Toward a Holistic Bikeability Framework: Expert-Based Prioritization of Urban Cycling Criteria via AHP

Abstract

This study applies a multicriteria decision analysis to explore how experts from different backgrounds assess traditional and emerging criteria for urban cycling. A hierarchical model with 7 main criteria and 31 subcriteria was evaluated by 30 specialists from academic, technical, and user-focused groups. Using pairwise comparisons and aggregated judgments, this study reveals points of agreement and divergence among expert priorities. Safety and infrastructure were rated as the most important factors. In contrast, contextual and technological aspects, such as Multimodality, Environmental Quality, Shared Systems, and Digital Solutions, received moderate to lower weights, with differences linked to expert profiles. These results highlight how different disciplinary perspectives influence the understanding of bikeability-related factors. Conceptually, the findings support a broader view of cycling conditions that incorporates both established and emerging criteria. Methodologically, this study demonstrates the value of the Analytic Hierarchy Process (AHP) as a participatory and transparent tool to integrate diverse stakeholder opinions into a structured evaluation model. This approach can support cycling mobility planning and policymaking. Future applications may include case studies in specific cities, combining expert-based priorities with local spatial data, as well as longitudinal research to track changes in cycling conditions over time.

1. Introduction

Cycling has emerged as a fundamental element of moving toward more sustainable, resilient, and inclusive cities, particularly considering growing concerns about climate change, air quality, urban congestion, and public health [1,2]. In response, bicycles have been increasingly integrated into mobility policies, with bikeability gaining traction in urban planning, transportation, and public health research [3]. This reflects a recognition of cycling as a feasible option for motorized transportation that is capable of delivering individual and collective benefits while contributing to several Sustainable Development Goals [1,4].

Regarding urban environments, bicycles are a versatile and low-cost solution. They require less space, reduce travel time for short distances, and facilitate seamless integration with various transport modes [5,6,7]. In addition to encouraging healthier routines and mitigating carbon emissions, cycling also supports local economic activity. Furthermore, when properly planned, cycling infrastructure is more economical and adaptable than car-centric alternatives [8,9].

It has been demonstrated that new phenomena, such as electric bicycles, electrified sharing systems, and digital technologies, have challenged traditional bikeability assessment models, prompting more inclusive, systematic, and methodologically robust approaches [3]. Among these emerging trends, the rising popularity of electrically assisted bicycles (e-bikes) is particularly notable. They combine energy efficiency, user accessibility, and extended reach, positioning themselves between active and motorized mobility options [10,11]. Research has shown that e-bikes attract new user profiles, including older adults, women, individuals with lower physical fitness, and commuters. This significantly broadens the potential user base for cycling mobility [12,13]. However, their higher average speed and increased weight require adaptations to cycling infrastructure, especially concerning safety, segregation, and interoperability with conventional bicycles [14,15]. In addition, the large-scale adoption of e-bikes raises challenges related to charging infrastructure, secure parking, and the environmental management of battery life cycles [16].

Bike-sharing systems (BSS), including electric variants, play an increasingly prominent role in urban mobility strategies. From the informality of early generations to the digital sophistication of present-day models, the schemes enhance their reach and operational efficiency [17,18]. Their integration has helped to reframe the bicycle as a practical and multimodal transport option. In doing so, BSS contributes to key urban objectives by reducing short car journeys, minimizing environmental impact, and reinforcing public transport connectivity. They are particularly effective in addressing the challenges of first and last-mile gaps [7,19,20]. E-BSS, in particular, offers significant advantages in cities with topographical constraints and among user groups that are less likely to cycle under typical conditions [17,21]. Still, despite their benefits, challenges and operational drawbacks remain. Key issues include unequal access across different urban areas, misuse and vandalism, regulatory complexities tied to dockless models, and challenges in fleet balancing [22,23].

Another transformative vector lies in the growing incorporation of digital technologies into the cycling experience. Tools such as mobile applications, Big Data platforms, the Internet of Things (IoT), and artificial intelligence now support better route guidance, performance tracking, and modal integration [24,25]. These digital solutions not only increase the convenience of cycling through features such as route planning, gamification, incentive schemes, and individual performance tracking but also generate valuable data for evidence-based urban planning [26,27]. Real-time data can guide more targeted interventions by mapping flows, conflicts, and infrastructure gaps. However, challenges remain, including platform compatibility, fragmented datasets, and ongoing privacy concerns [28,29].

While many studies have investigated cycling conditions, most focus on traditional infrastructure and safety-related metrics, often overlooking emerging factors. Additionally, there has been limited attention to how experts from different disciplinary and stakeholder backgrounds prioritize these factors. This study aims to address this gap by employing the Analytic Hierarchy Process (AHP) to integrate and compare the perspectives of academic, technical, and user-focused experts within a comprehensive hierarchical framework of urban cycling criteria. The insights gathered will be linked to implications for planning and policy. The remainder of this paper is structured as follows: Section 2 discusses the theoretical background on bikeability, multicriteria decision analysis, and the application of the AHP method. Section 3 outlines the case study and methodological approach, including expert consistency, group consensus, and prioritisation patterns. Section 4 discusses the main findings, and Section 5 concludes the paper by outlining potential directions for future research.

2. Background

2.1. Bikeability Assessment

Bikeability has gained ground as a strategic approach in sustainable mobility planning, aligned with efforts to foster cities that support cycling for everyday travel, recreation, and tourism. Its theoretical and methodological consolidation has enabled the identification of the elements that constitute a cycle-friendly environment and the development of diagnostic tools and decision-support mechanisms that inform public policies, cycling plans, and infrastructural interventions. Initially understood as the capacity of a given territory to accommodate bicycle use, the concept has evolved to encompass a diversity of definitions and analytical scales, ranging from more normative perspectives focused on infrastructure performance [30,31] to more perceptual approaches, which incorporate subjective and contextual factors related to user experience [32,33]. This variety of perspectives underscores how complex interactions with the spatial structure of cities shape cycling.

