Foresight for Sustainable Last-Mile Delivery: A Delphi-Based Scenario Study for Smart Cities in 2030

Abstract

This study aimed to investigate the future trajectories of last-mile delivery (LMD), and their implications for sustainable urban logistics and smart city planning. Through a Delphi-based scenario analysis targeting the year 2030, this research draws on inputs from a two-round Delphi study with 52 experts representing logistics, academia, and government. Four key thematic areas were explored: consumer demand and behavior, emerging delivery technologies, innovative delivery services, and regulatory frameworks. The projections were structured using fuzzy c-means clustering, and analyzed through the Technology Acceptance Model (TAM) and Innovation Diffusion Theory (IDT), supporting a systemic understanding of innovation adoption in urban logistics systems. The findings offer strategic insights for municipal planners, policymakers, logistics service providers, and e-commerce stakeholders, helping align infrastructure development and regulatory planning with the evolving needs of last-mile logistics. This approach contributes to advancing resilient, low-emission, and inclusive smart city ecosystems that align with global sustainability goals, particularly those outlined in the UN 2030 Agenda for Sustainable Development.

1. Introduction

The sector of last-mile delivery (LMD), which includes express, courier, and emerging startup services, has seen significant expansion driven by the surge in e-commerce. For example, global e-commerce sales are projected to increase nearly fivefold between 2014 and 2024, reaching USD 6388 billion [1]. The COVID-19 pandemic has further accelerated the shift toward online shopping [2], fueling the demand for innovative delivery options that promise fulfillment within 10 min to two hours. Nonetheless, LMD accounts for a substantial portion of shipping expenses, ranging from 41% to 50% [3], and contributes significantly to environmental degradation. The transportation sector globally has seen its carbon dioxide equivalent (CO₂e) emissions rise by roughly 80% from 1990 to 2018, amounting to 8.26 billion tonnes [4]. Road freight traffic is a major contributor to global greenhouse gas emissions, particularly in food logistics, which alone accounts for approximately 4.8% of global emissions [5,6]. These environmental impacts, along with growing concerns about congestion, air quality, noise, and traffic safety in cities, highlight the urgent need for more sustainable LMD systems. In addition to their ecological footprint, LMD operations raise public health concerns related to respiratory illness and accidents in densely populated urban areas.

In the context of LMD for healthcare, it is crucial to address the health risks associated with transportation by ensuring the safe handling and delivery of medical supplies, particularly in cities facing infrastructure strain. Meanwhile, the smart city population is projected to grow from 30% in 1950 to 68% by 2050 [7], amplifying the adverse effects of LMD in high-density environments. This population surge creates mounting pressure on urban healthcare systems and logistics infrastructure, demanding more integrated and sustainable planning approaches.

Logistics service providers (LSPs) are responding with innovations in digitization and e-mobility to enable more sustainable delivery operations. For instance, UPS has become the first LSP to receive comprehensive certification from the U.S. Federal Aviation Administration to operate commercial drone deliveries beyond the visual line of sight [8]. These technological advancements have the potential to transform the LMD landscape. However, the success of such innovations depends heavily on recipient acceptance, which is influenced by trust, convenience, and perceived benefit [9,10,11,12]. LSPs must adapt to shifting customer preferences while maintaining efficiency and affordability in a competitive marketplace. At the same time, healthcare providers and urban planners aim to reduce emissions and environmental risks, aligning logistics strategies with public health and sustainability objectives.

The rise of global e-commerce intensifies this pressure, as local producers now compete with international suppliers. For consumers, the perceived value of an e-commerce offering increasingly depends on the quality, sustainability, and reliability of the delivery service. Studies show that over 80% of customers choose e-commerce sellers based on their LSP’s performance [8]. Security, transparency, and environmental responsibility also play key roles in shaping customer loyalty [13,14]. As a result, LSPs are becoming not only logistical but strategic partners in meeting the expectations of increasingly sustainability-conscious consumers.

Simultaneously, municipalities must develop strategies to maintain efficient traffic movement and reduce urban air pollution. Some cities, such as London, have implemented congestion charges and are exploring dynamic tolling mechanisms based on location and time [15,16,17]. These urban policies reflect the growing need for integrated, forward-looking planning that balances environmental, economic, and social sustainability.

To provide strategic foresight for the LMD sector within this evolving landscape, we conducted a Delphi-based scenario analysis to explore possible futures for LMD by the year 2030. Involving 52 experts across academia, logistics, public policy, and healthcare, this method is well suited for highly dynamic and uncertain environments undergoing technological and regulatory shifts [18,19,20,21]. Considering these trends, this study addresses the following research questions:

Q1: How will consumer needs, public regulation, smart city planning, and healthcare service optimization requirements evolve concerning LMD by 2030?

Q2: What delivery modes and associated technologies will emerge, and what services will LSPs offer, considering the implications for smart city infrastructure and healthcare development?

The remainder of this paper is organized as follows. Section 2 reviews the relevant literature on last-mile delivery and sustainable urban logistics. Section 3 outlines the research methodology, including the Delphi study design and scenario development process. Section 4 presents the results, while Section 5 discusses their implications for smart city planning, healthcare logistics, and sustainability. Section 6 concludes the paper and suggests directions for future research. Section 7 outlines the study’s limitations and identifies further research opportunities to strengthen sustainable last-mile delivery and urban logistics planning.

2. Review of the Literature

2.1. Technology Acceptance Model (TAM) and Innovation Diffusion Theory (IDT)

In the sector of LMD, logistics service providers rely heavily on the utilization of their products and services by customers. The range of products includes not only traditional parcel delivery, but also the transportation of perishables, medication, and temperature-sensitive items. However, the sector is experiencing overcapacity, which is further exacerbated by new competitors like Amazon and Uber [22,23,24]. This situation has intensified competition and increased price pressure [25,26]. Therefore, LSPs need to predict emerging tendencies and customize their emerging delivery technologies as well as services to meet customer-specific demands [3,25]. Success will come to LSPs with the innovative capacity to exceed customer expectations while maintaining profitability.

In our study, we merged the Rogers innovation diffusion theory (IDT) [27] with the technology acceptance model (TAM) developed by Davis et al. [28] to explore the prospects of LMD. Numerous studies have demonstrated the interrelation between these theories (e.g., [9,10,11,12]). For example, Granić and Marangunić [29] integrated these theories in their empirical research, highlighting the benefits of this combined approach. They discovered that attributes of IDT, such as relative advantage, significantly influence TAM dimensions like perceived usefulness. Although IDT and TAM are commonly utilized in IT and business contexts [10,12], our research applies these theories to the sector of LMD, which is formed by client needs, regulatory frameworks, and technological advancements. Consequently, we broadened the range of technology diffusion theories by examining new aspects such as consumer demand and behaviour trends, service diffusion, and regulatory impacts. Our Delphi-based scenario analysis assesses these features to predict the future of the LMD industry, with a specific emphasis on smart city planning implications. Below, we provide a comprehensive introduction to IDT and TAM.

Rogers [27] investigates the “properties of innovation” and their influence on adoption, identifying five key characteristics: compatibility, relative advantage, observability, complexity, and trialability, which is defined as the user experience an innovation offers [9,11], including economic and social benefits such as pricing and/or status. According to Rogers [27], compatibility is “the degree to which an innovation is perceived as consistent with the existing values, past experiences, and needs of potential adopters.” Consumers tend to favor products and/or services that align with their comfort zone to minimize uncertainty. Complexity refers to how easily consumers can comprehend and utilize a new innovation. For example, Apple iPod’s success is often credited to its user-friendly interface and simplicity. Observability assesses how visibly others can see the outcomes of the innovation. Some products or services are noticeable and prompt inquiries, while others may remain unseen. Lastly, trialability refers to the likelihood of consumers adopting innovations after trying them out. As individuals become more familiar with the innovation, their confidence in buying it increases [9,10]. This study broadens the traditional focus from technology to encompass consumer demand and behavior, services, and regulation, while excluding observability and trialability, which fall outside the scope of our research.

Davis et al. [28] introduced the TAM, which is closely linked to and complements the IDT. Some researchers regard it as an aspect of perceived innovation characteristics [11,12]. TAM includes three key components: perceived ease of use, perceived usefulness, and behavioural intention to use. Perceived usefulness is the belief that a product or service will improve performance, such as text processing software that translates content efficiently and precisely. Perceived ease of use refers to the amount of effort required to utilize the product or service [28]. Behavioral intention to use is shaped by perceived usefulness and ease of use, with high evaluations in these areas leading to a strong intention to adopt the product or service.

2.2. Delivery of the Last Mile

Last mile is the concluding phase of the delivery process, involving the transportation of packages from a distribution center, like a warehouse or sorting facility, to the final recipient. Current trends in digitization, sustainability, and customer-centric approaches add further complexity for LSPs. Key LMD factors such as cost efficiency, time management, and quality present opportunities for ongoing optimization for LSPs [30,31,32]. Smart city planners must consider these trends and impacts to create supportive infrastructures and policies that facilitate efficient and sustainable last-mile deliveries. Smart city planners must consider these trends and impacts to create supportive infrastructures and policies that facilitate efficient and sustainable last-mile deliveries.

