Transportation Research Part A 130 (2019) 208–226
Contents lists available at ScienceDirect
Transportation Research Part A journal homepage: www.elsevier.com/locate/tra
The death and rebirth of bikesharing in Seattle: Implications for policy and system design Luke Peters1, Don MacKenzie
T
⁎
Department of Civil and Environmental Engineering, University of Washington, United States
A R T IC LE I N F O
ABS TRA CT
Keywords: Bikesharing Seattle Case study Policy Micro-mobility
What factors determine the ridership of micro-mobility systems such as bikesharing or scooter sharing? This paper presents a case study on bikesharing in Seattle, USA, which has the distinction of being both one of the few cities in the world where a modern public bikesharing system (named Pronto in Seattle) has been shut down, and the first US city to permit dockless bikesharing operations. These dockless services produced more bikesharing trips in four months than Pronto did in its entire two-and-a-half-year run, and in their first year 11 times more rides than Pronto had in its first year. The contrast in performance between these two systems provides a unique opportunity to test alternative theories for why Pronto’s ridership struggled where others have succeeded, offering more general insights into how system designers and regulators can avoid repeating the pitfalls of Seattle’s original bikesharing launch. This case study triangulates qualitatively between popular press reports, interviews with key stakeholders, an original survey of bikesharing users, and ridership data from multiple bikesharing systems to evaluate the contributions of eleven factors to Pronto’s low ridership. It concludes that the most important reasons for Pronto’s struggles were inadequate system scale, station density, geographic coverage area, ease of use, and pricing structure. Critically, these factors all represent explicit choices made by system designers and policymakers, rather than local market or environmental factors beyond their control. The higher ridership of dockless bikesharing in Seattle appears primarily due to differences in these factors, which are not necessarily exclusive to dockless services. The paper closes with a discussion of how policymakers can avoid condemning emerging micromobility services to Pronto’s fate of low ridership.
1. Introduction Bikesharing in the United States and globally is at a crossroads, between traditional docked systems and partially or fully dockless systems. The former typically have been owned and/or subsidized by government agencies or non-profits, while many of the latter are private, for-profit enterprises. The history of bikesharing in Seattle, USA is a microcosm of this evolution. Pronto Cycle Share, Seattle’s docked bikesharing system, was launched in October 2014 by a non-profit, and was eventually shut down rather unceremoniously in March 2017, when the city decided to zero out its budget for the struggling bikesharing system. Bikesharing returned to Seattle several months later, when three private companies - Spin, LimeBike, and ofo - launched their dockless services in the city. These dockless systems generated more bikesharing trips in Seattle in their first four months than Pronto did in its entire two-
⁎
Corresponding author at: Box 352700, Seattle, WA 98195-2700, United States E-mail addresses:
[email protected] (L. Peters),
[email protected] (D. MacKenzie). 1 331 Locust St, Fort Collins, CO 80524, United States. https://doi.org/10.1016/j.tra.2019.09.012 Received 26 November 2018; Received in revised form 8 August 2019; Accepted 9 September 2019 0965-8564/ © 2019 Published by Elsevier Ltd.
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 1. Timeline of Pronto Cycle Share’s history.
and-a-half-year run (Fucoloro, 2017h). Spin and ofo have since exited the Seattle market, while Jump has entered and continues to compete with Lime. This paper aims to understand why Pronto’s ridership languished where others have grown, as a case study to better understand how bikesharing and other micro-mobility systems can achieve sustained high ridership. It relies on interviews with individuals who have had close professional involvement with bikesharing in Seattle, a survey of Seattle bikesharing users, and ridership data from each system. Factors such as weather, topography, the city’s bike network, a mandatory helmet law, ease of use, pricing, transit competitiveness, system scale, geographic coverage, station density, and press coverage are examined to determine how they affected bikesharing ridership on Pronto and on the new dockless systems. 2. Background Fig. 1 provides a timeline of Pronto Cycle Share’s history in Seattle. Puget Sound Bike Share, a non-profit organization, was formed in 2012 by a coalition of government and private groups to launch and operate a regional bikesharing system for Seattle, USA (Fucoloro, 2012). Puget Sound Bike Share was guided by a business plan developed by Alta Bicycle Share, which planned an initial launch of a docked bikesharing system (Phase 1A) with 50 stations and 500 bikes in Seattle’s downtown, South Lake Union, Capitol Hill, and University District neighborhoods. The business plan projected first year ridership of 446,000 trips or 2.4 trips per bike per day, with 4000 annual members and 20,500 casual users providing $860,000 in user revenue (Alta Planning + Design, 2012). The plan also called for an expansion in year two (Phase 1B), which would add 60 more stations and 600 more bikes, slightly expanding the coverage area and increasing density to an average station spacing of 300 m, with further expansions (Phases 2 and 3) to follow. To address concerns related to King County’s helmet law, the plan included helmet vending machines at each station, and accounted for a lower usage rate (Alta Planning + Design, 2012). In 2012 and 2013, the system received much of the grant funding and sponsorships needed for a launch and Puget Sound Bike Share contracted with Alta Bicycle Share (which was renamed Motivate in 2015) to build and operate the system (Fucoloro, 2013). The Pronto Cycle Share system officially launched on October 13, 2014 with 50 stations and 500 bikes similar to Phase 1A proposed in Alta’s business plan (Fig. 2) (Fucoloro, 2014). Productivity, measured in trips per bike per day, is a widely used bikesharing performance metric (Alta Planning + Design, 2012; Bachman, 2015; Fishman et al., 2013; Fucoloro, 2015a,b; Motivate International Inc, 2018a,b,c; NACTO, 2019). In its first full year, Pronto tallied 144,000 trips, or about 0.8 trips per bike per day, with the peak ridership week occurring in July 2015 with about 1.3 trips per bike per day (Fucoloro, 2015b). These ridership numbers were similar to first year numbers from similarly sized systems in Denver (0.9 trips per bike per day) and San Francisco (1.2), but certainly lagged behind first year numbers of larger systems in New York City (4.3), Washington DC (2.5), and Chicago (1.5) as well as projections made by Alta’s original business plan (2.4) (Alta Planning + Design, 2012; Bachman, 2015; Motivate International, 2018a,b,c). Even before Pronto reached its first year the City of Seattle had proposed a major expansion of the system. Part of this proposal also included a plan for the City to contract directly with the system operator Motivate, rather than having the non-profit Puget Sound Bike Share contract with the operator (Fucoloro, 2015a). However, Seattle did not win a Federal grant for $10 million to expand Pronto, and the city instead had to plan to expand the system using $5 million in city funds and $1 million from Motivate (Fucoloro, 2015c). Meanwhile, Puget Sound Bike Share, which still owned the system and had been preparing for the planned city takeover of the system since May 2015, had lost its staff and much of its board by late 2015 and thus did not obtain new sponsorships or grants needed to keep the system financially viable (Fucoloro, 2016a). In January 2016, the Seattle Department of Transportation (SDOT) proposed to City Council spending $1.4 million of the 209
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 2. Pronto Bike Share Station Map (color) showing station locations at system launch.