Recent literature identifies a recurring set of dimensions for assessing bikeability: traffic and personal safety, cycling infrastructure, accessibility and connectivity, environmental characteristics, and behavioral factors. These dimensions are operationalized through both objective indicators, such as lane width, the presence of segregated cycle paths, pavement quality, or dedicated signage, and subjective ones, such as perceived comfort, feelings of safety, or the route’s overall attractiveness [34,35,36]. In such studies, the usual principles of cycling infrastructure design, such as safety, comfort, coherence, attractiveness, and directness, are broken down into indicators applicable at the street segment scale. Previous research has also proposed analytical frameworks integrating multiple domains. Since network analysis can be extended to a broader scale and oriented toward future scenarios, it may serve as a planning tool to support the development of cycling’s potential. In that context, a model based on fuzzy logic and geoprocessing was developed, incorporating six key factors: accessibility, infrastructure, parking, road features, urban surroundings, and traffic safety [37].

Within this context, the growing role of multicriteria approaches, such as the Analytic Hierarchy Process (AHP), also stands out in the structuring and prioritizing of criteria involved in cycle network evaluation. Although the methodological development of AHP is explored in this study, it is worth noting that the method has enabled the integration of the diverse factors involved in the decision-making process, reflecting the relative weight that different users and experts assign to the assessment criteria [35,38].

Among the studies, it is possible to identify the criteria most frequently used to evaluate urban bikeability. Safety, both in terms of traffic conditions and public perception, emerges as the dominant factor, followed by cycling infrastructure (presence, typology, continuity, and quality), accessibility and connectivity, environmental conditions (including topography, greenery, and air quality), and behavioral and social factors, such as users’ demographic profiles and their perception of risk. These criteria form a common core that underpins most assessment models. However, a lack of consideration for innovative parameters, such as electric bicycles, sharing systems, and digital solutions, was highlighted in previous work, despite their growing relevance in cycling mobility [3]. Ongoing shifts in urban mobility demand a revision of how cycling conditions are assessed. While conventional models still prioritize infrastructure and safety, recent studies have argued for including digital platforms, shared bicycles, and innovations such as electric bicycles [39,40].

Recent research has expanded the analytical scope of bikeability assessments, but traditional criteria still dominate. To illustrate this trend, we mapped the temporal distribution of seven evaluation criteria across 24 studies identified in two previous review papers [3,36]. Each study was classified according to our analytical framework.

Figure 1 shows the number of studies addressing each criterion (indicated in parentheses). Throughout the period analyzed, Infrastructure and Safety were the predominant focus areas, while Environmental Quality and Multimodality received limited coverage. Specifically, Environmental Quality was assessed through the sub-criterion of Air Quality (with no studies evaluating noise), and Multimodality was explored only in terms of Connection to Public Transport. Shared Systems were addressed solely through the Availability of Shared Bicycle Systems.

These patterns highlight the persistence of traditional focuses in bikeability assessments and the slower integration of emerging dimensions. Notably, Digital Systems were not mentioned in any of the reviewed studies, which points to their absence in established bikeability assessment frameworks.

2.2. Multicriteria Decision Analysis (MCDA) in Urban Planning

Multicriteria decision analysis (MCDA) provides tools for addressing choices involving multiple goals, which may sometimes conflict. Combining diverse analytical dimensions enables a systematic evaluation of alternatives [41]. The increasing complexity of urban problems requires decision models capable of simultaneously accounting for technical, social, environmental, and economic variables, thereby justifying the widespread application of MCDA in spatial planning. The formulation and structuring of the decision problem are central stages in the application of MCDA. The hierarchical decomposition of complex systems allows decision elements to be organized from general objectives to specific criteria, facilitating comparative evaluation among alternatives [42]. The method of transparency preserves the internal logic of evaluators’ judgments while also making it possible to replicate the model in varied decision-making contexts [41].

From a methodological perspective, multicriteria methods can be classified into two broad categories: Multi-Attribute Decision Making (MADM) and Multi-Objective Decision Making (MODM). The former are applied when dealing with a discrete and finite set of alternatives and are used to evaluate available options based on multiple criteria. MODM methods, on the other hand, are employed in continuous optimization problems, where the aim is to identify compromise solutions among competing objective functions, taking into account different dimensions and their interdependencies [41].

The Analytic Hierarchy Process, widely used among MADM techniques, provides logical clarity and a well-structured hierarchy to convert subjective assessments into numerical scores, enabling comparisons based on decision-makers’ views. A consistency check strengthens the process, making AHP a strong fit for decisions requiring openness, traceability, and stakeholder involvement [41,42].

Its applicability has been demonstrated across various domains: in strategic planning and security [43,44,45,46,47], project management [48,49], natural resource management and sustainability [50,51,52,53,54,55], and urban and transport planning [56,57,58,59,60,61,62]. Its capacity to operationalise complex decisions in a structured manner has contributed to its growing diffusion among researchers and decision-makers working in urban contexts.

2.3. AHP Applications in Bikeability Assessment

The analysis of urban bikeability has emerged as a central dimension of sustainable city planning, requiring methodological approaches capable of integrating multiple technical, environmental, social, and behavioral criteria. In this context, the AHP has distinguished itself, integrated qualitative and quantitative preferences, and produced robust and transparent rankings. The method’s flexibility has enabled its application across various active mobility scenarios, from prioritizing cycling interventions to assessing network or territorial suitability.