In the context of LMD, where profit margins are usually narrow, operational efficiency becomes critical. LSPs can achieve cost efficiency, or the minimum cost for shipping services, through three primary approaches. Firstly, optimizing the number of deliveries made at each stop, while reducing fuel consumption, delivery-time, and travel distance is vital. Since labor costs are the highest expense in LMD, effective management of personnel is crucial [33,34,35,36]. Innovations such as crowd-shipping have been introduced to give resources and cut expenses [37].

Secondly, digitization facilitates increased automation with the use of autonomous delivery vehicles, drones, Mobile Parcel Lockers (MPLs), and delivery robots, which can lower labor costs and improve efficiency. Mobile Parcel Lockers (MPLs) are transportable units that can be relocated to serve dynamic delivery points, whereas Smart Parcel Lockers (SPLs) are fixed, technology-enabled installations providing secure, automated parcel collection at designated locations. This distinction is important for understanding their respective roles in adaptive last-mile delivery strategies. Studies on smart parcel lockers (SPLs) and MPLs further support these technological advancements [33,34,35,36]. Automation offers practical solutions, especially with an ageing workforce [33,34,35,36].

Thirdly, AI can optimize routing and in-house processes at sorting centres. AI-enhanced routing, combined with autonomous vehicles, improves delivery efficiency by adapting to real-time traffic and customer demands. Enhanced route planning, incorporating weather data, is one example. Advanced robots, like those from Boston Dynamics, significantly outperform existing warehouse robots and transform parcel handling [37], presenting an opportunity for AI-driven home delivery automation.

Speed is an essential aspect of LMD. Quick delivery times have emerged as a crucial determinant for both B2B and B2C shipments. In regional areas, over 90% of parcels are successfully delivered to their destinations by the following day [38,39]. The growing volume of parcels is creating time pressures in the sector of LMD [34,36]. Many e-commerce sellers now offer same-day or next-day delivery, further increasing these time constraints. LSPs can distinguish their offerings by providing various delivery speeds, from express to standard, and by being adaptable to the recipients’ needs. Instant delivery services like Jahez and Noon deliver fast-moving consumer goods, groceries, and food in under 10 min [15]. To provide such services, LSPs need to develop highly adaptable and resilient operations, balancing delivery speed with vehicle efficiency. Smart city planning must support these different delivery speeds and ensure infrastructure facilitates rapid and efficient last-mile deliveries.

The quality and safety of LMD are of paramount importance. Beyond reducing the occurrence of lost or damaged parcels, the protection of personal information, such as detailed information on the geographical location and various delivery options available, has become critical. Excellent service performance is directly associated with financial success [40,41]. Samson and Terziovski [42] highlight that a strong customer orientation is a key predictor of organizational performance. Consequently, for logistics service providers (LSPs) to stay competitive, they must uphold high-quality standards, ensuring that delivery services are efficient and meet the expectations of recipients in collaboration with e-commerce retailers [8]. Eco-friendly delivery options can also enhance economic performance and meet customer demands [43]. Decarbonizing LMD is a core strategy for many LSPs, such as DHL [44]. Additionally, redesigning delivery routes, such as implementing linear freight routes, can improve connectivity and reduce emissions [24,45]. These improvements highlight the need for smart city planners to integrate environmentally friendly logistics solutions into city planning to reduce the ecological impact of last-mile deliveries.

Given the ever-evolving landscape of the LMD sector, researchers have investigated possible future developments. Shang [46], for instance, conducted a Delphi study targeting the logistics sector for the year 2025. This study involved 30 participants who created 41 projections across multiple dimensions. Similarly, Neghabadi et al. [47] pinpointed essential key factors for successful Smart city logistics by employing the Delphi technique, where 20 experts assessed 27 factors categorized under strategy, infrastructure, innovation, marketing, regulations, and environment. These research efforts highlight the crucial role of smart city planning in addressing infrastructure, regulatory, and environmental challenges, thereby facilitating sustainable growth in the LMD sector.

3. Methodology

3.1. Delphi Technique and Scenario Planning

There are several forecasting methods, including both quantitative and qualitative approaches like econometrics and expert workshops. In this research, we opted for the Delphi method because of its well-documented reliability and effectiveness in identifying long-term trends [48,49,50]. We developed a scenario study based on the Delphi method to envision potential future scenarios for LMD in 2030. A concise overview of the Delphi method and scenario planning is provided below.

Since the 1950s, Delphi research has been utilized across diverse fields like business, healthcare, and education [51,52]. It aims to achieve dependable group opinions and is valued for its effectiveness in predicting future outcomes based on expert opinions [52,53]. Delphi studies surpass traditional surveys due to the expert panels’ enhanced expertise and forecasting capabilities [54,55].

Based on the works of Dalkey and Helmer [56], Delphi studies are characterized by four essential features: anonymity, controlled feedback, iterative rounds and statistical group response. Anonymity allows participants to share their views freely, even if they differ from the majority opinion [57,58]. This method ensures that participants are not influenced by hierarchical status, which often occurs in face-to-face discussions [59], as the experts involved are not physically co-located. This approach also reduces financial and time-related constraints [58,60]. The iterative process, which involves at least two iterations, enables experts to refine their assessments without the risk of embarrassment [61,62]. Controlled feedback provided by the moderator after each round allows experts to interact indirectly with other participants, facilitating learning and adjustment of their evaluations [46,60,63,64]. The statistical group response involves summarizing the results numerically and graphically. Consensus among participants is measured using the interquartile range (IQR) [65,66], while changes in the standard deviation (SD) across different groups reflect the extent of alignment [24,67,68].

Scenario planning was developed to create various projections of the future, known as scenarios [50,69,70,71]. These scenarios are logical and comprehensible depictions of potential future occurrences that rely on several variables [72,73,74,75]. In our research, we utilized Amer’s [76] characterization of scenario planning as “a systematic approach to envisioning potential future scenarios” to inform organizational strategies and decision-making processes.

For practitioners, scenario planning presents a useful approach for exploring options amidst uncertain future outcomes [48,76]. This technique involves crafting structured scenarios to outline possible future conditions that serve as a foundation for strategic planning [63,64,72,73]. Scenario planning shifts the focus from uncertainty to potential opportunities enabling researchers and industry professionals to generate various scenarios that capture a spectrum of future possibilities [46,60,64,69]. The application of scenario planning in logistics has already been explored (see Tiberius and Weyland [63] and Shang [46] for a detailed overview). Often, it is integrated with other methods like the Delphi study [57]. In the context of smart city planning, scenario planning helps cities prepare for the impact of future emerging delivery technologies and services on smart city infrastructure and regulations.

While scenario planning holds promise, it needs to be undertaken carefully. The scenarios rely on a limited range of effecting factors and the process is demanding, necessitating continuous revisions [20,21]. To address these challenges, it is crucial to evaluate the scenarios for both coherence and feasibility [18,77]. A closely related concept, scenario thinking, is defined as “utilizing scenarios to foster innovative solutions for potential future contexts” [78]. Although scenario thinking can lead to potential innovations, our Delphi-based scenario research concentrates on assessing future possibilities depending on various projections, which might also consist of innovations. Consequently, there are substantial similarities between the two concepts. By integrating both methods, our Delphi-based scenario analysis adhered to five steps: developing Delphi projections, selecting panel experts, conducting the Delphi study, creating future scenarios, and deriving managerial insights.

Initially, we developed 14 Delphi forecasts. We then identified and invited qualified expert candidates to join our study. The first Delphi questionnaire was subsequently distributed to these participants. After evaluating the interim findings, we launched the second phase of the Delphi questionnaire, prompting the experts to adjust their earlier responses grounded on the collective feedback. In the fourth step, we developed three scenarios depending on both qualitative and statistical data. Lastly, we identified managerial implications for the pertinent groups.

To ensure the study produced valuable insights, we precisely defined its scope using three criteria [21,61,62]. Firstly, we chose a time frame of almost 5 years to encourage innovative thinking [46,74], with the year 2030 identified as an ideal target to allow experts to imagine future possibilities that differ significantly from current solutions [79,80,81,82]. Secondly, the study was concentrated on Saudi Arabia. Thirdly, we involved experts from the sector of LMD and related fields, including researchers, e-commerce sellers, suppliers, consultants, politicians and IT, as well as smart city planners. This comprehensive approach ensured that the findings are pertinent and valuable for smart city planning and policymaking.

3.2. Creation of Delphi Projections

The Delphi analysis entailed experts assessing likely future scenarios, which were articulated as brief statements referred to as projections [63,73,75]. This segment describes the formation of the Delphi projections, beginning with an explanation of the methodology and subsequently presenting the examined projections.

To identify the most relevant Delphi projections, we utilized three data sources [18,19,61]. Initially, we performed a comprehensive literature review, encompassing academic papers, studies, and white papers by highly regarded industry experts [20,62,77]. This review targeted consumer demand and behavior, emerging delivery technologies, innovative delivery services, and last-mile delivery regulations, in line with our two research questions. We concentrated on studies published in the past decade. Subsequently, we conducted an informal brainstorming sitting leveraging our business expertise. In early 2024, we organized expert workshops with professionals from academia and industry. These workshops allowed us to discuss, refine, and generate new projections. We then compiled a list of 51 projections formatted as closed-ended statements. This list was reviewed based on the features of IDT and TAM, considering their applicability and influence within the sector of LMD. Eventually, we nominated 14 projections to formulate a questionnaire with a suitable timeframe to minimize research exhaustion and participant dropout.