$5 million budgeted for Pronto on purchasing Pronto’s assets from the non-profit to keep the system afloat (Fucoloro, 2016a). In March 2016, after much debate, Seattle City Council voted for the city to buy Pronto’s assets (Fucoloro, 2016b). Around this time annual Pronto membership experienced a steep decline from a peak membership from more than 3000 in September 2015 down to nearly 1900 in February 2016 (Fucoloro, 2016c). Since Pronto launched in October 2014, the first annual subscribers’ passes expired in October 2015. Given that most of this membership decline - a loss of more than 900 subscribers - occurred in October 2015, it appears to be due to nonrenewals by the original adopters. The low renewal rate was attributed to the October launch of the system as opposed to a spring launch, initial members who didn’t end up finding the system useful enough, and doubts about the system’s future during the time City Council debated whether to buy the system (Fucoloro, 2016d). After the City took over the Pronto system, they began working on their plans to expand the system. The City received six bids from bikesharing vendors, including one from Pronto operator Motivate, but the winning expansion bid was from Quebec based company Bewegen with 100 stations and 1200 bikes, all of them with electric-assist to aid riders climbing Seattle’s many hills (Fucoloro, 2016e,f). This meant the existing Pronto system would be shut down and the city would sell the equipment to make way for the new all electric-assist fleet (Fucoloro, 2016e). However, in January 2017 Mayor Murray scrapped the bikesharing expansion 210
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
plan before it ever went in front of City Council. The bikesharing expansion funds were reallocated to Pronto decommissioning, Safe Routes to School projects, and other biking and walking projects. On March 31, 2017 this left Seattle without any bikesharing program just as many other US cities were launching systems, and made Seattle one of the few cities in the world where a modern public bikesharing system has been shut down (Fucoloro, 2017a). After two and a half years of operation Pronto had carried 278,143 total trips or about 0.6 trips per bike per day (Fucoloro, 2017h). While this relatively low amount of ridership is not unheard of for similarly sized systems (for example, systems in Pittsburgh and Milwaukee each with 500 bikes carried about 0.4 trips per bike per day in 2016), it was certainly disappointing for a bicycle friendly city like Seattle, especially considering the optimistic ridership projections from the Alta business plan (Healthy Ride Pittsburgh, 2018; Midwest Bikeshare Inc, 2018). After the saga and eventual shutdown of Pronto the city had no plans to revisit public bikesharing in the immediate future, but less than a month after the end of Pronto there was already interest from private dockless bikesharing companies to launch in Seattle (Fucoloro, 2017b). In June 2017, SDOT released a draft version of its dockless bikesharing pilot rules, and ten companies expressed interest in launching in Seattle (Fucoloro, 2017c). In July 2017, Spin and LimeBike became the first companies to apply for bikesharing permits and both launched in Seattle later that month, with ofo following in August (Fucoloro, 2017d,f). All three systems offered $1 rides for each 30 min on Spin or LimeBike, or 60 min on ofo, with services often offering free rides with sign-up (Fucoloro, 2017e). Within months, these operators had deployed thousands of bikes throughout Seattle (Fucoloro, 2017d). Spin in its first week achieved over 5000 trips, beating out Pronto’s best week by 300 trips with the same number bikes (Fucoloro, 2017e). By the end of November, Spin, LimeBike, and ofo had combined for 347,300 trips, eclipsing Pronto’s two and a half year total of 278,143 trips in just over three months (Fucoloro, 2017h). Altogether, dockless bikesharing in Seattle totalled over 1.7 million trips in its first full year, putting Seattle among the national leaders in total bike share trips (NACTO, 2019; SDOT, 2018b, 2019a,b,c,d,e; personal communication with Seattle DOT staff). This represents more than 11 times Pronto’s first year ridership of 144,000 (Fucoloro, 2015b). Pronto ultimately shut down due to a mutually reinforcing combination of financial struggles, political, and organizational challenges. While there were many underlying reasons for this, a central cause was low ridership that deprived the system of both revenues and stakeholder support. Therefore, this paper focuses on evaluating the factors that may have led to Pronto’s low ridership. 3. Possible explanations for Pronto’s low ridership Bikesharing experts and articles covering Seattle bikesharing have offered many possible explanations for Pronto’s low ridership. The eleven factors considered in this case study are based on a review of the bikesharing research literature, press reports covering Pronto, and Seattle bikesharing experts who were interviewed for this study. A 2011 review of the bikesharing literature concluded that although bikesharing systems show substantial variability in ridership, ranging from about 0.2 to more than 5 trips per bike per day, the reasons for this variation have not been fully explored (Fishman et al., 2013). The factors commonly identified as most important to bikesharing ridership are station density, system scale, ease of use, and coverage area, all of which relate to the proximity of stations to trip origins/destinations and system convenience (BachlandMarleau et al., 2012; ITDP, 2013; NACTO, 2015). Bachand-Marleau et al. (2012) examined factors impacting the likelihood of being a shared bicycle user and frequency of use of the BIXI program in Montreal and found that having a docking station close to home had the greatest influence on being a bikesharing user, with the presence of a station within 500 m of home increasing the likelihood of use by more than three times. The study also found that proximity of docking stations to destinations also increased odds of being a bikesharing user, but not as much as proximity to origins. Other factors found to be influential in increasing bikesharing use were the transportation habits of potential users, with transit users, people who combine cycling and transit for trips and those with a driver’s license more likely to use bikesharing, as well as individuals interested in the avoidance of private bike theft and maintenance, and individuals who like the design of bikesharing. A study of a low use bikesharing program in Brisbane, Australia, which featured ridership of 0.25 trips per bike per day in 2011, showed that users and potential users found a lack of accessibility and the ability to take spontaneous trips to be significant barriers to use, as a result of a mandatory helmet law in Australia, overnight closure of the system, and inability to sign up easily with a credit card swipe. Respondents also cited safety issues due to perceived limited motorist awareness and bicycle infrastructure to be barriers to use (Fishman et al., 2013). In 2013, the Institute for Transportation & Development Policy (ITDP) released The Bike Share Planning Guide (ITDP, 2013). Citing seven of the highest ridership systems around the world as models, the report identifies five elements critical to drive up key bikesharing metrics: (1) station density, with 10–16 stations per square kilometer and an average station spacing of 300 m; (2) bikes per residents, with 10–30 bikes per 1000 residents within the coverage area; (3) coverage area, with a recommended minimum area of 10 square kilometers; (4) quality bikes that are durable, attractive, and practical; and (5) easy to use stations that provide a simple bicycle check out process. Similarly, the National Association of City Transportation (NACTO) officials published a practitioner’s paper in 2015 (NACTO, 2015) which again indicated station density as a key factor in determining ridership. The report includes an analysis of North American bikesharing systems, which shows that ridership at a station (defined as number of trips that begin or end at the station) increases exponentially the more stations there are within a 15 min bike ride, and recommends a station density of about 28 stations per square mile (about 11 per square kilometer). Convenience, a factor which is closely associated with the proximity of stations to home and other destinations, has consistently 211
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