Several studies have recently employed AHP to support bikeability assessments in urban contexts. This method has been used to structure participatory decision-making processes involving experts and stakeholders in order to define priority measures to promote cycling, such as low-speed zones and educational campaigns [56]. It has also been applied to investigate specific aspects of cyclist interaction with other road users, particularly in shared spaces with pedestrians, thus contributing to the planning of safer and more inclusive infrastructure [63,64].

The methodology has also proven helpful in complementary domains, such as route selection for cycling tourism. In this context, AHP was applied to engage cycle tourists and guides in identifying decisive factors, including road quality, surrounding landscape, and local events [65]. Other studies used the method to evaluate green infrastructure principles with direct implications for integrating cycling corridors into urban space [66,67]. These examples illustrate AHP’s ability to encompass both technical dimensions and subjective perceptions within the realm of cycling mobility.

While the AHP has been applied across various contexts, its use in creating composite bikeability indices remains relatively limited. Most studies employ this method primarily to support rankings or isolated decisions. A notable exception has developed a spatial index that integrates slope, intersection density, and vegetation cover, utilizing AHP-derived weights within a Geographic Information System (GIS) environment. This demonstrates the recent emergence of such applications and highlights the benefits of integrating geoprocessing with relevant indicators [35]. This confirms the recent nature of such applications and the value of integrating geoprocessing and indicators. Typically, the expert samples in these studies are small, ranging from 7 to 17 participants [35,56,65,67], which can limit their representativeness. Broader and more diverse panels are likely to yield stronger models. The growing interest in expert-based and spatial tools indicates promising avenues for advancing cycling policy.

3. Case Study in Portugal

3.1. The Expert Panel

The selection process began by sending invitations to potential participants with recognized experience and knowledge in urban and transport planning. In addition to ensuring expertise in the subject matter, invitations were directed to individuals representing different stakeholder perspectives: academic, technical/professional, and user/advocacy, to capture a broad range of insights. Those who accepted the invitation were enrolled in the study and provided with detailed instructions about the evaluation process. All assessments were conducted online.

Before initiating the pairwise comparisons, participants completed a supplementary questionnaire designed to gather background information. This included age, gender, frequency and purpose of bicycle use, level of education, institutional affiliation, and experience in mobility-related projects. This information helped characterize the panel and allowed for the segmentation of responses based on expert profiles, as discussed later in the case study. The questionnaire also introduced the hierarchical model of the evaluation, consisting of seven main criteria and thirty-one subcriteria. Participants were asked to compare these elements using a nine-point scale to indicate their relative importance. The individual matrices were then aggregated for analysis, with consistency checks applied to ensure reliability.

The panel consisted of 30 experts with backgrounds in urban and transport planning. Based on the supplementary questionnaire, three distinct groups were identified: academics (12); professionals from public and private institutions (11), including engineers, architects, and technical planners; and user representatives (7), comprising seasoned cyclists and members of cycling advocacy organisations. Although no prior balancing was imposed during the selection process, the final composition reflected a wide range of perspectives and experiential knowledge.

Most participants (43%) fell within the 41–50 age range, generally corresponding to mature stages of professional experience. The panel included 11 women and 19 men. Bicycle usage patterns varied among the participants: while those from academic and professional backgrounds reported mixed frequencies, 71% of user representatives indicated they cycled three or more days per week. The size of the panel is significantly larger than what is usually observed in most AHP studies on urban cycling [35,56,65,67] and was considered sufficient to ensure robust results. Combined with the deliberate inclusion of different stakeholder profiles, this composition enhanced the diversity of viewpoints and strengthened the reliability of the prioritization outcomes.

3.2. Methodological Approach

This study employs the Analytic Hierarchy Process to assess factors affecting urban bikeability. AHP utilizes a structured decision-making framework that incorporates pairwise comparisons, allowing experts to prioritize criteria through matrices, derive priority vectors, and assess consistency [42]. The methodological steps are summarized in Figure 2.

The hierarchical model was developed based on previous studies [3,37] and the identification of gaps in existing bikeability indices. The model comprised seven main criteria: surrounding environment, infrastructure, safety, environmental quality, multimodality, shared systems, and digital systems, each broken down into operational subcriteria. The structure encompassed both traditional dimensions and new approaches, including criteria such as Environmental Quality (air pollution, noise, and allergenic factors), Shared Systems (coverage and integration of bike-sharing systems), and Digital Systems (information, technological features, and usage costs). Figure 3 presents the hierarchy.

The process began with the elaboration of the pairwise comparison matrices. Each expert was instructed on how to compare all elements at the same level in terms of their importance to the component immediately above. These comparisons utilized Saaty’s fundamental scale, which ranges from 1 to 9 [42]. The method relies on a square reciprocal matrix of pairwise judgments (1), where each element $a_{ij}$ represents the relative importance of element $i$ over $j$. By definition, $a_{ij}={\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$a_{ij}$}\right.}$ and $a_{ij}=1$.