Table 1. Future-proofing last-mile delivery and smart city planning projections for 2030.

No.	Category	Projection
Consumer Demand and Behavior
1	Recipients are likely to demand 15 min instant deliveries for essential items like food, repairs, and fast-moving consumer goods (FMCGs).
2	Customers will schedule 30 min slots for next-day deliveries, with penalties imposed on both the customers and logistics service providers (LSPs) if the delivery window is missed.
3	Consumers need to share personal data to tailor parcel delivery locations based on their daily movements.
Emerging Delivery Technologies
4	Mobile delivery systems will rely entirely on electric power generated from sustainable sources.
5	A substantial share of last-mile delivery (LMD) will be handled by delivery robots.
6	Mobile parcel lockers will dominate last-mile parcel deliveries.
7	Drones will be used for parcel deliveries only in remote areas.
Innovative Delivery Services
8	Cargo bikes are expected to become a favored method for deliveries.
9	Standard deliveries will require recipients to pick up parcels from designated locations such as gas stations and supermarkets.
10	Logistics Service Providers (LSPs) will offer an optional 24 h night delivery service.
11	LSPs will collaborate by sharing their delivery networks, including vehicles, parcel lockers, and pick-up/drop-off stations, as well as data, to efficiently manage deliveries within specific areas.
Regulation
12	Access to the city will only be granted to delivery vehicles that maintain a high capacity utilization, such as exceeding 90%.
13	Municipalities will mandate that LSPs work together, limiting each to serve particular areas on specific days.
14	Cities will be redesigned by municipalities into 15 min cities, significantly dropping the essential for online shopping.

presents an outline of these projections, categorized according to the research questions.

According to Sorvali et al. [54], it is appropriate to consider fewer than 20 projections. The selected projections for our questionnaire, though not exhaustive, addressed key future issues in the LMD sector. Following guidelines from Sorvali et al. [54], Fritschy and Spinler [83], and Rowe and Wright [55], we made sure the projections were clear, reasonable, and plausible, avoiding any irrelevant, emotional, or provisional statements. We verified the questionnaire with both seasoned professionals and non-experts, making minor adjustments based on their feedback. Participants indicated that the Delphi questionnaire could be finished within 30–45 min. The pre-test results verified the face validity and reliability of the content [84,85].

In the following part of this section, we outline the Delphi projections for assessing LMD by the year 2030 (refer to Table 1). Clients are anticipated to adopt innovative aspects of LMD. Recently, instant delivery services like Jahez and Noon have emerged in Saudi, targeting a specific market for quick grocery deliveries in the downtown areas of major cities. Our expert workshop participants questioned if, after the initial surge in popularity, consumers would continue demanding instant delivery services, such as those promising delivery within 10–15 min, potentially making it a standard service for particular products (projection 1). These logistics service providers offer deliveries in under 10-min for a nominal charge.

Several e-commerce sellers now provide the option to schedule delivery within specific time windows, a service that has long been standard in the express B2B last-mile delivery industry. For example, Amazon [15] allows customers to choose delivery windows of four to five hours. The use of time windows has been explored in previous studies [86,87]. By 2030, next-day delivery with limited time window booking might become a typical customer expectation (projection 2). Projection 3 explores the readiness of consumers to give personal information, such as real-time location data. Studies show that consumers are more likely to give their data if they realize a tangible advantage or if the recipients of the data are few [88,89,90]. This prompts the inquiry into whether consumers will be willing to share private data for last-mile delivery services.

As the movement towards electrification and decarbonization accelerates, substantial transformations are anticipated within the LMD sector [91,92]. Our expert panel was therefore asked to evaluate the probability of mobile delivery solutions utilizing sustainably generated electricity (projection 4). Innovative delivery technologies, such as the autonomous delivery robots created by the Starship company, have now emerged [24,39,45]. These delivery robots are capable of delivering a single parcel at a time, and studies have examined their environmental, economic, and operational advantages [93]. The effectiveness of this technology will become evident in the coming decades (projection 5). Projection 6 introduces the widely recognized idea of smart parcel lockers, such as DHL Packstations [15,44]. Studies have explored customers’ readiness to utilize parcel lockers and the optimization of their placement [84,85,86,87]. Although various technologies can be adapted to different environments, from small towns to major metropolitan zones, drones might be especially well-suited for rural regions (projection 7) [94]. Mucowska [95] examined the advantages of cargo bikes, a concept that has resurfaced in the last-mile delivery (LMD) sector, now featuring advanced models with greater capacity. Both start-ups and traditional logistics service providers (LSPs) are increasingly adopting this technology, particularly in bigger cities. It is yet to be determined if cargo bikes need to continue to be a favored future delivery method (projection 8).

Delivery technology and service are expected to undergo significant changes. Customers will be able to pick up their packages at specific collection points, a method already commonly used in Sweden and certain remote areas of the United States [96]. However, in less populated regions, these handover points are often located in larger cities, which can be less convenient for recipients. While this technique might deliver ecological and economic advantages for LSPs, experts in a recent workshop expressed doubts about its feasibility for widespread adoption in Saudi Arabia (projection 9).

In addition to traditional delivery schedules, LSPs might introduce night deliveries (projection 10) to distribute last-mile deliveries more evenly throughout the day [93]. This could be particularly beneficial for recipients with non-standard working hours, such as shift workers. The participants of the workshop raised questions regarding the future ownership of essential hardware and software infrastructure by LSPs (projections 11). The hardware elements involved are delivery vehicles, cargo bikes, delivery robots, parcel lockers, and drones.

The ultimate projections emphasize environmental regulations. Recently, city centers have been subject to restrictions like the implementation of low-emission zones, the enforcement of congestion charges, with London serving as a notable example [17]. These restrictions facilitate smoother commutes for residents using bikes or public transport [97,98,99]. During the expert workshop, participants considered whether allowing only high-capacity delivery vehicles to access city centres would be effective (projection 12). Additionally, if LSPs fail to collaborate, municipalities might mandate cooperation by assigning specific districts to a single LSP per day (projection 13). This regulatory approach could yield benefits like reduced emissions and congestion [23,36]. Smart city planners need to consider these regulations to balance the efficiency of LMD with the overall quality of smart city living. The last Delphi projection examines the concept of 15 min cities (projection 14), where a smart city is designed so residents can access essential services within their neighborhoods [93]. These cities are designed to fulfill six key roles: residential, employment, supply, healthcare, education, and recreation [98,99,100]. For instance, Paris has announced its ambition to become a 15 min city. Integrating LMD innovations into the planning of 15 min cities can aid smart city planners in developing sustainable, efficient, and livable smart city environments.

3.3. Expert Panel Selection

The success of Delphi studies hinges significantly on the selection of panelists, as these studies often face scrutiny for including inadequately qualified members or changing the panel composition [73,101]. Therefore, a meticulous process for selecting experts is crucial. Our methodology encompassed four main steps: formulating Delphi projections, choosing panel experts, conducting the Delphi study, and creating future scenarios along with managerial insights.

Beiderbeck [74] outlined three essential criteria to prevent biases in the selection of experts: diversity, relevant knowledge, and the size of the panel. To maintain a comprehensive perspective and integrate smart city planning insights, we included a variety of stakeholders such as LSPs, e-commerce sellers, research institutions, IT and software providers, suppliers, consultants, politicians, and smart city planners. Galarza-María [102] emphasized that a diverse group of experts ensures reliable and robust outcomes by counteracting individual cognitive biases. Additionally, it was crucial for the experts to have sufficient information on LMD. Therefore, we incorporated a self-assessment of expertise within the Delphi questionnaire to improve data reliability [52,54,74,103]. We also set a requirement of a minimum five years of experience in LMD or similar fields [67,68,104]. Researchers needed to have published relevant studies [52,60,103], while consultants were expected to have relevant project experience or publications. Those elaborate in creating the projections were not included as participants in this research.

In alignment with Culot et al. [105], we employed a rigorous and impartial process to select experts. This process entailed desk research into business networks, thorough analysis of relevant publications, and evaluation of the authors’ professional connections. Our goal was to assemble a diverse panel of experts by including stakeholders from various institutions, capturing a range of perspectives from both employees and managers [52,106,107]. Focusing on the management level, we aimed to obtain insights from the experts’ role in long-term strategic planning [69,70,71]. We identified 127 potential candidates.

After sending out invitations, 52 experts agreed to participate in both rounds of the Delphi study. Among these participants, 19 categorized their level of expertise as basic, 24 as advanced, and seven as expert. On average, the specialists had 15 years of experience in the field of LMD, with individual experience ranging from five to 40 years. The key stakeholder groups included 18 employees, 23 management experts, six researchers, and five government representatives.

3.4. Conducting the Delphi Study

To maximize participation, we carried out two rounds of the Delphi study based on evidence that the second round yields the most reliable results [50,55,63]. Our primary objective was to pinpoint projections that garnered both agreement and opposition. According to Tiberius and Weyland [63], expert panel disagreements often highlight critical issues. We utilized the online platform to maintain high data quality and user convenience. Unlike paper-based surveys, it mandates participants to evaluate all projections, ensuring comprehensive assessments.