emerged as a main factor motivating bikesharing use for various programs in North America, China, Europe and Australia (Fishman et al., 2013). NACTO also cited user convenience as a major factor in bikesharing use, based on survey responses of users in New York, Chicago, and Washington DC, and suggests stations should be placed no more than 1000 feet (305 m) apart within the program area, with 1000 feet being the convenient walking distance to a station (NACTO, 2015). Environmental factors such as weather and topography have been shown to play a large role in bicycle usage in general and also for bikesharing. Intuitively, studies from North America have shown that ridership generally increases with temperatures up to 90 degrees Fahrenheit and decreases as rainfall increases, with precipitation being more influential than temperature for biking (Lewin, 2011; Miranda-Moreno et al., 2011; Rose et al., 2011). Research has also shown that hills and steep grades discourage bike use, with cyclists more sensitive to grades than pedestrians, and more experienced riders more tolerant of grades (Cervero and Duncan, 2003). As cited in the study of factors influencing bikesharing ridership in Brisbane, Australia, helmets have been shown to be a barrier to bikesharing use, particularly in jurisdictions with mandatory helmet legislation. The vast majority of bikesharing systems do not provide helmets for users, meaning users have to bring their own if they choose to ride with a helmet (Fishman et al., 2013). In a survey of bikesharing users in Minneapolis-St. Paul, where there are no helmet laws, only 14% of respondents indicated they always wore a helmet while using bikesharing (Nice Ride Minnesota, 2010). Similar results were found in an observational study in Boston and Washington DC, with only 19% of bikesharing riders wearing helmets, a much lower rate than those riding personal bicycles who wore helmets 51% of the time (Fischer et al., 2012). In Melbourne, Australia, where helmets are required, 36% of users indicated difficulty in finding a helmet, and 25% indicated not wanting to wear helmet as reasons they did not use bikesharing (Alta Bike Share, 2011). In a survey of Capital Bikeshare users in Washington DC, only 17% reported always wearing a helmet while using bikesharing, and the main reason cited by users for not wearing a helmet was that their trip was unplanned and they were not carrying a helmet at the time (LDA Consulting, 2012). Two factors possibly impacting bikesharing usage which have received less attention are pricing scheme and relatedly bikesharing’s ability to compete with other transit options in cities. Traditionally, modern bikesharing systems have offered two basic pricing models: a long-term monthly or annual pass; and short-term daily, 3 day, or 7 day passes. Short rides (typically 30 or 60 min) are free with either pass. Shaheen et al. (2014) found in the 2012 season that among 22 US systems annual passes averaged $62.46 and the median was $65, and daily passes averaged $7.77 with a median of $5. Few US systems have offered a more affordable entry point to bikesharing than a $5 day pass. In contrast, 2016 median bus and rail transit fares around the US were $1.70 and $2.25 respectively, making short-term bikesharing passes significantly more expensive (APTA, 2017). Recently, private dockless bikesharing companies have offered bikesharing at a much more affordable and transit competitive entry point of $1 per ride, and often offer free rides to entice new users (Fucoloro, 2017g). The interactions of bikesharing and transit ridership has been studied by several authors. Campbell and Brakewood (2017) found that the introduction of bikesharing in New York City led to a reduction in bus ridership on routes near bikesharing docks. Barber et al. (2018) were able to identify spikes in bikeshare checkouts following train arrivals at adjacent light rail stations in Minneapolis, indicating that passengers were linking train trips with bikesharing. Barber and Starrett (2018) found that more bikesharing trips ending near Chicago’s El (rail) stations was associated with more rail boardings, indicating a complementary effect. They also found that more bikesharing trips originating near a rail station was associated with fewer rail boardings, indicating a substitution effect. While most of these factors are known to affect bikesharing ridership in general, the shutdown of Pronto in Seattle, followed quickly by the entry of private dockless operators, provides a unique opportunity to identify which factors were actually dispositive in Pronto’s low ridership. 4. Methods This paper triangulates qualitatively between data from four sources: (1) press reports documenting the history of bikesharing in Seattle; (2) interviews of experts familiar with bikesharing in Seattle; (3) a survey of Seattle bikesharing users; and (4) ridership data from Pronto and other bikesharing systems. These four sets of data were used to test and corroborate hypotheses, and draw conclusions about the relative importance of various factors in determining bikesharing ridership. This process relies on identifying recurring patterns in the separate data sources, as opposed to a quantitative weighting score. Press reports on bikesharing in Seattle were obtained through web searches, and includes many articles written on the Seattle Bike Blog (which has covered the topic since before Pronto’s launch through to the present) and The Seattle Times (a leading local newspaper). Semi-structured telephone interviews were conducted with seven key individuals who had close professional involvement with bikesharing in Seattle. Interviewees included transportation researchers, members of the Seattle press, and current and former employees of the Seattle Department of Transportation (SDOT), Puget Sound Bike Share, Pronto operator Motivate, and a dockless bikesharing company. Collectively, these individuals had senior roles in architecting, operating, regulating, and observing Pronto and dockless bikesharing. Interviews occurred from December 2017 - January 2018, and ranged from 30 to 60 min in length. The focus of the interview questions was on identifying the most important factors impacting ridership of Pronto and dockless bikesharing in Seattle, as well as the causes that led Pronto to shut down and dockless bikesharing to launch in Seattle. Interview responses were then analyzed through a comparison of responses, identifying similarities and differences between interviewees, with results tested against rival hypotheses, to the extent possible. A survey of 777 Seattle bikesharing users and potential users was conducted in February - March 2018 (Peters and MacKenzie, 2019). Respondents included users of Pronto (222 respondents) and dockless bikesharing (505 respondents), and potential Seattle bikesharing users who had not yet used any Seattle bikesharing offerings (244 respondents). The survey was hosted on Google Forms 212
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Table 1 Key demographic characteristics of all survey respondents, Pronto and dockless bikesharing users in this survey, and a representative sample of Seattle residents age 18 and older who reported cycling in the past 30 days (PRSC, 2017). All survey respondents
Pronto users in this survey
Dockless users in this survey
Seattle cyclists age 18 & older
Sample Size
777
222
505
N/A
Age (mean)
41
39
39
42
Gender (%)
53% male
65% male
59% male
55% male
Household income < $50,000 $50,000–$100,000 > $100,000 Prefer not to answer
15% 26% 45% 14%
11% 27% 50% 12%
13% 27% 50% 10%
14% 39% 46% 2%
Race White Asian African American Hispanic/Latino Other Prefer not to answer
79% 6% 0.5% 2% 5% 8%
80% 5% 0% 2% 5% 8%
80% 7% 0.5% 3% 5% 5%
82% 5% 5% 1% 1% 6%
Where did you hear about this survey? Seattle Bike Blog UW Today/UW News E-mail Facebook Twitter Other
28% 34% 11% 8% 5% 14%
47% 18% 10% 9% 8% 8%
40% 22% 9% 9% 7% 14%
N/A
and was distributed through several online platforms: UW Today/UW News, the Seattle Bike Blog, and social media. As a participation incentive, respondents were entered in a drawing for a premium bag (valued at $160) from Swift Industries, a Seattle company. The survey included sections on Pronto, dockless bikesharing, and demographics. Table 1 summarizes key demographics of the survey respondents and compares them with those of a representative sample of Seattle residents who reported cycling in the past 30 days, from the Puget Sound Regional Council’s 2017 Regional Travel Survey (PRSC, 2017). Pronto ridership data was downloaded from Seattle’s open data portal and included trip based data from Pronto’s entire life. Data for other US bikesharing systems was downloaded from a variety of online sources, often the system operators themselves. Disaggregate data for Seattle’s dockless systems were not available; instead, SDOT shared summary data reports, and other data on the dockless bikesharing were obtained from popular press reports. This paper makes use of the most detailed publicly available data about dockless bikesharing in Seattle. 5. Results This work examines 11 factors that have been floated as possible explanations for Pronto’s low ridership, evaluating each by qualitatively triangulating between three primary sources: system ridership data, user survey data, and interview data. Based on this analysis, the factors were divided into four categories of impact on ridership: (1) major impact, (2) substantial impact, (3) some impact, and (4) little impact. These results are summarized in Fig. 3 and key evidence related to each factor is summarized in Table 2. Details of the analysis are discussed in the remainder of this section.