$$A=\left[\begin{array}{cc}\begin{array}{ccc}\begin{array}{c}\begin{array}{c}1\\ {\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$a_{12}$}\right.}\\ {\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$a_{13}$}\right.}\end{array}\\ \begin{array}{c}⋮\\ {\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$a_{1n}$}\right.}\end{array}\end{array}& \begin{array}{c}\begin{array}{c}{a}_{12}\\ 1\\ {\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$a_{23}$}\right.}\end{array}\\ \begin{array}{c}⋮\\ {\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$a_{2n}$}\right.}\end{array}\end{array}& \begin{array}{c}\begin{array}{c}{a}_{13}\\ a_{23}\\ 1\end{array}\\ \begin{array}{c}⋮\\ {\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$a_{3n}$}\right.}\end{array}\end{array}\end{array}& \begin{array}{cc}\begin{array}{c}\begin{array}{c}\dots \\ \dots \\ \dots \end{array}\\ \begin{array}{c}\ddots \\ \dots \end{array}\end{array}& \begin{array}{c}\begin{array}{c}{a}_{1n}\\ a_{2n}\\ a_{3n}\end{array}\\ \begin{array}{c}⋮\\ 1\end{array}\end{array}\end{array}\end{array}\right]$$

(1)

The resulting priority vector $w={\left[w_1,w_2,\cdots ,w_n\right]}^T$, representing the relative weights of the elements, is obtained by solving the principal eigenvalue Equation (2):

$$A·w={\lambda }_{max}·w$$

(2)

This relationship establishes that the derived weights reflect consistent judgments when ${\lambda }_{max}=n$. In practice, however, slight inconsistencies are expected. The value of ${\lambda }_{max}$ is therefore used not only to verify the coherence of preferences but also as the basis for calculating the consistency indices described in the following section. This process ensures that the weight accurately reflects the expert’s pairwise preferences while preserving the reciprocal properties of the comparison matrix. The same procedure is applied for all matrices individually, and the derived weight vectors are subsequently assessed for consistency [41,42].

The reliability of individual judgments was assessed through the Consistency Index (CI) (3) and Consistency Ratio (CR) (4) [42]. A consistent judgment yields a principal eigenvalue equal to the matrix dimension. Deviations indicate inconsistencies in pairwise comparisons. The Consistency Index is computed as

$$IC={\displaystyle \frac{{\lambda }_{max}-n}{n-1}}$$

(3)

where ${\lambda }_{max}$ is the largest eigenvalue of the matrix, and $n$ is the number of compared elements.

To contextualize the magnitude of this deviation, the Consistency Ratio is calculated by normalizing the CI against a Random Index (RI), which represents the average CI of randomly generated matrices of the same order:

$$CR={\displaystyle \frac{IC}{RI}}$$

(4)

According to Saaty’s threshold, a Consistency Ratio below 0.10 is considered acceptable, indicating that the judgments made are coherent [42]. This consistency check enhances the reliability of individual prioritizations. It lays the groundwork for combining pairwise comparison matrices into a collective group matrix, which is essential for effective group decision making. The geometric mean method, which preserves the reciprocal property of judgments, is usually employed in this aggregation and is widely recommended in the literature as it attenuates the influence of individual extreme judgments [41,42].

The aggregation was performed by computing, for each $w_i$, the geometric mean of the corresponding weights provided by the target group of experts. This method avoids reconstructing the group matrix and directly produces a vector that reflects the shared judgment of the panel. The aggregated weight for each criterion is calculated as shown (5):

$$w_i={\left(\prod _{j=1}^n{a}_{ij}\right)}^{{\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$n$}\right.}}/\sum _{k=1}^n\left[{\left(\prod _{j=1}^n{a}_{kj}\right)}^{{\displaystyle \raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$n$}\right.}}\right]$$

(5)

where $a_{ij}$ denotes the weight assigned to criterion $i$ by expert $j$, and $n$ is the number of experts. The aggregated priority vector thus represents the consensus judgment across individual perspectives. This procedure preserves the relative importance across experts while maintaining scale consistency [42]. A detailed example of matrix aggregation is provided in Appendix A.

3.3. Individual Judgement Consistency and Group Consensus

The following sections present results from aggregating expert judgments and analyzing consistency and consensus across individual and group evaluations.

The overall consistency of individual judgments was deemed satisfactory. Most participants exhibited Consistency Ratios (CR) below the 0.10 threshold across the matrices. At the first level of comparison (criteria), no expert exceeded this threshold, reinforcing the reliability of the evaluations. However, greater inconsistencies were observed at the second level (subcriteria), particularly within the Infrastructure, Safety, and Environmental Quality matrices. Notably, two extreme cases with CR values exceeding 1.0 were recorded in the Environmental Quality matrix for experts 3A and 19B. Despite these outliers, all expert judgments were retained in the analysis. This choice aligns with AHP best practices, as employing the geometric mean aggregation method helps to soften the impact of inconsistent responses. Furthermore, each expert was used throughout the entire hierarchy, encompassing both criteria and subcriteria levels, which rendered the selective removal methodologically inappropriate. A comparative test excluding these two experts confirmed that the overall hierarchy and priority rankings remained stable, demonstrating the robustness of the aggregation approach.

Figure 4 presents boxplots depicting the distribution of Consistency Ratios (CR) across all comparison matrices, highlighting the variability in individual judgments and the proportion of values within the acceptable threshold (CR ≤ 0.10). The horizontal axis is limited to 0.7 for visual clarity. Values above 0.6 exist and were discussed above. Boxplots for the Consistency Index (CI) were not included, as they reflect the same pattern.

Individual judgments were aggregated using the geometric mean, a method that mitigates the impact of outlier values [42]. A table of CR and IC values for each judgment is available in Appendix B.

3.4. Criteria-Level Results

The global priorities based on the aggregation of responses from 30 experts reveal clear distinctions in how the criteria were ranked. Safety (0.297) and Infrastructure (0.256) are the most influential factors, together accounting for over half of the total weight. In contrast, Shared Systems (0.088) and Digital Systems (0.051) received the lowest scores. The aggregated comparison matrix was highly consistent (CR = 0.004; CI = 0.00528).

