126	0.126	0.120	0.120	0.114	0.108	0.103	0.097	0.077	0.077	0.077	0.067	0.059
Meso	C1	C21	C12	C16	C17	C20	C7	C22	C2	C15	C6	C18	C9	C8	C14	C23	C24	C25	C13	C19	C4	C5	C11	C3	C10
	0.431	0.289	0.253	0.245	0.245	0.245	0.204	0.204	0.189	0.189	0.168	0.168	0.154	0.141	0.135	0.135	0.116	0.116	0.100	0.089	0.065	0.065	0.065	0.045	0.045
Micro	C1	C2	C16	C17	C20	C7	C21	C22	C18	C8	C12	C19	C23	C24	C25	C13	C6	C9	C15	C14	C4	C5	C11	C3	C10
	0.409	0.289	0.245	0.245	0.245	0.220	0.212	0.197	0.189	0.168	0.161	0.147	0.147	0.147	0.147	0.141	0.122	0.116	0.116	0.100	0.075	0.075	0.075	0.042	0.042

, showing the differences in relative importance across scales. At the macro-scale, proximity to main arteries (C1), land rental values (C21), and population density (C16) emerged as the most significant factors. At the meso-scale, proximity to main arteries (C1) remains the dominant criterion, followed by land rental values (C21) and proximity to shopping malls (C12). At the micro-scale, proximity to main arteries (C1) again takes the lead, while proximity to intermediate arteries (C2) and population density (C16) rank second and third, respectively. In contrast, criteria such as proximity to railway routes (C3) and proximity to industrial zones (C10) consistently occupy the lowest ranks across scales, suggesting their limited influence on e-commerce logistics facility location decisions within the Istanbul context.

Figure 6 provides a comparative visualization of the inter-scale normalized weights for all criteria. The results confirm the internal consistency of the rankings presented in Table 5 and show that accessibility and socioeconomic factors have varying degrees of importance across facility scales within Istanbul’s e-commerce logistics network. The following section discusses these findings in relation to the evolving spatial logic of e-commerce logistics and its implications for urban planning and sustainability.

6. Discussion

This study’s findings provide empirical evidence that the location priorities of e-commerce logistics facilities vary significantly across facility scales, confirming that logistics networks operate through multi-scalar spatial logics rather than a single optimization principle. This discovery directly supports the arguments of Hesse and Rodrigue [17], who emphasized that logistics geography is increasingly multi-scalar as a result of the globalization of supply chains and the diversity of logistics facility types. Schorung et al. [8] have shown that global organizations like Amazon adopt a dual spatial approach consisting of large peripheral supply hubs and smaller urban delivery hubs. In the case of Istanbul, these findings align with the variable interaction of accessibility, land cost, and population dynamics at each stage.

At the macro-level, the dominance of proximity to main arteries (C1) and land rent values (C21) highlights the enduring nature of traditional cost-accessibility tradeoffs, as articulated by Dablanc and Ross [19] and Andreoli et al. [21]. Large logistics centers in Istanbul are clustered around peripheral highways and intercity corridors, as Xiao et al. [10] found in Shenzhen, where increasing urban land pressure and the need for spatial expansion are driving the relocation of large-scale logistics infrastructure to suburban or regional areas. The concomitant importance of population density (C16) demonstrates that, even at the macro-level, e-commerce logistics is linked not only to production or distribution networks but also to consumer markets, marking a significant departure from traditional industrial logistics models as seen by Kang [13]. This hybrid logic, combining demand proximity with regional connectivity, demonstrates how e-commerce is transforming the “core-periphery” dynamics of logistics in urban areas [9].

The mesoscale represents a transitional layer where accessibility dominates but interacts more intensively with urban land use and commercial activities. The importance of population density (C16), proximity to shopping malls (C12) and proximity to intermediate arteries (C2) at this level aligns with Wang et al. [5], who emphasize the increasing connectivity between retail and logistics environments in densely populated urban areas. Thus, meso-scale facilities hinge on network efficiency and access to consumption corridors. Furthermore, environmental and demographic factors (e.g., C18, C20) gain importance at this scale, supporting Sakai et al. [7] argument that sustainable e-commerce logistics increasingly depends on integrating social and environmental dimensions into planning decisions. The meso-level priorities identified in Istanbul exemplify a balanced approach between cost-effectiveness and urban cohesion, supporting the multi-actor perspective articulated by Boggio-Marzet et al. [33]. This facilitates the co-location of urban consolidation centers with mixed-use corridors and intermodal nodes to reduce empty trips and search for parking spaces.

The findings highlight the growing importance of logistics at the neighborhood level at the micro-scale, where factors such as proximity to intermediate arteries (C2), labor potential (C20), and educational status of the population (C17) emerge as critical variables. This shift toward social and operational proximity aligns with the concept of “logistical urbanism” as articulated by Fried and Goodchild [2], which conceives of micro-centers and distribution branches as core urban functions rather than peripheral industrial activities. The increasing importance of socio-demographic factors implies that logistics systems must adapt to local conditions and be tailored to accommodate the need for employee accessibility, customer proximity, and rapid and reliable service delivery. Risberg [12] similarly emphasizes that socially responsive logistics must align operational efficiency with urban livability; this dynamic is particularly evident in densely populated areas of Istanbul. The focus on people-centered criteria at the micro-scale demonstrates a shift in urban logistics from spatial optimization to service flexibility. Meanwhile, the low importance of proximity to railway routes (C3) and proximity to industrial zone (C10) at all scales marks a distinct departure from the traditional logistics frameworks established by Owen and Daskin [15] and Farahani and Hekmatfar [16]. As noted by Schorung et al. [8] and Sakai et al. [11], e-commerce logistics is increasingly based on flexible, road dependent networks and decoupled from traditional production clusters.

