Infrastructure: what has been found to work?

The purpose of this section is to lay out what is known to be successful with infrastructure related schemes and how we can learn from this.

  • Walking – Infrastructure was found to have little effect barring increasing the perception of safety for walking with children. Rather, land-use mix, walkable distances and access to public transport were found to be greater predicators of walking.
  • Bike Sharing Schemes – Bike sharing schemes have been shown to increasing cycling modal share in areas with low cycling uptake. However, success is often linked to existing cycling infrastructure and shared cycling docking density with most new trips coming at the expense walking and public transport rather than motor vehicles.
  • Cycling Lanes – The evidence suggests that individual lanes have a negligible effect, rather significant effects are found where a continuous connected network is formed. They also note that cycling infrastructure is a necessary but not sufficient condition to achieve modal shift in isolation, although once modal shift has been achieved it is heavily used by cyclists.
  • Cycling and Congestion – Evidence from Germany suggests that the speed of motor vehicle traffic has a large effect on cycling uptake, with slower moving traffic and higher congestion resulting in a greater increase in leisure cycling. Cycling speed and congestion was also found to have an effect, suggesting that the value of time influences modal shift significantly.
  • Multi Modal Integration and Cycle Parking – Integrating cycling with the public transport system in the Netherlands was found to increase cycling to train station as well as train modal share with a much more muted effect for buses.
  • Gender Gap – Female cyclists self-report a preference for segregated on-road cycle paths. This appears to be corroborated by a study drawing on data from a travel app to analyse cycle route choice by gender that observes female cyclists changing their route to include segregated on-road paths.
  • Cycling City and Towns (CCT) Evidence – Evidence from a large cycling scheme in England noted that awareness of off-road cycle paths was lacking, and that advanced stop lines were popular with regular cyclists but less so with motorists. However, new cyclists were still observed dismounting for junctions instead. They also noted that whilst cycling parking was helpful in encouraging cycling, cycle theft was not tackled and remained high on people’s minds with one respondent stating it prevented them from cycling into town.
  • Expert Interviewee Advice – Advice from several cycling practitioners noted that interventions should be planned to last at minimum 5 years, with shorter schemes having little practical value. They also noted that whilst the mix of revenue to capital spend was not especially relevant, revenue spend should always be a part of any capital spend.


Infrastructure has a much more muted effect on walking than cycling (Forsyth & Krizek, 2009). When analysing travel diaries and built environment using GIS, Forsyth and Krizek (2009) noted that land-use mix was a significant predictor of walking trips, and that built environments with destinations closer together was correlated with higher rates of walking for transportation. In general, infrastructure investments may be useful in changing the perceptions of walkability among children and their parents. However, for adults the relationship between infrastructure and increased travel walking is not strong (Forsyth & Krizek, 2009).

Knuiman et al. (2014) also found that density and land use were among the key predicators of walking for travel, although they also note that the number of destinations that are accessible as well as the number of bus and train stops within 1.6km were significant predicators. Individuals who had 30 or more bus stops within 1,600 m of their homes were almost twice as likely to walk for transportation than those who had 0–14 bus stops. The presence of a train station within 1,600 m increased the likelihood of walking for transportation by approximately 50%.

Bike sharing schemes

There has been some success in individual cycling projects, particularly in bike sharing schemes. A paper looking at the successes in cycling projects in Spain by Marqués, et al. (2015) and Anaya & Castro (2012), notes that their success was fairly uneven but the one main factor that was noted as a difference between those succeeding and those not was the density level of the bike stations as well as other present infrastructure. It appears that further increasing the density and having a cycle friendly city aided in the schemes becoming successful, suggesting that there needs to be a critical mass in order for the schemes to become successful. The literature from Spain recommends a docking station every 300 metres in order for a bike sharing scheme to have a strong chance of success (Anaya & Castro, 2012).

The Danish Cycling Embassy (2012) also notes that cities such as Lyon, Paris and Barcelona managed to boost bicycle traffic significantly by introducing public bikes, but they attribute this to the fact that hardly anyone used a bicycle before.