In addition to analyzing the overall sample, this study examined how each group prioritized the main criteria of the assessment model. This segmentation provides a more nuanced understanding of how different stakeholder profiles assign importance based on their backgrounds and experiences.

Table 1. Weights assigned to the main criteria.

	Global	Academic	Technical	Users
Safety	0.297	0.290	0.261	0.346
Infrastructure	0.256	0.266	0.199	0.330
Surrounding Environment	0.106	0.089	0.124	0.101
Multimodality	0.104	0.117	0.121	0.062
Environmental Quality	0.097	0.118	0.094	0.070
Shared Systems	0.088	0.072	0.136	0.056
Digital Systems	0.051	0.048	0.066	0.036

Bold indicates the highest weight; darker shades represent higher values.

presents each group’s normalized weights for the first-level hierarchy. While there are areas of convergence across the groups, several meaningful differences emerge, reflecting their distinct professional perspectives.

The table design features a horizontal division separating the aggregation from the three expert groups. ‘Global’ represents the prioritization of all 30 expert responses, while the other columns show results by stakeholder profile. Color gradients indicate the weight distribution within each group independently. Bolded values highlight the highest weight assigned to each subcriterion across the expert groups.

The comparative analysis of the three expert groups highlights differing priorities for bikeability criteria based on professional backgrounds. Safety (0.297) and infrastructure (0.256) emerged as core dimensions, accounting for over half the total weight. User representatives and cycling advocates assigned the highest values to both, reflecting daily experience with risk and infrastructure quality. Academic experts followed closely, while technical professionals attributed slightly lower weights, indicating a more balanced planning-oriented view.

A second set of contextual criteria, Surrounding Environment (0.106), Multimodality (0.104), and Environmental Quality (0.097), ranked at an intermediate level, with notable variation in group valuations. Technical professionals prioritized the Surrounding Environment (0.124), likely due to its relevance for infrastructure planning, whereas users (0.101) and academics (0.089) gave it less emphasis, possibly favoring structural aspects. Regarding Multimodality, technical professionals again assigned the highest weight (0.121), followed by academics (0.117), with users significantly lower (0.062), suggesting limited engagement with strategic integration across modes. In terms of Environmental Quality, academics attributed the most significant importance (0.118), focusing on air and noise pollution. At the same time, technical professionals (0.094) and users (0.070) appeared more focused on immediate infrastructure and safety concerns.

Shared Systems (0.088) and Digital Systems (0.051) received the lowest global weights but were valued differently across groups. Academics assigned the highest weights to both Shared Systems (0.121) and Digital Systems (0.074), indicating openness to emerging mobility trends and technological integration. Users followed (0.094 and 0.048), likely reflecting personal exposure and practical engagement, whereas technical professionals provided the lowest values (0.045 and 0.028), possibly due to concerns about feasibility or operational complexity.

In summary, while strong convergence was observed around the structural foundations of bikeability, namely, Safety and Infrastructure, the analysis reveals meaningful divergence in how complementary and emerging factors are valued. These findings underscore the importance of incorporating multiple expert viewpoints when evaluating cycling conditions in urban settings.

3.5. Subcriteria-Level Results

3.5.1. Weight Distribution in Surrounding Environment

The prioritization of subcriteria under the Surrounding Environment criterion reveals how experts perceive the role of contextual factors in urban cycling. Among the four elements, Land Use received the highest emphasis (0.320), followed by Climatic Protection (0.290) and Adjacent Green Areas (0.270). Water Bodies were rated as the least relevant (0.119), indicating minimal influence on bikeability (

Table 2. Weights assigned to the subcriteria of the Surrounding Environment.

	Global	Academic	Technical	Users
Land Use	0.320	0.404	0.216	0.363
Climatic Protection	0.290	0.290	0.317	0.228
Adjacent Green Areas	0.270	0.210	0.307	0.308
Water Bodies	0.119	0.096	0.160	0.101

Bold indicates the highest weight; darker shades represent higher values.

). Land Use was particularly valued by academic experts (0.404) and user representatives (0.363), emphasizing the importance of land use diversity and active urban frontages in shaping functional and attractive cycling environments. Technical professionals assigned a lower weight (0.216). Climatic Protection was more highly valued by technical professionals (0.317) than the other groups, suggesting a focus on creating sheltered infrastructure, such as tree-lined routes or built shading.

In contrast, frequent cyclists appeared less concerned with such features. For Adjacent Green Areas, technical professionals (0.307) and user representatives (0.308) assigned the highest weights, reinforcing the perceived role of greenery in enhancing route appeal. Water Bodies received the lowest weights across all groups: 0.160 (technical), 0.101 (users), and 0.096 (academics). While such features may enhance urban aesthetics or environmental ambience, they were not considered to influence cycling decisions significantly. The expert panel’s aggregated comparison matrix showed a Consistency Ratio (CR) of 0.011 and a Consistency Index (CI) of 0.0099, indicating strong internal coherence. The small matrix size may have facilitated this alignment, even though the lower prioritization of Water Bodies may have introduced some divergence.

3.5.2. Weight Distribution Across Infrastructure

The assessment of subcriteria under the Infrastructure criterion primarily focuses on the physical features of the cycling network. Segregation Level was the top priority (0.197), followed by Corridor Integration (0.162), Pavement Smoothness (0.147), and Street Geometry (0.147), while Supporting Infrastructure (0.065) and Public Lighting (0.066) were deemed less necessary (

Table 3. Weights assigned to the subcriteria of Infrastructure.