Fig. 3. Bikesharing system factors impact on ridership (color). 213
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Table 2 Summary of key evidence on determinants of ridership from survey, expert interviews, and system data (Bluebikes, 2019; City of Seattle, 2018; Ford GoBike, 2018; Healthy Ride Pittsburgh, 2018, Motivate, 2018b). Factor
Survey
Expert interviews
System data
System scale
Station Proximity to Home ranked #2 most important factor in determining use of Pronto. Station Proximity to Work ranked #3 Station Proximity to Other Destinations ranked #1 Station Proximity to Home ranked #2 most important factor in determining use of Pronto. Station Proximity to Work ranked #3 Station Proximity to Other Destinations ranked #1 Coverage area ranked #4 factor in determining Pronto use Median respondent said Pronto stations were “poorly aligned” with their travel patterns Ranked #5 most important factor Dockless perceived as easy to use by users (4.3 average out of 5) Pronto perceived as moderately difficult to use by users (2.5 average out of 5) Dockless short term pass perceived as high value by users (4.4 average out of 5) Pronto short term pass perceived as low value by users (2.0 average out of 5) Pricing structure was ranked as the #8 most important factor determining Pronto use. Ranked #7 most important factor determining Pronto use.
High Importance - All interviewees listed as a top factor
Dockless (1000–10,000 bikes) greater than 7× ridership first full year compared to Pronto (500 bikes)
High Importance - Most interviewees listed as a top factor
Dockless: 12–119 “stations” (i.e. bikes) per square mile, greater in high use areas Pronto: 10 stations per square mile
Medium Importance - A few interviewees listed as a top factor, others listed as a significant factor
Many neighborhoods with high dockless ridership omitted from Pronto service area
Helmet Law
Ranked #11 most important factor determining Pronto use.
Medium Importance - a few interviewees listed as a significant factor
Weather
Ranked #10 most important factor determining Pronto use.
Low Importance - no interviewees listed as a top factor
Quality of Bike Facilities
Ranked #6 most important factor determining Pronto use. Ranked #9 most important factor determining Pronto use.
Low Importance - no interviewees listed as a top factor Low Importance - no interviewees listed as a top factor
Ranked #12 most important factor determining Pronto use.
Low Importance - a few interviewees listed as some impact
Station density
Geographic Coverage Area
Ease of Use
Pricing Structure
Transit Competitiveness
Topography
Press Coverage
Medium Importance - A few interviewees listed as a top factor, others listed as a significant factor
Medium Importance - several interviewees listed as significant factor
Pronto: $8 per day, no per-ride option Dockless: $1 per ride
Low Importance - a few interviewees listed as some impact
Pronto: $8 per day, no per-ride option Dockless: $1 per ride Transit: $2.25-$3.25 per trip Helmet law in place for both dockless and Pronto. Dockless: no helmets provided with bikes Pronto: cleaned, shared helmets available for rent Similar weather for both dockless and Pronto. Pronto: 0.6 rides/bike/day in 2016 Boston: 2.2 (2016) Chicago: 1.5 (2016) Similar bike facilities for both dockless and Pronto. Similar topography for both dockless and Pronto Pronto: 0.6 rides/bike/day (2016) Pittsburgh: 0.4 (2016) San Francisco: 1.0 (2016)
Factors in the “major impact” category were ranked as the most influential in our user survey. These generated consensus or nearconsensus among expert interviewees as being top factors, and system data from Seattle and other cities support the idea that they played a major role. Factors in the “substantial impact” category ranked as somewhat less influential in the user survey, and were cited by only some expert interviewees as being important. System data are consistent with these factors playing a role. Factors in the “some impact” category were cited by fewer survey respondents as being “important” or “very important” determinants of their Pronto ridership. Few or no experts identified them as top factors, and comparisons between Pronto, dockless, and other cities contradict the idea that these factors played a major role. Factors in the “no impact” category have been previously cited as potential explanations for Pronto’s low ridership, but we found no evidence that they had a meaningful effect on ridership. 214
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Although the directionality is intuitive for many of these effects (e.g. that a larger system scale would result in more bikesharing trips than a smaller system, or that an easy to use system would garner higher ridership than a hard to use system), the value of this analysis is in understanding the relative importance of these factors and the specific reasons that dockless bikesharing ridership has outperformed that of Pronto in Seattle, and the insights this provides for bikesharing policy and system design more generally. 5.1. System scale and station density Among the factors considered here, overall system scale and station density stand out as the most important in terms of their impact on ridership. These two factors are also closely correlated, since as system scale increases within a given area, station density also increases. The effect of overall system scale and station density on ridership can be seen in data from Seattle and in other US cities. The three largest US bikesharing systems by total bikes in 2016, Citi Bike in New York City (10,000 bikes), Divvy in Chicago (5800 bikes), and Capital Bikeshare in Washington DC (3700 bikes), also featured the highest 2016 ridership, and were all in the top ten in rides per bike per day, with 3.8, 1.5 and 2.5 respectively (Motivate, 2018a,b,c). Additionally, NACTO’s analysis of North American systems showed that ridership at a station increases exponentially with station density (NACTO, 2015). In Seattle, Pronto launched with 500 total bikes and 50 stations, spread over 5 square miles, for a station density of 10 stations per square mile, well below NACTO’s recommended station density of 28 stations per square mile (NACTO, 2015). While many attempts were made to expand the system scale and increase density, it closed two and half years later without a significant expansion. Dockless bikesharing meanwhile launched in July 2017 with 1000 bikes, or 12 bikes per square mile (over the city’s 84 square miles), and had more than 9000 bikes between three companies by the end of the year, more than 100 bikes per square mile (SDOT, 2018a). That said, the functional density is variable throughout the city, as many high ridership areas typically feature far more bikes than low usage areas. Additionally the dockless model may amplify density as bikes are spread out rather than grouped; with trips able to begin or end anywhere in the service area, each bike can act as its own “station,” increasing effective density of a system. Increased density appears to be one of the key reasons that dockless bikesharing has seen much more ridership in Seattle than Pronto, which featured a low density of stations. With that said, the dockless model does introduce additional unpredictability into the system, as riders cannot be assured of always finding a bike at the same location. Stations and bikes which are located a short convenient walk