3.4.1. First Delphi Round

The core of the questionnaire comprised 14 projections, each displayed on a separate page to ensure focused responses and avoid information overload [54,55]. These projections were assessed using a seven-point Likert scale, which gauged probability, impact, and desirability, with boundary values ranging from 1 (not probable, very weak impact, very undesirable) to 7 (very probable, very strong impact, very desirable).

To quantify expert assessments, we calculated the average score for each projection and dimension using the arithmetic mean:

$$\underset{\_}{x}={\displaystyle \frac{1}{n}}\sum _{i=1} x_i$$

where $\underset{\_}{x}$ represents the average rating, $x_i$ is the individual rating provided by expert $i$, and $n$ is the total number of experts. This metric helped us identify projections with relatively high or low perceived likelihood, impact, or strategic value.

Experts were also encouraged to provide qualitative feedback, which offered valuable insights into their reasoning and interpretations. This open-ended feedback helped refine the second round of the study and added depth to the interpretation of quantitative trends.

To assist expert understanding of emerging technologies, visual aids were included for projections 4 to 7, which focused on innovative delivery solutions such as drones, delivery robots, and mobile parcel lockers. These visualizations provided concrete examples of potential applications and increased participant engagement.

Additionally, panelists were asked to assess the primary deployment contexts for these technologies by the year 2030. They selected from a range of urban typologies including rural areas, small towns, medium-sized cities, and major metropolitan zones to reflect the diverse applicability of LMD innovations in smart city planning.

The questionnaire concluded with socio-demographic questions, including years of professional experience, field of expertise, and current organizational role. These data supported the validation of the panel’s expertise and diversity.

The first Delphi round was conducted from early January to early February 2024. To ensure a high response rate, email reminders were sent to non-respondents after one and two weeks, respectively. The high level of engagement and completion rate in the first round provided a solid foundation for the iterative feedback and convergence process in the subsequent round.

3.4.2. Interim Evaluation and Second Delphi Round

Subsequently to the initial phase of expert assessments, we carried out quantitative analyses in Excel, including creating histograms. Based on von der Gracht’s [108] advice on providing controlled feedback, we shared preliminary findings with the expert panel. These findings included histograms, averages, and minimum and maximum values, and highlighted individual expert ratings. Personal statements were sorted into supportive and opposing remarks, emphasizing the crucial role of qualitative feedback in the expert re-evaluation process [46,50,103,108].

Mid-February 2024, we distributed the second Delphi survey along with the preliminary findings. We included an example at the beginning to illustrate how to understand the interim outcomes, followed by one projection on each page. Based on feedback from the first round, we made a slight adjustment to the additional question, requesting experts to evaluate the proportion of e-commerce in relation to sales. This allowed panelists to reconsider their replies in light of group feedback. Many participants responded positively, noting the importance of the topic and the clarity of the presentation. The inclusion of smart city planning considerations in our projections likely enhanced this engagement, as experts recognized the implications for city infrastructure and policy. This high level of engagement contributed to the good-quality outcomes we achieved.

3.4.3. Final Evaluation and Conclusion of the Delphi Study

By early June 2024, we completed the second round of the Delphi analysis. We carefully examined the data for any errors and reached out to several experts to confirm their experiences with LMD [58,108].

For each projection, we calculated standard descriptive statistics, including the mean, median, interquartile range (IQR), and standard deviation (SD). The IQR was used to measure the degree of consensus among panelists, defined as

$$IQR=Q_3-Q_1$$

where $Q_3$ is the third quartile and $Q_1$ is the first quartile of the ratings. Following previous Delphi studies [84,85,86,87], a threshold of IQR ≤ 2.0 was used to indicate sufficient consensus across expert evaluations.

To assess the convergence of responses between the first and second rounds, we examined changes in standard deviation, calculated as

$$SD=\sqrt{{\displaystyle \frac{1}{n}}\sum i=1n{(x_i-\underset{\_}{x})}^2}$$

where $x_i$ is an individual rating, $\underset{\_}{x}$ is the mean rating, and $n$ is the number of experts. A reduction in $SD$ from Round 1 to Round 2 indicated greater alignment among expert opinions, suggesting that the Delphi process contributed to refinement and consensus.

Evaluations were conducted using a seven-point Likert scale, with a consensus threshold set at 2.0, consistent with prior research employing similar scales (e.g., [84,85,86,87]). Following von Briel’s [107] methodology, we performed both syntax and content analysis, which highlighted a high level of engagement from the expert panel. The feedback and insights on smart city planning challenges and solutions emphasized the importance of this aspect in our study. Since participants were informed that the Delphi analysis would conclude after the second round, we were satisfied with the outcomes. The confidentiality of participants’ personal information was preserved. The last outcomes of the Delphi analysis were then shared with the expert panel.

3.5. Formulation of Future Scenarios

Delphi analysis frequently serves as a foundation for scenario development, allowing researchers to structure expert judgments into coherent representations of possible futures [53,109]. The process of developing these scenarios typically involves two main stages: plotting and clustering.

In the first stage, we plotted the mean values for likelihood and impact from the second round of the Delphi study for each of the 14 projections. To avoid the influence of subjective bias, desirability scores were excluded from the clustering process, as recommended by earlier studies [103,110,111]. Since all projections were evaluated using a seven-point Likert scale, the values were already standardized, and no further normalization was required.

In the second stage, we applied the fuzzy c-means clustering method using the R programming language. This non-hierarchical clustering algorithm was chosen for its suitability with smaller datasets and its capacity to represent overlapping clusters, a desirable feature for exploratory scenario work. Unlike hard clustering approaches (e.g., k-means), fuzzy c-means allows for each data point (projection) to belong to more than one cluster with varying degrees of membership, reflecting the multidimensional nature of future uncertainties.

The degree of membership of each projection $x_j$ to cluster iii was calculated using the following formula:

$$u_{ij}={\displaystyle \frac{1}{{\sum }_{k=1}^c {(\frac{\parallel x_j-c_i\parallel }{\parallel x_j-c_k\parallel })}^{\frac{2}{m-1}}}}$$

where

• $u_{ij}$ is the membership degree of projection $j$ to cluster $i$;
• $c_i$ is the centroid of cluster $i$;
• $m$ is the fuzziness coefficient (typically set to 2);
• $\parallel x_j-c_i\parallel$ is the Euclidean distance between projection $j$ and cluster centroid $i$.

The centroids of the clusters were updated in each iteration using

$$c_i={\displaystyle \frac{{\sum }_{j=1}^n {u_{ij}}^m\cdot x_j}{{\sum }_{j=1}^n {u_{ij}}^m}}$$

This approach allowed us to generate soft partitions of projections that could be interpreted flexibly during scenario formulation. For example, a projection with a high membership value across two clusters could be incorporated into multiple future narratives, reflecting its influence across various plausible pathways.

The choice of fuzzy c-means is further supported by its adoption in contemporary Delphi-based foresight studies focused on emerging technologies, logistics, and urban innovation [23,24,36,67,104]. Its flexibility in handling overlapping, complex, and interrelated trends made it ideal for our study on the future of last-mile delivery (LMD) and smart city integration.

Following clustering, we reviewed the grouped projections for thematic coherence and labeled the clusters according to their underlying dynamics. This interpretation was guided by the mean values of likelihood and impact, as well as expert commentary from the Delphi rounds. The scenarios were then formulated by combining projections within each cluster into narrative arcs that represent distinct, plausible futures for LMD in Saudi smart cities by 2030.

These scenarios serve multiple purposes: they inform policymakers and urban planners of emerging delivery challenges, help logistics firms assess their preparedness, and guide investment in smart infrastructure aligned with evolving consumer and regulatory expectations.

4. Results

4.1. Quantitative Analysis

The outcomes of the quantitative Delphi study are detailed in

Table 2. Summary of quantitative Delphi study findings.

No.	Projection	Probability Round 1 (n = 54)—IQR	Probability Round 2 (n = 52)—Median	SD Change (Mean)	Impact (SD)	Desirability (IQR)
Consumer Demand and Behavior
1	Instant Delivery Demand	2.10	4.9	4.60	1.70	1.90
2	Time Window Booking	2.90	4.4	4.40	1.80	1.80
3	Personalized Delivery Points	2.00	6.1	5.50	1.60	1.60
Emerging Delivery Technologies
4	Sustainable Electric Delivery	2.00	6.1	5.70	1.70	1.10
5	Delivery Delivery Robot Usage	3.10	4.1	3.80	1.70	2.30
6	Mobile Parcel Lockers	1.80	2.0	3.00	1.80	2.00
7	Drone Delivery in Remote Areas	2.00	4.1	4.10	1.60	2.00
Innovative Delivery Services
8	Cargo Bikes for Delivery	2.60	6.1	5.10	1.90	1.90
9	Collection Point Usage	2.80	5.1	5.00	1.80	2.10
10	Night Delivery Option	2.70	4.1	4.00	1.70	3.20
11	LSP Infrastructure Ownership	4.10	3.8	4.00	1.90	2.60
Regulation
12	High-Capacity City Access	2.60	5.0	4.70	2.00	2.50
13	Mandated LSP Collaboration	3.80	4.0	4.00	1.80	4.0
14	15 Minute City Planning	2.00	3.8	3.80	1.70	2.60

Note: probability was rated on a seven-point Likert scale (1 = very unlikely, 7 = very likely).

, with the results organized by research theme for clarity and consistency with Table 1.