	Global	Academic	Technical	Users
Segregation Level	0.197	0.211	0.198	0.163
Corridor Integration	0.162	0.132	0.184	0.171
Pavement Smoothness	0.147	0.135	0.127	0.200
Street Geometry	0.147	0.161	0.158	0.106
Intersections Typology	0.127	0.143	0.093	0.159
Bicycle Parking	0.089	0.076	0.104	0.086
Public Lighting	0.066	0.091	0.055	0.050
Supporting Infrastructure	0.065	0.051	0.081	0.065

Bold indicates the highest weight; darker shades represent higher values.

). The results underscore the significance of structural components for route safety and continuity, particularly the need to minimize interactions with motorised traffic. Group-level analysis revealed differences in emphasis among experts. Academics rated Segregation Level highest (0.211), followed by technical professionals (0.198) and user representatives (0.163), indicating shared recognition of the importance of segregated infrastructure.

For Corridor Integration, technical professionals assigned the highest weight (0.184), while user representatives and academics rated it lower, suggesting that practitioners place greater emphasis on network continuity. Users prioritized Pavement Smoothness (0.200), reflecting their experiences with surface conditions, whereas other groups rated it lower. Supporting Infrastructure (e.g., maintenance points or e-bike charging stations) received its highest score from technical professionals (0.081). Public Lighting was generally deprioritized, possibly viewed as a complementary rather than core element, with academics assigning it a slightly higher weight (0.091). The panel’s aggregated comparison matrix yielded a Consistency Ratio (CR) of 0.006 and a Consistency Index (CI) of 0.00846, indicating high coherence in judgment.

3.5.3. Weight Distribution Across Safety

Subcriteria for cyclist safety were prioritized as follows: Traffic Exposure (0.282), Crash Frequency (0.256), and Intersection Conflicts (0.179) received the highest weights, while Conflicts with Parking Lanes and Bus Stops (0.083), Crime and Security (0.130), and Conflicts with Building Access Points (0.069) were less prioritized (

Table 4. Weights assigned to the subcriteria of Safety.

	Global	Academic	Technical	Users
Traffic Exposure	0.282	0.299	0.250	0.305
Crash Frequency	0.256	0.260	0.282	0.208
Intersection Conflicts	0.179	0.180	0.154	0.219
Crime and Security	0.130	0.110	0.149	0.137
Conflicts with Parking Lanes and Bus Stops	0.083	0.081	0.098	0.066
Conflicts with Building Access Points	0.069	0.071	0.068	0.065

Bold indicates the highest weight; darker shades represent higher values.

). This distribution highlights concerns with objective physical threats. Traffic Exposure was prioritized across all groups, defined by the presence, speed, and volume of nearby motorized vehicles. It reflects the demand for physical separation and reduced interaction with traffic. Crash Frequency indicates the relevance of historical safety data, while Intersection Conflicts relate to perceived risk at junctions. Meanwhile, subcriteria such as Crime and Security and both types of spatial conflict were considered less influential, possibly due to their subjectivity or context dependency.

Traffic Exposure was consistently prioritized, with user representatives assigning the highest weight (0.305), followed by academic experts (0.299) and technical professionals (0.250). This near-unanimous ranking underscores its perceived immediacy, especially among experienced cyclists. Crash Frequency was rated highest by technical professionals (0.282), slightly ahead of academics (0.260), while users assigned the lowest value (0.208), perhaps reflecting a stronger response to perceived rather than statistical risk. User representatives (0.219) emphasized Intersection Conflicts most, followed by academic experts (0.180) and technical professionals (0.154), reinforcing concern about navigating mixed-traffic environments.

Crime and Security received relatively similar weights across groups: 0.149 (technical), 0.137 (users), and 0.110 (academics), indicating moderate but consistent attention to social safety. The two lowest-ranked subcriteria, Conflicts with Parking Lanes and Bus Stops and Conflicts with Building Access Points, showed close values across all groups, though slightly higher among technical professionals (0.098 and 0.068, respectively). These factors appear to be seen as contextual inconveniences rather than primary safety risks. The panel’s aggregated matrix achieved a Consistency Ratio (CR) of 0.004 and a Consistency Index (CI) of 0.00496, indicating strong coherence across expert perspectives.

3.5.4. Weight Distribution Across Environmental Quality

Expert assessments under the Environmental Quality criterion show strong concern for health impacts, prioritizing Air Pollution (0.531) over Noise Pollution (0.302) and Allergenic Factors (0.168). This indicates a focus on quantifiable environmental risks affecting cyclists’ comfort and safety (

Table 5. Weights assigned to the subcriteria of Environmental Quality.

	Global	Academic	Technical	Users
Air Pollution	0.531	0.514	0.534	0.539
Noise Pollution	0.302	0.266	0.331	0.315
Allergenic Factors	0.168	0.220	0.134	0.146

Bold indicates the highest weight; darker shades represent higher values.

). Air Pollution was the highest priority, reflecting awareness of the dangers associated with exposure to particulate matter, especially in high-traffic areas. Noise Pollution, beyond discomfort, may be seen as a proxy for traffic intensity and urban stress. Allergenic Factors, by contrast, were perceived as less immediate or as having lower attribution to cycling conditions.

Weight distributions across groups were relatively consistent. Academics assigned the highest weight to Air Pollution (0.553), followed by technical professionals (0.511) and user representatives (0.492), showing broad agreement on its critical impact. For Noise Pollution, values were more dispersed: 0.340 (academics), 0.299 (users), and 0.265 (technical), suggesting slightly greater sensitivity among academics and users. Allergenic Factors were ranked low across all groups: 0.106 (users), 0.180 (academics), and 0.224 (technical), indicating a shared view of their marginal role, possibly due to individual variability and seasonal variation. The aggregated comparison matrix showed perfect consistency (CR and CI equal to zero). The relatively small matrix size may have facilitated alignment, supported by the apparent dominance of air pollution in expert judgments.