from user origins and destinations are key to maximizing ridership. Since dockless bikes can be parked anywhere, bikes naturally arrive in locations of use, there is greater flexibility for bikes to be moved around and new locations tested, and system planners don’t have to outguess users when siting stations. Dockless bikesharing delivered four times as many rides in its first full month as Pronto did in its best month. As fleet size and density for dockless bikesharing in Seattle grew each month, July to December, total trips also grew each month until ridership dipped in November and December (Fig. 4) (City of Seattle, 2018, SDOT, 2018a). The decline in November and December is most likely due to seasonal variation in cycling demand (Fig. 5) overtaking the growth in adoption of dockless bikesharing. User survey data similarly showed the importance of system scale and density with several users citing “more bikes” or similar responses as key to what dockless bikesharing is doing better than Pronto; other examples included “9000+ bikes is better than 500 bikes”, “massive availability”, “always a bike close by”, and “ubiquity”. Some 30% of respondents provided answers relating to the number, density, or availability of bikes. Proximity of stations to home, work/school, and other destinations, which are directly related to system scale and station density, were also seen as the three most important factors on Pronto use, when users were asked, “To what extent did each of the following factors affect the amount that you used Pronto?” Distributions of responses are shown in Fig. 6, along with mean values of responses based on a 0–3 scale. Interview responses from industry professionals further highlighted the importance of system scale and density, which were the factors listed more than any others among interviewees for their effect on ridership. Industry professionals all agreed that Pronto needed to grow in size to attain higher ridership, “fundamentally the issue with Pronto was scale”, “[Pronto] was never big enough to have the utility needed to be successful”, “the lack of stations is easily the number one downfall of Pronto”, “had Pronto…expanded per the business plan…would it have been successful?”, “[Pronto] needed an opportunity to grow, and it didn’t get that opportunity”. Similarly interview responses indicated that with the low station density on Pronto users had to walk too far to get to a station, and stations didn’t take users close enough to destinations. “Pronto needed more density of stations to make them more convenient.” One interviewee summed it up, “[Pronto] did not have the scale and density to succeed, not much else mattered.” In talking about dockless bikesharing in Seattle, the larger scale was again noted by interviewees as key factor in higher ridership, with the large number of bikes providing easy access to the system. The high density and ubiquitous nature of bikes in dockless bikesharing also adds convenience with bikes often located within a block of origins, and complete flexibility of destinations. 5.2. Geographic coverage, ease of use, and pricing scheme After system scale and density, the most important factors in determining bikesharing ridership in Seattle were geographic coverage area, ease of use, and pricing. The vast increase in ridership among Seattle’s dockless bikesharing systems compared with Pronto can be partially attributed to a larger service area, the perception that the dockless systems are much easier to use, and the perception that the pricing scheme on dockless systems provides much more value. 5.2.1. Geographic coverage Closely related to overall system scale, geographic coverage area also appears to be a major advantage for dockless bikesharing in 215
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 4. Seattle Dockless bikesharing trips and fleet size, by month. Source: Seattle Department of Transportation (SDOT, 2018a).
Seattle compared with Pronto. Pronto had an approximate coverage area of five square miles, while dockless services are serving more or less the entire city (84 square miles). While Pronto’s coverage area featured many of the highest usage areas for dockless bikesharing in Seattle, it missed out on several neighborhoods along the city’s heavily used Burke Gilman Trail cycling corridor (Fig. 7) (City of Seattle, 2018; SDOT, 2018a). The neighborhoods of Fremont, Wallingford, and Ballard have seen more ridership on dockless than portions of Pronto’s service area, particularly Pronto’s eastern fringe including Capitol Hill, Eastlake, and Laurelhurst. User survey data also showed the impact of geographic coverage on bikesharing usage, with Pronto’s service area ranking fourth among factors affecting bikesharing use (Fig. 6). The only factors that ranked higher were proximity of stations to typical destinations, which also relate to coverage area. The survey also revealed that only 44% of respondents reported living within a 20 min walk of a Pronto station, and just 15% within a 5 min walk (the latter is similar to NACTO’s standard for convenient walking distance to a station) (NACTO, 2015). When asked, “How well aligned were Pronto stations with your travel patterns?” on a five-point scale (1 = poorly aligned, 5 = well aligned), the average score given was a 1.9, with a median of 1.0, or poorly aligned (Fig. 8). Three industry professionals also listed coverage area among the top factors impacting Pronto’s ridership, noting that unless you were taking a trip that began and ended in Pronto’s relatively small service area, Pronto could not serve you. Also cited was the lack of flexibility Pronto had with its coverage area as grant funding required a number of stations in South Lake Union and the University District. These University District stations in particular were low performing since they were separated from the primary downtown network. In contrast, interviewees cited the ubiquity of bikes throughout the city as a major advantage for dockless bikesharing. In general, interviewees defended the service area chosen by Pronto, though some said stations in Capitol Hill may have received more ridership if they had instead been located in flatter and lower areas, such as near the Burke Gilman Trail. 5.2.2. Ease of use Traditional docked bikesharing systems and dockless bikesharing operate under very different models, which can make for a significant difference in the ease of use for each type. Results from the user survey showed that users and potential users of bikesharing in Seattle found dockless systems far easier to use than Pronto. When users were asked, “To what degree did you find Pronto Bike Share easy to use?”, on a five-point scale (1 = difficult, 5 = easy), users on average gave Pronto a 2.5, with half of users indicating “1” or “2”. Meanwhile, responses to the same question for dockless bikesharing averaged 4.3, meaning users have found dockless bikesharing far easier to use (Fig. 9). When asked an open-ended question about what dockless services were doing better than Pronto, 34% of respondents mentioned 216
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 5. Monthly bicycle counts on three key bicycle facilities in Seattle: 2nd Avenue downtown cycle track (City of Seattle, 2019a); Spokane Street Bridge linking Seattle with West Seattle (City of Seattle, 2019b); Fremont Bridge linking Seattle with North Seattle (City of Seattle, 2019c).