However, it is important to consider where these new journeys come from. Fishman, et al. (2015) analysed data from cycle hire schemes and concluded that where bike share trips replace existing trips, the main mode shift comes from public transport, then walking, followed by cars, private bikes, and finally taxis as shown in Figure 1.

Figure 1 – Cycle hire trips by original mode of transportation
Figure 1 – Cycle hire trips by original mode of transportation

Source: (Fishman, et al., 2015)

Cycling lanes

The Danish Cycling Embassy (2012) notes that “Danish bicycle counts show that installing a cycle lane on a single road has a negligible effect, a mere 0.5% increase. If, however, an entire network of cycle lanes is established in an urban area without cycle tracks the number of cyclists will increase.” As such it seems that fixed infrastructure interventions have very little impact in isolation but when combined have a significantly larger impact. This is echoed by other works and evidence reviews (Pucher, et al., 2010; DfT, 2016). Pucher, et al. (2010) goes on to say that some specific programs which appear to have a negligible impact when examined in isolation can have a significant impact when implemented comprehensively. Indeed the most compelling evidence came from communities that implemented a fully integrated package of strategies to increase cycling such as city wide continuous segregated lanes. These can be seen under ‘Successful Projects’.

Interestingly it appears there is a significant lag in effect from any cycling infrastructure. Goodman, et al. (2014) found that living nearer the infrastructure did not predict changes in activity levels at 1-year follow-up but did predict increases in activity at 2 years relative to those living farther away. This feeds in to other studies that suggest that infrastructure is a necessary but not sufficient condition for increases in cycling. Rather it is disruption in people’s lives due to new work or moving house that has an effect which is more successful in areas which already have infrastructure, thus explaining the lagged effect.

Beenackers, et al. (2012) looked at built environment factors effect on cycling uptake when moving to a new home in Australia. They found that residential density, number of recreation destinations nearby and access to a park were all positively associated with cycling uptake for transport, with parks in particular resulting in a strong influence towards cycling. However, the survey did not include cycling infrastructure either due to the lack of infrastructure at the location or simply by omission. Regardless it shows that the environment factors are significant in helping trigger points be effective.

Using longitudinal panel data, Song, et al. (2017) found that self-reported use of cycling infrastructure (such as casual or leisure cycling) was significantly associated with a modal shift towards active travel after controlling for personal and household characteristics, but that passive exposure (residential proximity to the infrastructure) was not directly associated with a modal shift. Other studies have argued that the provision of dedicated cycling facilities is critical for achieving a higher level of cycling, based upon aggregate international data (Pucher, et al., 2010). Song, et al.’s (2017) study suggests that while infrastructure provision may not be a sufficient condition to achieve modal shift, it may well be a necessary condition, as those who casually used cycling infrastructure became more regular cyclists over those who did not.

Dill (2009) used GPS to collect trip route data from 166 regular cyclists in Portland, Oregon, over seven days (1,800 trips). The cyclists were selected from a sample of cyclist survey respondents who were stratified by geography and sex, and then randomly selected to receive GPS trackers. It was found that almost half the miles of bicycle travel ‘occurred on-roads with bicycle lanes, [separated] paths, or bicycle boulevards’, facilities that made up only 8% of the available road network. A follow-up survey indicated that the top reasons for choosing paths were directness and avoiding traffic.

Cycling and congestion

Goestzke & Rave (2011) looked at the effect of city congestion on cycling modal share. They note that the more congested roads are, the more people use bicycles for shopping and errands. These shopping trips are most affected by changes in average motor vehicle speed. The evidence suggests that a 10% increase in mean automobile speed, for example from the a mean of 20 km per hour to 22 km per hour, would halve the modal split for shopping cyclists, from about 27% to 13.5%. However, commuter cyclists did not appear to be responsive to the change in average motor vehicle speed.