These patterns collectively point to structural changes in how logistics systems function. Three interrelated mechanisms clarify the evolving logic of e-commerce logistics. Temporal competitiveness has increasingly replaced production efficiency, as delivery time guarantees now outweigh traditional concerns for arterial accessibility (C1, C2), particularly in high-rent urban contexts. At the same time, platform coordination through digital pooling and algorithmic demand balancing has reanchored logistics within neighborhoods rich in both consumers and labor (C16, C17, C20). In parallel, diversification across macro-, meso-, and micro-scales enables firms to mitigate risk and remain responsive to fluctuations in demand.

Rather than replacing classical location theory, these dynamics reinterpret it through a multi-scalar lens that integrates temporal and social dimensions. At the macro-level, cost and accessibility trade-offs (C1, C21) still prevail, consistent with traditional models optimizing fixed assets and trunk corridors. At the meso- and micro-scales, however, temporal proximity and social accessibility dominate, reflecting a relational form of proximity that classical cost and distance models explain inadequately for dense clusters of consumers and workers, even though they remain relevant for large-scale facility siting. E-commerce logistics therefore extends classical theory by embedding geographical rationality within a temporal and socio-digital framework.

The results obtained from the Fuzzy SAW analysis closely match Istanbul’s spatial configuration. At the macro-scale, the high weights assigned to proximity to main arteries (C1) and land rental values (C21) mirror the city’s east–west linear development pattern, where large logistics and industrial facilities are concentrated along the southern transportation corridors. Istanbul’s northern zone (defined by forest and water-basin conservation areas) creates physical constraints for large-scale facilities that require extensive land. At the meso-scale, the growing influence of proximity to commercial areas (C12), population density (C16), and labor potential (C20) coincides with the clustering of transfer centers in mixed-use and densely populated districts. At the micro-scale, human-centered criteria such as proximity to residential areas (C7), labor potential (C20), population density (C16) and educational level (C17) parallel the location of cargo branches and mini-hubs within the urban fabric, particularly in high-demand neighborhoods.

These spatial shifts carry direct sustainability implications: a scale-calibrated redistribution of functions advances SDG 11 (spatial efficiency), SDG 8 (local jobs), and SDG 13 (shorter trips, lower emissions). This sustainability dimension highlights that spatial efficiency and social equity are not separate objectives but interdependent outcomes of e-commerce logistics restructuring. In practice, pairing meso-hubs with mixed-use corridors while deploying micro lockers or bike-based delivery in high-demand districts can curb cruising and missed deliveries. Incorporating scale-sensitive insights into urban planning can enhance the environmental efficacy, social equality, and resilience of e-commerce logistics networks.

Overall, our multi-scalar data necessitates planning instruments that concurrently enhance macro-connectivity, meso-corridor integration, and micro neighborhood accessibility. This indicates a gradual transition from a regional, cost-oriented logistics logic to a people-centered urban logistics paradigm and highlights the need for multi-scalar planning strategies that balance efficiency, equity, and environmental sustainability. Recognizing these spatial dynamics is essential for formulating adaptive and resilient e-commerce logistics networks that can respond to both market demands and urban sustainability goals.

However, it is necessary to acknowledge certain limitations of the study. First of all, expert opinions are inherently subjective and can be affected by the experience and operational constraints of the experts. Moreover, the exclusion of emerging delivery technologies (e.g., drones, autonomous couriers) limits the ability to anticipate future logistics transformations. Another limitation is the lack of data integration regarding ownership and zoning, which are critical for location decisions. Furthermore, the findings are specific to the context of Istanbul, and while the methodological approach can be replicated, it may not be directly applicable to cities with different spatial, economic and institutional structures and planning approaches. In addition, the assessments assume equal decision-making power among logistics actors, i.e., they do not take into account potential power asymmetries.

7. Conclusions

This research operationalizes the core premise that e-commerce logistics functions not under a single optimization logic but through a multi-scalar spatial configuration. By integrating this theoretical framework with the Fuzzy Multi-Criteria Decision-Making (Fuzzy SAW) method, it demonstrates how logistics facility location priorities shift from cost-based efficiency at the macro-scale to socially and service-oriented accessibility at the micro-scale. In doing so, the notion of scale sensitivity was translated into a quantifiable analytical framework through which spatial disparities among facility types were systematically assessed.

Its originality lies in a dual contribution at both the conceptual and methodological levels. Conceptually, the research revisits classical location theory within the context of digitalized, consumer-centric, and time-sensitive logistics networks. Methodologically, it integrates expert judgment within a fuzzy logic framework, allowing the analysis of uncertain and multi-layered urban decision-making processes. This approach provides a transparent and reproducible decision-support tool, advancing both theoretical debates in urban logistics and the practical implementation of spatial planning strategies.

The research process faced several challenges. The integration of heterogeneous data sources within a multi-scalar framework, the limited accessibility of zoning and property data, and the rapidly evolving technological landscape of e-commerce partially constrained the model’s predictive capacity. Nevertheless, these limitations underscore the necessity for adaptive, expert-driven, and updatable tools in future logistics planning and evaluation processes.

Beyond its empirical results, the study contributes a framework for future research and policymaking. The findings emphasize that logistics should be recognized as a core component of urban infrastructure, supporting the development of cross-scalar, sustainable, and socially equitable planning strategies. Future research is expected to test and refine this multi-scalar framework across diverse urban contexts, generating new models that strengthen the spatial sustainability and resilience of logistics systems.