convenience or ease of use. Further, user survey results showed that ease of use was one of the most important factors determining Pronto use, ranking fifth, behind service area and the three factors pertaining to stations proximity to origins and destinations (Fig. 6). Responses in interviews with industry professionals also highlighted this difference in ease of use and a few interviewees cited ease of use as a top factor impacting Pronto use. “Pronto was hard to use…barriers to entry and use were high.” Interviewees indicated the convenience of having dockless bikesharing bikes everywhere, as opposed to often inconvenient station locations on Pronto, make dockless easier to use. Responses also highlighted advantages dockless bikesharing has in the speed and ease of transactions, which are made on a smartphone, as opposed to transaction speeds for casual Pronto users taking around two minutes. Our survey did not probe for further details on what specific features make dockless bikesharing easier to use than Pronto, but we can speculate. For one, dockless bikesharing is entirely smartphone based, and allows the user to go directly to the bike, rather than interacting with dock hardware. Second, a dockless bike trip can be ended anywhere, which demands less pre-planning than a dockbased trip that can only be ended at a station. In an open-ended question about what dockless services are doing better than Pronto, the most common response (36% of respondents) related to the ability to drop bikes off anywhere. Additionally, short-term Pronto users had to go through a menu of choices at the docking station each time they checked out a bike, such as what type of pass they wanted to purchase, if the user wanted to rent a helmet, and which bike the user wanted to check out. Pronto users then needed to swipe their credit card. Users who chose to rent a helmet had the additional task of entering a code on a keypad at the helmet dispenser, which could then be opened to get a clean helmet. In contrast, repeat dockless bikeshare users simply open their app, scan a QR code on the bike, and begin riding. 5.2.3. Pricing scheme Another major difference between Pronto and dockless bikesharing has been their pricing schemes. Pronto’s pricing scheme, which closely matched most traditional docked bikesharing systems nationally and internationally, was membership based offering day passes for $8, with regular system users getting a much better deal if they purchased an annual pass for $85. Pronto did not offer a per-trip pricing option. Dockless bikesharing on the other hand encouraged more casual use of the system by offering short rides for $1 each, monthly plans at around $30 a month, and annual passes on Spin for $99 a year. User survey responses showed that users found dockless bikesharing’s pricing structure for short-term use provides far more value compared with Pronto’s short term passes, while Pronto’s long term passes provided similar value to dockless monthly and annual plan offerings. When asked, “How would you rate the value of short-term Pronto passes?” on a five-point scale (1 = poor value, 5 = great value), the mean and median response among users was a 2.0. In contrast, users gave the value of $1 dockless rides at a 217
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 6. User Survey Responses: To what extent did each of the following factors affect the amount that you used Pronto? 0 = not at all important, 1 = somewhat important, 2 = important, 3 = Very Important. (color).
mean score of 4.4, with the median score a perfect 5, or great value. When asked about the value of long-term Pronto passes, Pronto had a mean score of 3.1, and dockless long-term offerings had a mean score of 3.2. Although respondents stated that Pronto provided mediocre value, they also indicated that the impact of pricing structure on Pronto use was relatively unimportant compared with top factors, ranking it as eighth most important to their use of Pronto among 12 factors (Fig. 6). Several industry experts considered pricing structure to be a relatively important variable impacting bikesharing use. Interviewees noted the high costs of Pronto’s short-term passes as a significant deterrent for casual users, with the pricing structure mostly appealing to regular users of the system, who were long-term members, and tourists who were willing to pay $8 to rent a bike. This left out many potential users who might have wanted to use it as one mobility option among many. “If you don’t have an annual pass, you’re not going to pay $8 just for a bike ride. That’s a huge price.” Rather than users being able to take spontaneous rides when needed, Pronto was only affordable if users made a full year commitment. In the parlance of diffusion of innovation theory (Rogers, 1962), Pronto had low trialability, which tends to impede adoption. Other responses noted that while pricing structure may have played a role in ridership, it was much less important than overall system scale and density, which significantly limited Pronto’s overall utility with few potential origins and destinations. 5.3. Other factors considered A number of other factors including transit competitiveness, helmet laws, weather, quality of bike facilities, and topography, appear to have some impact on ridership, while press coverage appeared to have marginal or no impact on ridership. 5.3.1. Transit competitiveness Related to pricing structure, the ability for bikesharing to compete with other transit offerings in a city would seem to impact its ridership potential. Pronto’s $8 day pass was much more expensive than King County Metro bus and Link light rail fares in Seattle, which range from $2.25 to $3.25, while dockless bikesharing has offered rides at a more affordable rate of $1. 218
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 7. First Year Trip Starts for Pronto and Dockless Bikesharing by Area (color).
Fig. 8. User Survey Responses: How well aligned were Pronto Stations with your travel patterns?
As shown in the previous section, responses to the user survey showed that users found far more value in dockless bikesharing’s pricing structure for short term use, compared with Pronto. Users perceived transit competitiveness as somewhat important to use of Pronto ranking it seventh among 12 factors (Fig. 6). Interviewees suggested the impact of transit competitiveness to be similar to pricing scheme. “It’s hard to justify an $8 single ride on Pronto, when you could take the bus for $2 or…an Uber or Lyft for something competitively priced.” Unless users made the commitment to an annual pass, Pronto was not going to be cost competitive with other transit options. Another interviewee suggested that Pronto, with its limited number of origin destination pairs (due to a small overall scale and coverage area), had a limited number 219
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 9. User Survey Responses: To what degree did/do you find Pronto/dockless bikesharing easy to use? (color).
of trips that could compete with transit. 5.3.2. Helmet law The City of Seattle is located in King County, which mandates helmet use for cyclists of all ages. Many jurisdictions around the country have helmet laws that apply to minors, but Seattle is one of very few cities with an all ages helmet law. This marked a significant challenge to bikesharing in Seattle. Pronto launched in 2014 with the most comprehensive approach to providing helmets to users, with helmet vending machines providing cleaned helmets at each station. While there have been no other US examples of bikesharing systems with a helmet law, in Australia where there are helmet laws, bikesharing systems have struggled. Notably, systems in Brisbane and Melbourne featured 0.3 and 0.6 trips per bike per day respectively in 2012 (Fishman et al., 2014). The effect of the helmet law on Pronto’s ridership is difficult to determine from looking at ridership data alone, as many other factors, including those discussed here, shaped Pronto’s ridership. While Pronto provided helmet dispensing machines at each station and informed users of King County’s helmet law, dockless bikesharing in Seattle has taken a significantly different approach to the helmet law. Helmets are not provided with the bike and users are simply informed of the law. However one interviewee explained that under both systems, although the helmet law is in place, “for the most part Seattle Police do not prioritize the helmet law”. Seattle police issued only 12 citations for helmet law violations in the first half of 2017 (Gutman, 2017). Survey data provides interesting insight into helmet usage on Pronto and the dockless systems. When Pronto users were asked, “Did you wear a helmet when you rode Pronto Bike Share?” 42% indicated they always wore a helmet, 31% said sometimes, and 27% indicated they never wore a helmet on Pronto (Fig. 10). Among dockless bikesharing users, who are not provided with helmets, the “never” group was 25 percentage points higher, while the “always” group was 26 percentage points lower. As shown by the lower two panels in Fig. 10, this distribution was essentially the same among the subset of 195 dockless riders who had previously been Pronto users. This suggests that the difference in helmet use on each system is not due to a difference in user base, but more likely due to the difference in helmet provision, with users much more likely to wear a helmet when they are provided. When asked “To what extent did [King County’s helmet law] affect the amount you used Pronto?” Results showed that the helmet law was a relatively unimportant factor on the amount of Pronto use for users and potential users, ranking eleventh of 12 factors (Fig. 6). Among the seven interviewees, responses varied on the degree to which the helmet law impacted Pronto’s ridership. Some interviewees indicated that the process of getting a helmet on Pronto acted as one more deterrent for ridership, and while users could choose to ride without a helmet, the helmets “were in your face.” Pronto’s approach to the helmet law by providing a helmet dispensing system, which included capital costs of helmets and bins and operating costs to collect, clean, and distribute the helmets, was also cited by several individuals knowledgeable about Pronto operations as a significant factor in making Pronto financially unsustainable, “the operational cost behind [providing helmets]…was a huge burden for sure.” Conversely, Pronto’s helmet dispensing system was cited by a few interviewees as innovative: “at the time they executed what was seen as potentially the best practice nationally, especially for a community with a helmet law.” Several individuals knowledgeable on the history of Pronto also suggested that politically at the time, Pronto “could not launch without a helmet solution,” and to repeal 220
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 10. User Survey Responses: Helmet Use on Seattle Bikesharing Systems (color). Approximately 25% more Pronto users reported always wearing a helmet, and 25% fewer reported never wearing a helmet, compared with dockless bikesharing users (upper panel). This pattern holds when we consider only the subset of respondents who have used both Pronto and DBS (lower panel), indicating that this difference is not due to selection bias.