Other factors affecting the success of network systems include a reduction in the stops along a cycle route increases cycling. Rietveld & Daniel (2004) found evidence from Holland that suggests that routes which do not require cyclists to make stops increase cycling mode share: 0.3 fewer stops per km along a route was associated with a 4.9% higher share of cycling. As is shown above the mean speed is an important factor in modal shift, keeping car speeds stagnant while improving bicycle speeds appears to increase the cycling share.

Multi modal integration and cycle parking

Cycling integration into the public transport system can increase use of both systems. A pilot project in the Netherlands in the 1990s found that there was a significant increase in both public transport use and bicycling, but mainly for bicycle trips between home and the suburban rail station (access trip) and far less for bicycle trips between the terminal station and the activity end of the trip (egress trip) (Pucher, et al., 2010). This was achieved by the increased provision of bicycle parking both at working and at public transport hubs (Martens, 2007).

They also found that bike lockers at bus stops were hardly used by bus passengers which they attributed to the large number of bus passengers being students and the high price of lockers relative to the value of the bike (Martens, 2007), suggesting that cheaper bike parking at bus stops would be better or simply focus on providing bike lockers at train stations. An interesting point to note is that standard cycle parking is more popular among users than bicycle lockers, which tend to be under-utilised (although it is worth noting that the evidence for this comes from Holland and may not apply to the UK).

Gender gap

Women self-report a preference for segregated on-road cycle lanes over mixed traffic cycling (Pucher, et al., 2010; Moore, 2019). However the preference for segregated cycling paths does not extend to off-road cycle paths, with safety concerns being cited as a possible reason. This appears to be reflected in analysis conducted by Strava where they investigated the effect of new segregated lanes on route choice. However, the analysis should be heavily caveated as it fails to disclose key data such as the sample size and demographics, as well as drawing data from a voluntary app which could lead to selection bias.

A study that utilised Strava data of cyclists in New York showed that women are a lot more attracted to segregated safe cycling paths than mixed traffic (Moore, 2019). “Strava’s study analyzed 11,416 bike trips by women in Queens between 2015 and 2019. The results were clear: Making streets safer increased women’s use of them dramatically. In 2019, women made nearly 40% more trips on 43rd and nearly 50% more trips on Skillman than they had in 2018 (two streets with new segregated lanes)” (Moore, 2019). Although this increase was predominantly displacement from other streets, it was noted that these streets were being avoided prior to this, suggesting that the new lanes allowed for more direct cycling for many cyclists, thus making cycling a more attractive prospect. Additionally it was noted that men’s ridership did not change after the bike lane upgrades. A separate review by Strava in Philadelphia found the same results, with female cyclists moving to the segregated cycle paths while male cyclists continued on their previous routes (Bunn, 2019).

Expert interviewee advice

Slowman, et al. (2014) conducted interviews with those responsible for cycling in Seville and Odense amongst other successful cycling towns. The table below outlines the experts’ opinions on best practice, indicating among other things that:

  • The programme duration should be at least five years. Long-term funding settlements - whatever the proportion of revenue and capital – are considered of greater value than short term settlements. Five years was considered the absolute minimum programme time with short-term investment programmes considered to have no value.
  • Larger urban areas may require a higher proportion of capital investment than smaller urban areas due to their need for quality infrastructure, with smaller towns expected to only need some minor adaptations to streets such as 20mph zones or traffic calming. If the conditions for cycling are already fairly suitable the priority should be on behavioural spend.
  • Overall scale of Investment: Smaller budgets might involve a higher proportion of revenue than larger budgets as there is little point in building a section of path that is not connected to the rest of the cycle network. In this situation revenue measures would be better suited to create a groundswell of support in preparation of future larger spends.
  • Cultural starting point is important. The proportion of revenue funding may need to be higher in places which lack a cycling culture.
  • The proportion of revenue funding may change during the course of the programme. Being high at the early and late stages of a long-term (20-30 year) programme due to behaviour interventions, but lower in the middle period due to infrastructure interventions.
  • Both capital and revenue are important, although the exact split was considered less crucial.