In the first round of the Delphi survey, seven out of 14 projections (50%) reached a consensus. This number increased to 11 projections (79%) in the second round. This consensus rate is notably higher than the average of 30% found in other Delphi studies [50,65,66,103]. The IQR for projections 6 (Mobile Parcel Lockers), 10 (Night Delivery Option), 11 (LSP Infrastructure Ownership), and 12 (High-Capacity City Access) saw slight increases, while projection 14 (15 Minute City Planning) had a significant rise. Our primary aim was to identify projections that evoked both agreement and disagreement. Therefore, we emphasize projections 5 (Delivery Delivery robots Usage), 10 (Night Delivery Option), 11 (LSP Infrastructure Ownership), 12 (High-Capacity City Access), and 14 (15 Minute City Planning) based on their IQR in the second round. Despite the lack of full consensus, these projections are valuable due to the extensive debate they generated [65,66].

In our Delphi study, the average impact ratings for various projections varied from 4.11 for Mobile Parcel Lockers (projection 6) up to 5.78 for High-Capacity City Access (projection 12), with an overall average impact that is approximately close to 5. This suggests that the panel of experts considered the selected projections to be significant. Notably, Sustainable Electric Delivery (projection 4) received a desirability score as high as 6.25 and exhibited the smallest interquartile range (IQR) of 1.00. In contrast, Mobile Parcel Lockers (projection 6) had the lowest desirability mean at 3.33, indicating expectations for its use by 2030. The projections showing the highest convergence were Sustainable Electric Delivery (projection 4), as well as Drone Delivery in Remote Areas (projection 7), with reductions of 30.0% and 23.1%, respectively. On the other hand, LSP Infrastructure Ownership (projection 11) and High-Capacity City Access (projection 12) showed the least convergence, with reductions of 3.1% and 1.9%, respectively.

On average, each participant reviewed their initial probability scores for 7.4 projections. Of these, 3.9 projections were assigned a higher likelihood, while 3.7 projections were considered less likely. Unlike previous research (e.g., [21,112]), our expert panel also re-assessed the influence and desirability of each projection, resulting in an average of 6.7 and 6.9 revised evaluations per participant, respectively. Additionally, we asked the experts to give recommendations for projections 5 to 9, considering different regional contexts such as rural regions, small to large cities/metropolitan areas, for the implementation of emerging delivery technologies.

The findings reveal that the expert panel predominantly recommended deploying delivery robots (projection 5) in big cities, with potential applications in small- and medium-sized cities as well. Mobile Parcel Lockers (projection 6) were considered suitable for all regional categories. Experts suggested that Drone Delivery in Remote Areas (projection 7) is mainly appropriate for remote locations. Cargo Bikes for Delivery (projection 8) were expected to be used primarily in small to large cities. Collection Point Usage (projection 9) was projected to be mainly utilized in small to big cities.

These insights are essential for smart city planners to effectively integrate emerging delivery technologies into smart city infrastructures, ensuring sustainability, efficiency, and minimal disruption to smart city environments.

4.2. Stakeholder Group Analysis

Suppliers and smart city planners found the projections to be overly optimistic, while a think tank representative considered them rather conservative. Other stakeholders, including logistics service providers, e-commerce sellers, consultants, researchers, IT and software providers, and politicians, deemed the projections moderately likely. Projection 4 (Sustainable Electric Delivery) garnered widespread agreement among stakeholders, except for the think tank, which expressed strong disagreement. Politicians, smart city planners, and consultants were optimistic about projection 6 (Drone Delivery in Remote Areas), whereas logistics service providers and researchers were more skeptical of these technologies.

E-commerce sellers and the think tank rated the influence on the sector of LMD lower than other groups on average. In contrast, suppliers and smart city planners gave the highest impact ratings, while others viewed the influence as moderate. Projection 3 (Personalized Delivery Points) was perceived as having a significant influence on the sector of LMD by all experts except the politician and the think tank. Additionally, the think tank rated projection 12 (High-Capacity City Access) as having a negligible impact, whereas all other expert groups saw it as moderate. This divergence in influence ratings highlights the necessity of integrating smart city planning perspectives to address the required infrastructure and regulatory changes for LMD innovations.

Unlike the estimates for probability and impact, projection 4 (Sustainable Electric Delivery) was universally deemed desirable by all stakeholders. On the other hand, projection 13 (Mandated LSP Collaboration) was generally viewed as less desirable, with the exception of suppliers, researchers, and IT and software providers. According to Dabic-Miletic’s study [112], personal interests of participants can influence their evaluations. As a result, the desirability measurement was omitted from the scenario development method.

The intra-group analysis shows that LSPs demonstrated strong consensus on projection 4 (Sustainable Electric Delivery) but showed varying opinions on projection 3 (Personalized Delivery Points). Consultants exhibited extreme views on projection 13 (Mandated LSP Collaboration) while reaching a strong consensus on projection 4. Researchers, on the other hand, strongly agreed on projections 1 (Instant Delivery Demand) and 3 (Personalized Delivery Points), with varied assessments of the other projections.

These divergences likely reflect underlying stakeholder biases shaped by institutional priorities. Public sector participants, such as policymakers and smart city planners, may emphasize regulatory measures and social equity, leading them to rate collaborative initiatives (e.g., projection 12) as highly impactful. In contrast, private sector stakeholders and think tanks may prioritize operational autonomy and market efficiency, resulting in lower impact ratings for regulatory projections. These differences underscore the importance of multi-stakeholder engagement in interpreting Delphi results and suggest that the scenarios, while robust within the expert panel, may have varying levels of generalizability across different real-world contexts. Smart city planners, therefore, play a critical mediating role in reconciling these perspectives to design balanced urban logistics strategies.

4.3. Future Scenarios

Scenarios are essential for aiding companies in their long-term planning processes. Additionally, they provide smart city planners with critical insights to develop sustainable and efficient infrastructure and policies. We developed future scenarios for the sector of LMD using fuzzy c-means clustering.

Table 3. Membership degree and projection assignment in fuzzy C-means clustering.

#	Projection	Cluster 1	Cluster 2	Cluster 3
Consumer Demand and Behavior
1	Instant Delivery Demand	0.7020	0.2320	0.0660
2	Time Window Booking	0.3880	0.5050	0.1070
3	Personalized Delivery Points	0.7470	0.1850	0.0680
Emerging Delivery Technologies
4	Sustainable Electric Delivery	0.6100	0.2500	0.1400
5	Delivery Delivery Robot Usage	0.0040	0.0060	0.9900
6	Mobile Parcel Lockers	0.1200	0.1700	0.7100
7	Drone Delivery in Remote Areas	0.0600	0.0800	0.8600
Innovative Delivery Services
8	Cargo Bikes for Delivery	0.1950	0.1800	0.6250
9	Collection Point Usage	0.9300	0.0400	0.0300
10	Night Delivery Option	0.0550	0.8900	0.0550
11	LSP Infrastructure Ownership	0.6000	0.1600	0.2400
Regulation
12	High-Capacity City Access	0.0060	0.9900	0.0040
13	Mandated LSP Collaboration	0.0250	0.9400	0.0350
14	15 Minute City Planning	0.1050	0.3150	0.5800

displays the membership degrees for each projection individually.

Projections were categorized into clusters according to the degree of their highest membership. Cluster 1 depicts the infrastructure and services state, Cluster 2 emphasizes collaboration and regulation, and Cluster 3 involves recipient pick-up solutions, which include projections with relatively low likelihood and moderate impact.

We assessed the scenarios for reliability and credibility using the criteria set by Arend’s [113] and Wu et al. [114], analyzing trends, combinations of outcomes, and key stakeholder responses. This process included verifying the alignment of outcomes with the overall timeline, identifying any contradictory outcome combinations, and evaluating stakeholder reactions. Each scenario is detailed below and contextualized using IDT or TAM frameworks. It incorporates favorable and dissenting perspectives from expert panelists collected during both Delphi phases.

4.3.1. Scenario 1: Infrastructure and Services

This scenario encompasses eight forecasts related to infrastructure and services. Experts assessed this scenario as having the highest anticipated probability, with scores ranging from 5.7 to 6.2, suggesting a strong likelihood of these events happening by 2030. Importantly, there was agreement on six out of the eight forecasts, indicating a common viewpoint among the experts. However, consensus was not reached for projections 10 (Night Delivery Option) and 11 (LSP Infrastructure Ownership). The average impact of these projections is between 4.1 and 5.6, reflecting a significant effect on stakeholder operations.

The expert panel predicts that customers will soon expect extremely fast delivery times, such as within 15 min. Some specialists pointed out that companies like Jahez and Noon are already providing such services, indicating a potential rise in consumer demand for rapid delivery. However, one expert mentioned that this expectation might be limited to smart cities, and another noted that the demand for one-hour delivery has not met expectations at their company. Furthermore, the LMD sector may encounter increased cost challenges, including higher labor costs and congestion fees. An additional expert highlighted that advancements in software, especially artificial intelligence, will facilitate more precise predictions of consumer request and behaviour, making rapid delivery crucial primarily for essential items. Thus, the main factors will be the relative advantage and perceived usefulness from the customer’s perspective. Additionally, instant delivery could simplify daily routines for recipients. These shifts will necessitate smart city planners to design infrastructure that can support quick delivery services both efficiently and sustainably.