3.5.5. Weight Distribution Across Multimodality

The subcriteria under Multimodality received closely weighted responses, indicating general agreement on their relevance. Connection to Public Transport was the highest priority (0.373), followed by Bicycle Parking at Stops and Terminals (0.335) and Regulation of Bicycle Transport in Public Transit (0.292) (

Table 6. Weights assigned to the subcriteria of Multimodality.

	Global	Academic	Technical	Users
Connection to Public Transport	0.373	0.466	0.305	0.323
Bicycle Parking at Stops and Terminals	0.335	0.215	0.368	0.293
Regulation of Bicycle Transport in Public Transit	0.292	0.319	0.326	0.385

Bold indicates the highest weight; darker shades represent higher values.

). This emphasizes the importance of integrating cycling with other transport modes. The high value assigned to Connection to Public Transport suggests the need for effective intermodal links, supporting bicycle use over medium to longer distances. Secure parking at terminals was also considered critical, while regulation within transit systems was slightly less emphasized, possibly due to its perceived complexity.

Academics prioritized Connection to Public Transport most (0.466), compared to lower values from technical professionals (0.305) and user representatives (0.323). This may reflect an understanding of Multimodality as a core element of sustainable mobility systems. For Bicycle Parking at Stops and Terminals, technical professionals attributed the highest weight (0.368), consistent with a focus on infrastructure provision. Users followed (0.293), while academics assigned the lowest value (0.215), possibly preferring systemic over facility-based solutions. Regarding the Regulation of Bicycle Transport, weights were more balanced: 0.385 (users), 0.326 (technical), and 0.319 (academics), indicating shared recognition of its role in enabling intermodal travel. The panel’s aggregated matrix showed satisfactory consistency (CI = 0.00754; CR = 0.013), indicating a coherent set of judgments, with some variation possibly influenced by differences in professional background or planning context.

3.5.6. Weight Distribution Across Shared Systems

Subcriteria related to Shared Bicycle Systems (BSS) showed a moderate hierarchy of importance. Coverage of Points of Interest (POIs) was prioritized highest (0.365), followed by Availability of Systems (0.347) and Attributes of the Systems (0.288) (

Table 7. Weights assigned to the subcriteria of Shared Systems.

	Global	Academic	Technical	Users
Coverage of Points of Interest by BSS	0.365	0.313	0.382	0.381
Availability of Shared Bicycle Systems	0.347	0.282	0.319	0.488
Attributes of Shared Bicycle Systems	0.288	0.405	0.299	0.132

Bold indicates the highest weight; darker shades represent higher values.

). This distribution highlights the perceived relevance of spatial reach and accessibility. The prioritization of BSS coverage at strategic locations, such as travel generator zones and transport hubs, suggests that their integration into everyday travel flows is essential. Availability followed closely, reflecting concern with service presence and distribution. System attributes, such as bicycle type, interface, and reliability, were considered somewhat less central, though still relevant.

Among expert groups, user representatives assigned the highest weight to Availability (0.488), likely reflecting the importance of tangible presence in influencing daily use. Academic experts prioritized Attributes (0.405), indicating conceptual attention to BSS design and operation. Technical professionals assigned the highest value to POI Coverage (0.382), aligning with a planning logic favoring integration with key activity nodes. Despite differing emphases, the three groups assigned relatively similar weights to POI Coverage: 0.313 (academics), 0.382 (technical), and 0.369 (users), showing convergence on this spatial dimension. Conversely, user representatives assigned Attributes the lowest weight (0.132), indicating that functional features are secondary to availability in daily use contexts. The panel’s comparison matrix showed perfect consistency (CR and CI equal to zero), possibly facilitated by the limited number of elements in the matrix and the shared valuation of system coverage in Portugal, where BSS are still developing and are mainly found in larger urban areas.

3.5.7. Weight Distribution Across Digital Systems

The global priorities based on expert assessments under the Digital Systems criterion show a balanced prioritization, with Usage Cost ranking highest (0.299), followed by Service Functions (0.249), Digital System Solutions (0.240), and Trip Information (0.212) (

Table 8. Weights assigned to the subcriteria of Digital Systems.

	Global	Academic	Technical	Users
Usage Cost	0.299	0.354	0.226	0.315
Service Functions	0.249	0.247	0.281	0.188
Digital System Solutions	0.240	0.162	0.332	0.257
Trip Information	0.212	0.237	0.161	0.241

Bold indicates the highest weight; darker shades represent higher values.

). These results suggest that while digital dimensions are not central, they are not dismissed outright in assessing cycling conditions. Comparisons across groups revealed notable differences. Academic experts assigned the highest weight to Usage Cost (0.354), emphasising equity and inclusion in digital access, followed by user representatives (0.315) and technical professionals (0.226). For Service Functions, technical professionals gave the highest weight (0.281), while user representatives assigned the lowest (0.188), possibly perceiving such features as less essential.

Digital System Solutions received the highest weight from technical professionals (0.332), reinforcing their focus on system integration and accessibility. Academic experts assigned the lowest value (0.162). Lastly, Trip Information was prioritized mostly by academics (0.237), with technical professionals (0.161) and user representatives (0.241) assigning similar but slightly lower values. The aggregated matrix showed excellent consistency (CI = 0.0009; CR = 0.001), with only slight variations observed among expert judgments.