King County’s helmet law, “wasn’t looking like it was an option.” It does not appear that King County’s helmet law alone was one of the top factors limiting bikesharing usage in Seattle, but by pushing shared helmets Pronto may have marginally deterred some ridership. And while Pronto’s helmet dispensing system increased helmet usage among users, the hidden costs of added time to check out a bicycle and higher operating costs hurt Pronto overall. “The helmet piece of Pronto…was a part of that story. Cleaning, maintaining, redistributing helmets, ended up being a rather large financial lift.” 5.3.3. Weather, city bike network, and topography Three environmental factors considered, which essentially are the same for both Pronto and Seattle’s dockless bikesharing systems, are weather, the city’s bike infrastructure network, and topography. While these factors appear to have some influence on bikesharing ridership, they play little to no role in the vast difference in ridership between Pronto and Seattle’s dockless systems, as weather conditions, the city bike network, and topography faced by each system are very similar. The effect of weather can be easily seen in bikesharing data in Seattle (Figs. 11 and 12) (City of Seattle, 2018, SDOT, 2018a). Monthly ridership data for Pronto and Seattle’s dockless systems show a strong correlation between ridership, temperature and precipitation. As expected, both systems received much more ridership in Seattle’s warmer, drier months, April through September, compared with colder, rainier months, October through March. The correlation coefficient between temperature and monthly trip count is 0.68 for dockless and 0.91 for Pronto; between precipitation and trip count it is −0.60 for dockless, −0.74 for Pronto. The lack of perceived safe bicycle routes has also been found as a significant barrier to bikesharing use (Fishman et al, 2013). However, it is difficult to draw conclusions about bikesharing ridership relative to quality of bicycle infrastructure, due to lack of granular trip data for the dockless services. The heat map from SDOT of dockless bikesharing trip starts (Fig. 7) suggests the city’s Burke Gilman Trail, may be a generator of dockless bikesharing trips (SDOT, 2018a). Pronto generally missed out on this potential with its limited service area. Previous research has also shown that steeper topography generally discourages bicycling compared with flatter topography (Cervero and Duncan, 2003), and one study ranked Seattle second only to Pittsburgh in ‘hilliness’ among large US cities (Pierce and Kolden, 2015). As was expected, the first year of Pronto data showed that 56% of trips made were downhill, compared with 8% which 221
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 11. Pronto Monthly Ridership, Seattle Monthly Temperatures and Precipitation (color). The decline in ridership from 2015 to 2016 may be due to the drop in annual membership subscribers between fall 2015 and spring 2016 (Fucoloro, 2016c).
were flat and 35% which were uphill. It is clear from the literature that weather (Lewin, 2011, Miranda-Moreno et al., 2011, Rose et al., 2011) affects bikesharing usage month to month, higher quality bike facilities encourage bikesharing ridership (Fishman et al., 2013), and steep topography promotes downhill trips over uphill ones (Cervero and Duncan, 2003). Nevertheless, dockless bikesharing has shown the high ridership potential in Seattle despite these environmental conditions. User survey data showed that users found weather and topography to be relatively unimportant factors on their usage of Pronto. Of the twelve factors that the survey asked about, topography ranked ninth and tenth, respectively, most important to use of Pronto, while city bike network ranked slightly higher, sixth (Fig. 6). Expert interviewees generally downplayed the importance of these environmental factors in shaping bikesharing ridership. Respondents pointed to cities with worse bicycling weather (Chicago, Washington DC), similar hills (San Francisco), and similar bicycle infrastructure (Boston) that have had more successful bikesharing programs than Pronto. (Bachman, 2015, Motivate International Inc, 2018b,c, Bluebikes, 2019)
5.3.4. Press coverage The final factor examined here for its impact on bikesharing ridership in Seattle, is press coverage of bikesharing. While the independently run Seattle Bike Blog has continuously advocated for bikesharing in Seattle even as Pronto struggled, other members of the Seattle press, such as The Seattle Times and its editorial board, have presented a more negative view of bikesharing, especially Pronto. Results of the user survey strongly suggested that press coverage of Pronto had little to no effect for most users on the amount they rode Pronto, ranking last of 12 factors (Fig. 6). While users did not find the press coverage to be influential in their Pronto use, the negative press Pronto received was widely cited among industry professionals interviewed. Several agreed that press coverage has little effect on users’ day to day transportation choices: “I don’t think anyone cares what the local press is saying when they are thinking about making a trip.” At the same time, some interviewees suggested negative press coverage played a role in a decline in Pronto’s annual membership and political support, with one arguing that “there wasn’t a very long period of time when Pronto was touted a success of any kind.” Another summed it up as: “Pronto got caught in a bad media cycle. Pronto was doing poorly, there was a homelessness crisis. It was a tough sell to invest more money in the system, even if it was the right answer.” 222
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fig. 12. Seattle Dockless Bikesharing Ridership, Seattle Monthly Temperatures and Precipitation (color).
Overall, press coverage likely has very little to no direct impact on individual users’ transportation choices, but media coverage certainly did not help the political will to invest and expand the system, which might have helped to grow ridership.