The emergence of drone technology (projection 7) has also been emphasized. The expert panel, with a moderate probability score of 5.0 and supported by 34 votes, anticipates that drones will primarily be employed for commercial purposes in rural regions. Although there are current regulatory challenges, some experts believe that by 2030, drone technology will have progressed considerably. These advancements are expected to include reduced noise levels, compliance with rigorous safety standards across various weather conditions, reduced operational expenses, and enhanced load capacities. One expert highlighted the limited practicality of drones in densely populated smart cities due to the high volume and density of deliveries. Participants discussed the potential for establishing networks of regular routes for remote areas, often combining drones with electric delivery vans, and noted together the benefits and challenges of this technology. However, some experts voiced concerns about obtaining large-scale regulatory approval for drones. Panelists also pointed out that even within large cities, there are remote locations like skyscrapers, but factors such as weather conditions, reliability, and overfly rights could restrict these applications. The acceptance of drone technology will depend on the perceptions of LSPs, suppliers, municipalities, and recipients concerning its usefulness, user-friendliness, and their intention to adopt it. Smart city planners will need to integrate drone delivery systems into city logistics frameworks, considering regulatory and safety standards.

The expert panel predicts that by 2030, cargo bikes (projection 8) will likely become a popular choice for deliveries, with a moderate probability score of 5.0. While bike deliveries are currently being utilized, experts emphasized the advantages of cargo bikes, such as their small size, eco-friendliness, capacity to alleviate traffic congestion, and versatility. The majority of experts considered this scenario highly desirable, highlighting that the introduction of new networks of micro depots, along with advanced designs featuring higher payloads and electric power, could enhance their perceived utility and user-friendliness [115,116]. A specialist forecasted that by 2030, city centers would no longer permit regular traffic, which could encourage logistics service providers (LSPs) to embrace this technology. Cargo bikes are specifically advantageous in smart city settings, but their benefits diminish in rural regions. However, some experts have raised concerns about the potential for this technology to result in low-paying jobs and have criticized the partial capacity of cargo bikes. Furthermore, the required infrastructure is not yet completely established in all smart cities. To fully realize the benefits of cargo bike delivery, smart city planners must develop appropriate infrastructure, such as dedicated bike lanes and micro depots, ensuring seamless integration into city logistics.

The expert panel had varying opinions on the desirability of night delivery services by LSPs by 2030. This projection (number 10) received an average score of 5.1 and a median score of 6.0. Advocates suggested that night delivery could distribute delivery volumes more evenly across 24 h, improve customer service, and potentially alleviate traffic congestion during peak hours. An expert pointed out that night delivery could remain a niche service, complementing existing pick-up and drop-off options. They noted that overnight delivery is already well-received in regions like New York City, where recipients find it convenient [114]. On the other hand, some experts questioned the profitability of night delivery due to potentially low volume and cited challenges such as regulatory issues and difficulties in recruiting qualified staff [114]. They also pointed out that most recipients are likely to be asleep at night; this option is particularly well-suited for B2B deliveries. Smart city planners will need to address the implications of night deliveries on smart city infrastructure, including noise regulations, traffic patterns, and safety measures.

Experts considered it somewhat likely that LSPs, including delivery robots, parcel lockers, and drones, will become more prominent by 2030 (projection 13). One participant emphasized that by owning assets, LSPs can attain seamless end-to-end integration, which enhances their control and visibility. The high use of assets in the sector of LMD is seen as a key advantage for LSPs, making the ownership of infrastructure a vital factor for success. However, other participants argued that LSPs might favor leasing infrastructure, such as parcel lockers, rather than owning them, as ownership could introduce complexity and reduce their perceived usefulness. A panelist noted that e-commerce businesses could consider investing in their own hardware and delivery infrastructure to reduce reliance on third-party logistics providers. Moreover, two experts suggested that local governments should deliver white-label delivery services, similar to the logistics hotels in Paris, to enhance smart city logistics. It was noted that smart city planners and municipalities need to determine their roles in providing or facilitating access to these essential infrastructures to support efficient and sustainable smart city logistics.

In scenario 1, the final forecast (forecast 14) addresses software infrastructure. This forecast sparked the most disagreement among experts, reflected by an interquartile range (IQR) of 5.0, owing to the software sector’s highly innovative nature. Some experts believed that LSPs would also dominate software infrastructure, while others disagreed, resulting in an average expected probability of 4.7. Proponents highlighted the advantages of having in-house tools for cost evaluation, route planning, and optimizing network efficiency. Some experts mentioned the potential for a few software solutions, typically three or four, are dominated by joint venture partnerships, forming an oligopoly. On the other hand, some participants argued that major software companies would develop the software, with LSPs subscribing to specific solutions. Thus, the decisions made by LSPs regarding software ownership are shaped by the complexity and the perceived value of the software. Smart city planners must consider incorporating these software solutions into smart city infrastructure to improve the efficiency and sustainability of last-mile delivery systems.

4.3.2. Scenario 2: Collaboration and Regulation

Scenario 2 focuses on collaboration and regulation, featuring four distinct forecasts. The likelihood of these forecasts materializing ranges from 4.1 to 4.7. Their average impact, however, is notably higher than in alternative scenarios, with values ranging from 5.2 to 5.8. Consequently, these projections are relatively probable and expected to have a substantial impact on the sector of LMD by 2030. Consensus has been achieved on two of these projections, while the remaining two require further expert deliberation.

Study participants assessed Projection 2, which deals with delivery time windows, as moderately likely and significantly impactful for the sector of LMD. This projection is closely linked to Cluster 1, depicting the services scenario and infrastructure. Several panelists proposed that implementing 30 min time slots for delivery by the next day could yield both sustainability and economic benefits through increased transparency and order consolidation. Recipients would advantage by better organizing their schedules, thereby reducing complexity and enhancing perceived usefulness. The process will be further supported by new drop-off standards and automated delivery systems. Smart city planners must incorporate these innovations into city logistics to ensure smooth operations and minimize disruptions.

Although some experts forecasted that consequences for missed delivery deadlines would be implemented by 2030 due to fierce competition, others disagreed. They argued that penalty fees could detract from the delivery experience for recipients and that logistics service providers (LSPs) cannot always predict unforeseen issues such as traffic congestion or adverse weather. Another expert proposed that recipients might favor flexibility over fixed time slots. Additionally, new smart city storage services will be needed, and innovations such as refrigerated packages and home reception boxes will make precise delivery windows more prevalent.

Study participants considered projection 2, which addresses time windows, to be moderately likely with a significant influence on the sector of LMD. This projection is closely associated with Cluster 1, emphasizing the infrastructure. Some panelists suggested that introducing 30 min delivery time slots for next-day shipments could provide both sustainability and economic advantages by improving transparency through order consolidation. Recipients would benefit from being able to plan their timetables more accurately, dropping complexity and increasing perceived utility. The implementation of new delivery protocols and automated logistics systems will facilitate this transition. Smart city planners will need to integrate these innovations into city logistics to ensure smooth operations and minimize disruptions. Some experts anticipated that penalties for failing to meet delivery windows would be enforced by 2030 due to intense competition. However, other experts argued that such penalties could negatively impact the delivery experience for recipients, as LSPs cannot predict issues like traffic congestion or inclement weather. Another expert mentioned that recipients would favor having flexible options rather than rigid time slots. Additionally, new smart city storage solutions will be crucial.

Regulating city access based on High-Capacity City Access (projection 12) received a neutral probability rating of 5.1 from the expert panel. Some experts emphasized the potential benefits of this measure, including reduced traffic, environmental improvements, a better cost–benefit ratio, consolidated routes, and increased operational efficiency, which could streamline delivery processes. Effective implementation of such regulations will heavily depend on smart city planning policies. Conversely, another group of experts challenged these benefits, suggesting that alternative measures like congestion fees, quotas for eco-friendly vehicles, limits on drop densities, and taxation could potentially be more effective. They also argued that with the growing volume of parcels, logistics service providers (LSPs) could optimize vehicle use without requiring such interventions.

Both sets of projections received a mean probability and median score of 4.0, along with similar influence assessments. Municipalities experiencing significant traffic problems may benefit from this approach, as it could streamline deliveries and reduce the number of logistics service providers (LSPs) involved, enhancing perceived ease of use. Some experts argue that government involvement is crucial for maintaining smooth traffic flow. On the other hand, others advocate for market-driven approaches, like variable pricing during peak and off-peak hours, as being more efficient.

They argued that instructed collaboration would complicate the management of parcel flows for LSPs and decrease perceived ease of use. Smart city planners must weigh the trade-offs of these regulatory measures to find a balance between efficiency and practicality. Since postal monopolies were abolished to encourage competition, experts viewed mandated collaboration as a step backward that could harm service quality. Additionally, one expert raised concerns about the authority and support municipalities would need to enforce such a regulation.

4.3.3. Scenario 3: Recipient Pick-Up Solutions

Scenario 3 examines five forecasts related to recipient pick-up solutions. In this scenario, the predictions have the lowest mean probability, ranging from 2.8 to 4.2, and the lowest average influence, with values ranging from 4.2 and 5.3. As a result, these projections are considered less likely to happen. Consensus was reached on three of the projections, while experts remained divided on the other two.