4. Discussion

The expert panel’s evaluation shows consistent prioritization patterns, providing insight into the valuation of both traditional and emerging factors when assessing urban bikeability. Even with varying professional backgrounds and experiences, there was a notable agreement on several key dimensions. Safety and infrastructure were the most prioritized criteria across all groups, reaffirming their foundational role in cycling policy.

The second tier of criteria, comprising Surrounding Environment, Multimodality, and Environmental Quality, reflects a growing awareness of contextual and integrative factors. Although not perceived as central, these elements shape the broader cycling experience in urban areas; features related to this criterion contribute to usability and long-term sustainability. Air pollution emerged as the most influential subcriterion within Environmental Quality, underscoring its significance for planning low-emission routes and safeguarding public health.

In contrast, Shared Systems and Digital Systems received lower weights, suggesting more reserved expectations regarding these emerging components. Regarding shared mobility, the limited presence of bicycle-sharing systems (BSS) in Portugal may have influenced participants’ evaluations. Still, these systems represent an emerging dimension of bikeability, with potential to broaden the functional scope of cycling, diversify user profiles, and enhance flexibility and connectivity in urban mobility. Despite their potential to expand access and promote cycling, such systems remain heavily dependent on the quality of cycling infrastructure and the perceived safety of the network. As such, their effectiveness is seen as contingent rather than autonomous.

For Digital Systems, the subcriteria selected may not have fully captured the potential relevance of technology to bikeability. As currently framed, the focus was placed on the cost of use, service features, and trip data, elements that, although useful, are often considered peripheral by frequent cyclists. It is plausible that if digital features had included aspects directly linked to infrastructure or safety, such as environmental sensing systems, real-time hazard alerts, or advanced rider-assist functions, these elements might have received greater attention. For example, cyclist-carried devices capable of receiving accident warnings or systems akin to vehicular driver-assistance tools could be perceived as strong enablers of safety and confidence in complex urban settings.

Furthermore, for experienced cyclists or those familiar with local travel routines, navigational apps are unlikely to be used in day-to-day travel unless route modifications are needed. In this context, the perceived utility of basic digital navigation functions without contextual alerts may have contributed to the lower prioritization of this criterion. The distinctions among expert groups offer additional detail. User representatives highlighted practical considerations, such as availability, surface conditions, and traffic exposure, which indicate firsthand experience with the cycling environment. Technical professionals maintained a balanced perspective, concentrating on network design and the feasibility of implementation. On the other hand, academic experts gave higher importance to conceptual and emerging issues, particularly concerning environmental quality, shared systems, and digital integration.

These findings have clear implications for urban mobility planning. They reinforce the need to prioritize high-quality infrastructure as the backbone of cycling systems while integrating environmental and intermodal considerations to enhance user experience. From a policy perspective, understanding the relative importance of each criterion can guide investment strategies: factors with high perceived importance but low current performance may be prioritized for improvement. At the same time, less relevant elements may be addressed in a secondary phase. Conversely, strengthening highly valued aspects that already perform well can generate significant returns in user satisfaction and network performance. Additionally, peripheral, shared, and digital systems innovations should not be overlooked. Their gradual integration, notably when aligned with established infrastructure and safety standards, can support more inclusive, adaptive, and technologically enabled mobility models. In summary, the prioritization patterns observed reflect a balance between foundational needs, contextual considerations, and emerging solutions. Capturing the perspectives of diverse expert groups, including academic, technical, and experiential profiles, is essential to ensure responsive, evidence-based, and forward-looking cycling policies.

5. Conclusions

This study examines how urban and transportation planning experts perceive the relevance of traditional and emerging criteria in evaluating cycling networks. Using the Analytic Hierarchy Process (AHP), we elicited and compared the priorities of 30 specialists from academic, technical, and user-based profiles. The findings offer a structured and multidimensional view of how bikeability is evaluated by stakeholders with diverse backgrounds and viewpoints.

All expert groups prioritized traditional criteria such as Safety and Infrastructure. At the same time, Shared and Digital Systems were perceived as complementary elements, receiving lower weights overall, although academic experts valued them more. The limited prioritization of emerging criteria appears to reflect contextual factors, such as the availability of shared systems or the framing of digital tools around ancillary features rather than core infrastructure or safety. Still, these additions did not significantly alter the hierarchy of priorities.

This study emphasizes the potential of AHP as a transparent and replicable method for evaluating urban cycling conditions, particularly when implemented through digitally supported procedures. From a planning standpoint, it reinforces the importance of investing in robust infrastructure and physical safety. Additionally, contextual and integrative elements such as environmental quality and multimodality should be considered.

Though currently less prioritized, shared mobility systems, e-bikes, and digital tools challenge traditional bikeability assessment models by introducing dimensions that go beyond physical infrastructure and safety. They expand the functional reach of cycling; diversify user profiles; and incorporate new aspects of attractiveness, flexibility, and connectivity. Their importance may grow as these technologies are more explicitly linked to safety-related functions and embedded within established infrastructure standards. This integration could strengthen connections with other transport modes, attract new cyclists, and support more inclusive and adaptive mobility models. The involvement of various expert profiles was crucial in creating a nuanced and balanced assessment that captures both current limitations and future potential.

However, there are limitations to consider. These include reliance on expert judgment, a geographically restricted sample, and the static nature of perceptions captured at a specific moment. These factors suggest opportunities for further research, such as applying the framework at local levels, conducting comparative studies, integrating empirical data, and using GIS tools to apply prioritization weights to real cycling networks. This approach could help generate spatialized bikeability indices and inform targeted interventions. For example, integrating expert-weighted criteria with geospatial data such as cycling flow patterns, accident locations, infrastructure coverage, and environmental quality indicators could enable location-specific planning and policy development.