Cycling Cities and Towns evidence (CCT)

Cycling Embassy (2012) carried out research with regular and new cyclists to gauge reaction to interventions introduced through the Cycling Cities and Towns programme. The table below lays out the reaction to hard interventions, noting that cycling infrastructure was well received by cyclists, but proved unpopular with car drivers e.g. advanced stop lines. Cycle parking was found to be useful but there was still significant concerns regarding theft and vandalism with one respondent stating it prevented them from cycling into town.

Cycling infrastructure and facilities

Residents tended to have noticed new cycling infrastructure in their town (both on- and off-road) and valued the improved cycling experience that resulted. Although improved infrastructure was generally viewed positively, there remained negative perceptions of discontinuous routes, narrow lanes and lack of routes segregated from traffic.

It also had an impact on social/cultural and journey perception issues (with visible investment in cycling presenting an image of cycling as a supported, feasible and popular option)

Advanced Stop Lines (ASL)

Non-cyclists: ASL were unpopular as cyclists were sometimes perceived as an obstruction to motorists. There were reports of cyclists damaging vehicles or appearing in vehicle blind spots whilst approaching the turning box.

Cyclists: ASL were positively viewed and enhanced the cycling experience for regular cyclists by reducing potential risks.

Despite the positive views of cyclists, some New Regular cyclists were unsure how to use ASL and reported dismounting and using pedestrian facilities to cross a road.

Awareness of the investment in off-road cycling facilities

In Stoke and Woking, where investment focused on improving canal tow paths, non-cyclists tended to be unaware of the improvements as they were not visible to drivers. Knowledge of off-road cycling facilities was often gained by word of mouth when friends, families and/or colleagues had discussed their experiences of using the facility as a cyclist or pedestrian.

In contrast, in Blackpool, Southend and Southport where strategies included investment in dedicated off-road cycle lanes on the sea front, awareness of the investment was generally high and this was mainly due to improvements being highly visible to all residents.


In all CCTs, infrastructure has been supported by the provision of enhanced signage, often with average cycle times rather than miles to destination. Continuing Regular Cyclists were those most likely to be aware of cycle route signage and so often noticed new signage when travelling on their current cycle routes.

Awareness of route signage was low amongst other cyclist groups. Overall, few participants expressed strong views regarding cycle route signage. Despite being regarded as a ‘good idea’ in principle, most perceived the signs being aimed at visitors or tourists rather than residents.

Cycle Parking

Across all CCTs there was a noted increase in the number of parking facilities in the city/town centres, workplaces, schools and rail stations and this was often attributed to local authority policies to encourage cycling.

The provision of cycle parking facilities within the workplace had contributed to a small number of participants cycling to work (alongside other factors).

Finding a place to park your bicycle was therefore not a problem for many cyclists; however, cycle security was a concern in most CCTs, particularly Chester, Colchester and York, where bicycle theft was perceived as high and many had experienced theft or vandalism. Improvements to cycle parking security were therefore important to many participants.


Infrastructure investments for walking were not seen as particularly necessary, with land planning and strong public transport connections considered more important in achieving modal share.

The evidence for cycling was more comprehensive, suggesting that infrastructure works best as a comprehensive package and not as a system of small improvements. As such the literature does not suggest singular measures but rather a raft of measures to increase cycling, although what share each of the measures should have is not stated. The one exception to this is in the style of cycling lane, with the literature strongly pointing to segregated on or near traffic lanes as being the best to attract female and casual cyclists.

Additionally the evidence suggests that the infrastructure is a necessary but not sufficient factor in isolation to increase cycling. Instead, natural disruptions in the form of moving home, or changing job, along with other modes journey times and environmental factors play a greater role in achieving modal shift. Some of these happen organically and are lagged with time, which can be overcome with some behavioural measures such as personalised travel plans, others will inhibit modal shift if not overcome such as adverse topography and weather.