The expert panel expressed a subtle inclination to think that delivery robots (projection 5) would not have a significant impact on LMD. They evaluated that delivery robots do not perform well across all TAM dimensions and are considered a niche solution. Most experts thought that delivery robots would primarily be applicable in densely populated areas or controlled environments like university campuses. One expert suggests that the use of delivery robots should be reserved for explicit urgent products, like medication, which contributes to the perception that their relative advantage is low.

Experts have highlighted several issues with delivery robots, including their operational complexity, limited range, and the congestion they cause on sidewalks. They also encounter challenges with unpredictable car traffic and adverse weather conditions. Additionally, parcel handovers are less convenient compared to traditional home delivery, as recipients must retrieve their packages directly from the droid, which cannot navigate stairs. Smart city planners must consider these limitations when integrating droid delivery systems into city infrastructure, ensuring they complement broader smart city mobility strategies.

The experts unanimously agreed that SPLs will remain relevant through 2030 (projection 6). Only seven experts believed MPLs would become more appealing, and one participant suggested SPLs might decrease by 55%. However, most panelists emphasized the comparative advantages, perceived usefulness, and ease of use of SPLs. They highlighted SPLs as efficient logistics centres for last-mile delivery that are cost-effective, clean, quick, and reduce city congestion. Recipients also showed a strong intention to use SPLs due to the convenience of picking up and returning parcels at their leisure. SPLs can be strategically integrated into smart city planning to reduce traffic congestion and environmental impacts and improve the effectiveness of smart city logistics.

For projection 7 regarding MPLs, the expert panel collectively agreed that while they are a feasible delivery option, they are not expected to become the main method, receiving an average rating of 4.0. They suggested smart city implementation due to the increased flexibility and suitability for recipients. The experts highlighted the comparative advantage, perceived usefulness, ease of use, and customers’ willingness to adopt MPLs. Though, some experts mentioned that MPLs might not be suitable for all sorts of packages, potentially complicating deliveries. One panelist compared the technology to delivery robots, noting potential geographical limitations within cities. Another expert raised concerns that changing delivery locations could disrupt natural human habits, thus diminishing the perceived benefits of MPLs. Smart city planners are encouraged to assess the suitability of MPLs for different smart cities, ensuring alignment with residents’ needs and smart city mobility patterns.

The idea of establishing collection points (projection 13) for regular deliveries received mixed reviews from the experts. Some panelists highlighted that situating collection points mainly in residential neighborhoods could boost their perceived utility. One expert anticipated that municipalities would urge logistics service providers (LSPs) to support centralized collection points to reduce smart city congestion. On the other hand, some participants suggested that distinct pricing models might develop for home deliveries versus collection points. Despite these discussions, the panelists concurred that home delivery would continue to be the preferred choice. Nonetheless, municipalities can use collection points in smart city planning to manage congestion and enhance the efficiency of smart city freight systems.

Projection 14 focuses on the 15 min city idea, which aims to decrease reliance on online shopping. The expert panel rated the likelihood of its realization as low, with a mean probability of 3.1. While some experts supported the idea, noting that residents would appreciate short travel distances to essential services, others suggested that stores would become concept showrooms with products delivered to customers’ homes, highlighting the advantages of this smart city design. Despite the support for 15 min cities, most panelists did not trust that online sales would decrease. Additionally, some experts mentioned that implementing such a concept could take generations, impacting its feasibility in the short term. Smart city planners need to balance the long-term benefits of the 15 min city concept with current online shopping trends, incorporating this vision into sustainable smart city development strategies.

5. Discussion and Implications

The sector of last-mile delivery (LMD) has experienced significant development in recent years, and this upward trend is likely to persist. Factors such as digitization, heightened awareness of sustainability, and emerging technologies have the capacity to significantly alter the sector. To investigate possible future developments, we conducted a Delphi-based scenario analysis, focusing on changes in consumer demand and behaviour, emerging delivery technologies, innovative delivery services, and regulations concerning innovation diffusion and technology acceptance. Our study generated 14 projections for the LMD sector by 2030. We engaged 52 experts from various fields, including e-commerce retailers, consultants, research institutions, IT and software providers, politicians, and smart city planners. The experts evaluated the projections based on their probability, impact, and desirability. To ensure methodological rigour, we followed a strict procedural protocol. We aimed to pinpoint projections that had a consensus, as well as those where expert opinions varied. To accomplish this, we conducted descriptive and statistical analyses, along with a qualitative review of expert feedback. These results contribute to the understanding of how last-mile delivery can evolve toward more sustainable and resilient configurations. Practitioners, including smart city planners, are advised to integrate these findings into their strategic planning to ensure environmentally responsible, resource-efficient, and socially inclusive smart city logistics systems. The study highlights the importance of aligning last-mile delivery innovations with broader sustainable development goals, such as emissions reduction, improved urban livability, and equitable access to services. The next section concludes the paper with a summary of key insights and recommendations for future research on sustainable last-mile delivery.

In addition to technological and operational innovations, the success of future last-mile delivery systems depends heavily on governance dynamics and the interaction of key actors. The scenarios identified in this study reveal the need for adaptive governance models capable of managing complex urban logistics ecosystems. Centralized governance approaches may provide uniform regulatory frameworks and coordinated investments in infrastructure, such as city-wide networks of smart parcel lockers or drone corridors. Conversely, polycentric and collaborative governance models—where municipalities, private actors, and citizens share decision-making responsibilities—can foster greater innovation and context-specific solutions. Public–private partnerships (PPPs) are particularly critical in implementing sustainable delivery technologies and services, as they enable resource sharing and align private sector agility with public sector oversight. For example, the successful deployment of mobile parcel lockers and night-time autonomous deliveries may require multi-stakeholder agreements balancing efficiency, equity, and environmental considerations. Therefore, governance strategies must be designed to integrate centralized oversight with local flexibility to support sustainable, resilient, and inclusive urban logistics systems.

Institutional feasibility and power dynamics are crucial determinants of whether sustainable last-mile delivery scenarios can be realized in practice. In the Saudi context, institutional enablers include Vision 2030’s emphasis on smart cities, digital transformation, and public-private partnerships, which create a favorable environment for adopting innovative logistics solutions. However, potential barriers remain, such as centralized decision-making structures and limited local governance autonomy, which may constrain context-specific adaptations. Power asymmetries between large global logistics providers, emerging local players, and municipal authorities could also shape implementation outcomes. For example, while multinational LSPs may have the resources to deploy advanced technologies like drones or autonomous vehicles, smaller local firms could face challenges in accessing such innovations without supportive regulatory frameworks and collaborative governance models. Addressing these asymmetries requires policies that encourage equitable participation and capacity-building among diverse actors in the LMD ecosystem.

Implications

This section examines the implications for management and policy arising from the proposed scenarios for key stakeholders in LMD. These stakeholders encompass LSPs, policymakers, local governments, e-commerce companies, IT and software firms, and suppliers of LMD technologies like cargo bikes and drones.

LSPs must focus on three main areas for the future of LMD. First, accelerating the transition to sustainable logistics through decarbonization is crucial. Major companies have already made significant investments in electric delivery fleets, such as DHL and Amazon, which ordered 100,000 electric LMD vehicles [117]. LSPs need to maintain this momentum and continue transforming their fleets swiftly. While these investments target large vehicles suitable for long-distance routes, LSPs should also consider environmentally friendly solutions for inner-city deliveries. Our research suggests the implementation of cargo bikes for deliveries involving small quantities and high-value items. In the courier sector, freelance couriers frequently handle deliveries, and their decision to use electric cargo bikes is impacted by factors such as the bikes’ range and cost [39,118,119]. Therefore, LSPs in this segment should offer additional motivations to couriers to encourage the use of electric cargo bikes, reducing emissions and other LMD-related externalities [120,121,122,123].

Additionally, LSPs should focus on expanding their network of smart parcel lockers (SPLs). In the study context, installing designated lockers in blocks of flats in major cities, similar to the system in China, could be particularly effective. LSPs should also offer incentives to encourage customers to use collection points more regularly. Improving the sustainability of delivery operations by optimizing collection point usage can help reduce unnecessary vehicle miles and fuel consumption. Enhancing the location, safety, and accessibility of these points, especially when recipients are not home, will make the service more appealing. Research has shown that these factors significantly influence the use of pick-up and drop-off points [124,125]. Furthermore, LSPs should incorporate mobile parcel lockers (MPLs) into their technology portfolio and minimize reliance on delivery robots, except for specific low-disruption environments like company premises and hospital areas. Another potentially viable application is the use of autonomous delivery services for nighttime deliveries, which can reduce congestion and support more sustainable traffic flows. However, in some cities, delivery surcharges might be necessary to maintain the profitability of the service.

Second, LSPs should focus on improving and expanding their range of services. The conventional division into express, courier, and parcel services might not be adequate anymore. It is crucial to clearly differentiate the sector of LMD. For example, LSPs could provide delivery within specific time slots as a typical service for certain product categories, such as high-demand consumer goods. According to Kiba-Janiak et al. [37], this approach can ensure efficient delivery processes while remaining cost-effective. Recipients are unlikely to face penalties for not meeting time windows, and LSPs might offer compensations instead. Instant delivery services, such as grocery deliveries, are expected to stay a niche market. Companies in this sector must balance speed and cost as recipients tend to prefer the fastest service available. By refining service segmentation and integrating sustainable delivery models, LSPs can reduce operational strain and improve overall system performance.

Innovative delivery technologies can enhance the segmentation process and help minimize failed deliveries, reducing environmental impacts. According to previous research, these technologies can decrease the rate of late instant deliveries by as much as 49% [126,127]. LSPs can benefit from collaboration for specific product ranges. Our study emphasizes the importance of clearly defining the collaboration’s scope to ensure success. For example, collaborating on the delivery of specific goods, such as refrigerated items, might not impact the core business of competitors but could improve the utilization of refrigerated delivery vehicles. Previous research has shown that co-opetition requires significant coordination but can also foster innovation [128,129].

Third, LSPs should meticulously specify hardware and software requirements, as IT infrastructure will be crucial in the LMD area by 2030. Addressing consumer habits effectively is essential, and LSPs should capitalize on receivers’ willingness to give personal data. Transparency about the organization receiving the data and its intended use is crucial, as customers are less likely to share personal information with unknown entities [88,130]. By leveraging their wide data networks, LSPs can develop new services that align with evolving consumer demand and behaviours and strengthen their contribution to sustainable urban logistics.

Public authorities are uniquely positioned to act as enablers of innovation in last-mile delivery by creating regulatory frameworks, providing infrastructure investments, and fostering collaborations between private logistics providers and local communities. At the same time, they may need to mediate potential conflicts between stakeholders, such as balancing the interests of multinational logistics firms with those of local businesses or ensuring equitable access to delivery innovations across different urban populations. Effective policymaking in this context requires a proactive, multi-stakeholder approach that integrates technological advancement with social and environmental considerations.

Politicians and municipalities should focus on two primary goals: developing the necessary infrastructure and outlining the critical components of a policy framework. Our research anticipates that by 2030, Smart Parcel Lockers (SPLs) will be extensively utilized. Nonetheless, experts highlight that SPLs are mostly installed on private properties such as gas stations or supermarkets, encountering significant hurdles in public areas due to complicated bureaucratic approval processes. Therefore, local governments should ease the installation of SPLs in public spaces by streamlining permit and approval processes, reducing bureaucratic barriers, and designating municipal land for shared locker hubs. Examples include allocating areas near public transport stations, libraries, or community centers as potential SPL sites. These measures would promote equitable access and operational efficiency while reducing traffic congestion from failed deliveries [124,131].

Municipalities need to establish specific regulatory guidelines for the sector of LMD, including those involving drone deliveries. Although people may be receptive to receiving packages by drone, there are still public concerns that need to be addressed [132,133,134]. Therefore, municipalities might consider allowing drone deliveries for essential items like medicine and emergency supplies. Regulations may be needed to control instant delivery traffic, giving municipalities a reason to restrict such services by, for example, applying white-label solutions or promoting collaboration among providers. Our study also suggests that policymakers could encourage night deliveries to help manage daytime traffic, provided there are measures in place to minimize noise. This supports the results of Lemardelé et al. [135], who indicated that nighttime delivery can be an effective strategy for achieving sustainable LMD practices. However, our study advises against regulating city access based on van capacity usage.

Despite the robust partnerships e-commerce companies have with LSPs—including offering shipment tracking through their apps—there is still room for enhancing this collaboration. As previously noted, the traditional sector of LMD, encompassing courier, express, and parcel services, is becoming more specialized. E-commerce businesses can offer greater customer flexibility by introducing new services like immediate or late-night deliveries. Enhancing the interfaces between e-commerce platforms and LSPs can significantly improve the overall experience of the customer throughout the entire delivery process. For instance, enabling customers to schedule specific delivery time slots directly through the e-commerce platform can make the process more convenient. Some e-commerce companies, like Amazon and Zalando, have established their own LMD distribution networks. While this trend may increase competition, it also presents an opportunity to promote more sustainable delivery strategies through integrated platform–LSP partnerships.

IT and software vendors must illustrate use cases that show how the increased volume and diversity of recipient data, when integrated with weather information, can improve efficiency and customer satisfaction. According to our research, recipients are increasingly willing to give personal data, which urges software vendors to develop innovative features that enhance the delivery services provided by LSPs. One example of such a use case is the optimized location-based routing that adapts to recipients’ changing locations, particularly useful for delivering food or medication. Additionally, there are new opportunities to integrate delivery bots into existing IT systems for specialized deliveries, which could support sustainable micro-logistics in dense urban zones.

While this study highlights the potential of smart city technologies to enhance last-mile delivery systems, it is important to acknowledge critical perspectives from urban studies that caution against over-reliance on technological solutions. Issues such as data governance challenges, risks of citizen exclusion, and the potential for increased surveillance must be considered to ensure that future urban logistics systems are not only efficient but also equitable and socially inclusive.

Public authorities are uniquely positioned to act as enablers of innovation in last-mile delivery by creating regulatory frameworks, providing infrastructure investments, and fostering collaborations between private logistics providers and local communities. At the same time, they may need to mediate potential conflicts between competing stakeholders, such as balancing the interests of multinational logistics firms with those of local businesses or ensuring equitable access to delivery innovations across different urban populations. Effective policymaking in this context requires a proactive, multi-stakeholder approach that integrates technological advancement with social and environmental considerations.

6. Conclusions

The sector of last-mile delivery (LMD) has experienced remarkable growth recently, driven primarily by the surge in e-commerce. This sector remains fiercely competitive, with new entrants targeting untapped market segments, while advancements in digitization and sustainable delivery practices continue to reshape industry standards. Through a Delphi-based scenario study focused on the year 2030, this paper has explored future trends in the LMD sector, particularly in relation to smart city planning and healthcare logistics.

Our research forecasts future trends in consumer demand and behavior, emerging delivery technologies, innovative delivery services, and regulatory measures, offering valuable insights for sustainable smart city infrastructure and policy development. These forecasts were confirmed through a two-round Delphi study with 52 experts from the LMD sector, academia, and government. Utilizing fuzzy c-means clustering, we identified and examined three future scenarios using Innovation Diffusion Theory (IDT) and the Technology Acceptance Model (TAM).

The three scenarios present distinct pathways for the evolution of urban logistics. The first scenario envisions centralized governance and large-scale technological deployment, enabling uniform urban delivery networks but potentially limiting flexibility at the local level. The second scenario emphasizes polycentric governance models and public-private partnerships, fostering innovative, context-specific solutions while requiring robust coordination across stakeholders. The third scenario reflects a slower adoption trajectory shaped by societal hesitance and regulatory caution, highlighting the need for incremental policy adjustments and adaptive infrastructure planning.

For governments and policymakers, these findings underscore the need to proactively design regulatory frameworks that balance innovation with social equity and environmental protection. Public health logistics providers are encouraged to leverage emerging technologies such as drones and autonomous vehicles to improve access to critical healthcare services, especially in underserved urban areas. Technology developers, meanwhile, can seize opportunities to create scalable, user-centric solutions that align with evolving urban logistics systems and smart city infrastructures.

By framing last-mile delivery as a key component of sustainable urban development, this study emphasizes the importance of multi-stakeholder collaboration and data-driven planning to create resilient, inclusive, and low-emission logistics ecosystems. Future research directions, including addressing identified limitations and integrating quantitative modeling, are discussed in the next section to build on these insights.

7. Limitations and Future Research

Numerous pathways for further research are evident. Initially, this study concentrated on the effects on key participants involved in facilitating and supporting last-mile delivery, with a focus on the supply side, thereby excluding parcel recipients from the expert panel. Future studies should address this limitation by improving stakeholder representation and including underrepresented groups such as parcel recipients, rural residents, small local businesses, politicians, urban sustainability specialists, and public health professionals. The absence of these perspectives in the current study may bias the scenarios toward supply-side priorities, potentially overlooking user-centric factors such as accessibility, affordability, and equity of services. Examining the demand side, particularly the experiences and preferences of end-users, is vital for developing more inclusive and citizen-centered smart city logistics systems. To achieve this, future research could employ concrete methods such as discrete choice modeling or stated preference surveys to quantify user preferences and trade-offs regarding innovative delivery solutions like mobile parcel lockers (MPLs), autonomous delivery robots, and night-time delivery services.

Secondly, while the Delphi-based scenario analysis provides valuable qualitative foresight, future research could enhance robustness by integrating quantitative data. This may include sustainability indicators such as carbon emissions, delivery efficiency metrics, and equity of access. Such integration would enable a more comprehensive evaluation of environmental, economic, and social impacts, providing smart city planners and policymakers with precise data for evidence-based decision-making.

Thirdly, to improve scenario validation and reduce potential biases, future studies could employ more structured validation frameworks. Applying methods such as the Büchel and Spinler ranking approach [50] would strengthen the objectivity and reliability of scenario prioritization, especially in complex, multi-stakeholder environments.

Lastly, although the expert panel in this study was notably diverse, with a higher representation of logistics service providers, consultants, and researchers, the underrepresentation of parcel recipients, rural residents, and other key groups may limit the inclusivity and relevance of the scenarios. Addressing this in future research would help avoid sectoral bias and ensure a wider range of perspectives in shaping resilient and equitable LMD strategies.

By addressing these areas—broader stakeholder engagement, integration of quantitative data, a stronger end-user perspective, and enhanced scenario validation—future studies can build on this foundation to develop more resilient, equitable, and sustainable last-mile delivery systems.