6. Limitations There are several key limitations related to the interviews, user survey, and system data that form the basis of this case study. One limitation is that our team was constrained to relying on publicly released summary reports on dockless bikesharing activity. The City of Seattle requires permitted bikesharing operators to report high-resolution system data to a third party repository, but these data are tightly controlled and only available to staff charged with conducting specific authorized analyses. This limited our ability to analyze dockless usage patterns at a smaller scale than the zones reported in the City’s summary reports, or to examine dockless usage within a specific radius around former Pronto stations. In contrast, dock-based bikesharing operators have traditionally been very open with their data. Access to data is likely to be a challenge in future research involving dockless bikesharing as well, as private companies have little incentive to release data more widely unless compelled to do so by regulators. A second set of limitations relates to the user survey. Although our sample matched Seattle cyclists reasonably well in terms of age, income, and race, it is nevertheless a convenience sample, and it is plausible that survey responses are biased toward individuals with strong feelings for or against bikesharing in general or dockless bikesharing in particular. In addition, it is difficult to completely disentangle some of the 11 ridership determinants from one another in survey responses. For example, proximity to home, work, and other destinations relate to both system scale and system density. Similarly, when respondents tell us that dockless services are more convenient, it is difficult to determine whether they are referring to the app-based process for signing up and checking out bikes, or to the ability to drop bikes off anywhere in the city. A third set of limitations relates to the interviews, and particularly the time when they were conducted. The interviews occurred in December 2017 and January 2018, after the entry of dockless bikesharing and before Spin and ofo exited the Seattle market. At that point, dockless bikesharing seemed to be going very well, so it is possible that interviewees would be inclined to identify dockless system attributes or practices as the best way to go, even if those practices might not be sustainable in the long term. Finally, the success or failure of a bikesharing system depends on more than just raw ridership. As documented in the Background section of this paper, Pronto Cycle Share’s closure after just two and half years of operation in Seattle resulted from a combination of financial struggles (caused in part by low ridership), political, and organizational reasons. Pronto’s initial launch scale was too small to garner significant ridership or revenue, and its inability to expand beyond this initial deployment was the primary reason for its 223
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
demise. The obstacles to expansion were mainly due to political and organizational factors, which proved to be mutually reinforcing with the low ridership numbers. Thus, while the introduction of dockless bikesharing provides a useful opportunity to study determinants of ridership, the pursuit of ridership at all costs does not necessarily lead to a sustainable system. As shown by the experience of Spin and ofo in Seattle, a sustainable bikesharing system likely requires a balancing of system scale and bike density against bike productivity and demand for bikesharing. 7. Conclusions The closure of the Pronto docked bikesharing system, followed soon after by the entry of dockless bikesharing services into the Seattle market, provides a unique opportunity to evaluate the impacts of various factors on bikesharing ridership. Among the 11 factors considered in this paper, overall system scale and density of stations/bikes stand out as having a major impact on bikesharing ridership in Seattle. The next tier of impacts includes geographic coverage area, ease of use, and pricing structure. These three factors all appear to have a substantial impact on bikesharing ridership, but were not as dispositive as system scale and density in driving Pronto’s low ridership. Factors that appear to have had a lesser impact on ridership include competitiveness with transit, helmet laws, weather, bike facilities, and topography. Negative press coverage, while perhaps undermining political support for Pronto, appears to have had little direct impact on ridership. Although dockless bikesharing has garnered much higher ridership in Seattle than the Pronto docked system ever did, this is not necessarily due to it being dockless per se. All five of the top factors that hurt Pronto’s ridership represent system design and business model decisions made by the system operator and/or regulators. They are neither intrinsic characteristics of a dock-based system, nor fixed characteristics of Seattle or its environment. A dock-based system that offered significantly greater system scale and station density, more bikes and broader geographic coverage, and easy, affordable access for new and casual users, would likely have attracted much greater ridership than Pronto did. With that said, dockless bikesharing services do appear to have some inherent advantages over dock-based services. When asked what dockless services are doing better than Pronto, the top answer from our survey respondents (36%) was the ability to drop bikes off anywhere. While more docking stations would erode this advantage, a dock-based system will always have a finite number of potential dropoff locations (though hybrid docked/dockless systems are also possible). Relatedly, bikesharing docks are costly. The original Pronto business plan (Alta Planning + Design, 2012) projected setup costs of $3.1 million for 500 bikes and 50 stations, similar to published equipment and installation costs of $53,000 for a 10-bike/19-dock station from Capital Bikeshare in Washington DC (District Department of Transportation, n.d.). In contrast, Lime’s bikes reportedly cost $300 to manufacture, with a retail price equivalent of $900–1000 (Recode, 2017). Conservatively assuming that the bikes cost $1000 each, the setup cost for a dockless service appears to be about 80% lower than for a docked service with the same number of bikes. Moreover, the dockless bikes can be much more readily redeployed based on learned demand patterns, whereas dock locations will be largely fixed once installed. These results have several implications for the design and regulation of bike sharing systems and other micro-mobility services, assuming the goal is to achieve high ridership. First, governments should welcome providers who can deploy at scale, since system scale, density, and geographic coverage are key to achieving high ridership. Relatedly, they should avoid policies that deter largescale, widespread market entry. If a city is not prepared to commit to the permanent introduction of a large number of shared micromobility vehicles, a limited-time trial is likely a better approach than a limited-scale or limited-area trial, since scale, density, and coverage are so important to ridership. As long as the service entails little in the way of up-front costs to users, riders will not be deterred just because the service may be gone in the future. On the other hand, limiting the scale or area of deployment will render the system intrinsically less useful and will likely limit ridership. The lower capital cost, and the ability to experiment and learn demand patterns, may make dockless services more suitable for such-time limited trials. Second, policymakers and companies should strive to ensure that micro-mobility vehicles can be picked up and dropped off in places that are convenient to users. The misalignment of station locations with users’ travel needs was a major cause of Pronto’s low ridership. Closing off high-demand areas to bikes and scooters will greatly diminish their attractiveness. With that said, our survey showed that the top complaint about dockless bikesharing in Seattle was bikes being parked in inappropriate places. To grow ridership without crowding the right-of-way and engendering a public backlash, one approach might be to transition to a docked or hybrid system as the market matures and demand patterns are learned. If dockless vehicles are to be used, then clearly stated and consistently enforced parking rules may be preferable to blanket closures of area to such vehicles, since closing high-demand areas would diminish the utility of the system and deter ridership. Finally, policymakers may need to recognize and grapple with the complicated relationship between helmet use and ridership. Pronto’s system appears to have been effective at increasing helmet use, and survey respondents indicated that the helmet law had a small effect on their use of Pronto. Yet, the helmet management system was costly to operate and retrieving a helmet takes time, and convenience and cost are important factors in ridership. Moreover, enforcement of the helmet law in Seattle was lax. Strict helmet policy enforcement, whether by companies or police, might deter ridership more than occurred in the Pronto case. This evidence suggests that providing helmets without strictly enforcing helmet laws could be a way to realize increased (though probably not universal) helmet use without deterring ridership. However, it would likely be cost prohibitive (if even feasible at all) to supply clean helmets along with dockless bikes or scooters, due to the lack of a centralized storage, dispensing, and retrieval system. While companies have experimented with shipping free helmets to users, the literature (e.g. Fishman et al., 2013) suggests that the biggest barrier in this area is the need to carry a helmet in case a spontaneous trip is desired. Transportation planners and regulators should consider local goals and priorities when making decisions about micro-mobility system design. On the one hand, dockless systems may be more amenable to time-limited trials and experimentation, and offer riders 224
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
the convenience of ending trips anywhere. On the other hand dockless systems present more challenges for managing parked vehicles, and dock-based systems appear more conducive to helmet use. Ultimately, the most important key to strong ridership is ensuring that a system is deployed with sufficient scale, density, and coverage to align with prospective users’ travel patterns. Acknowledgments We thank the Pacific Northwest Transportation Consortium (PacTrans) for its support of this work through a fellowship for Mr. Peters. Appendix A. Supplementary material Supplementary data to this article can be found online at https://doi.org/10.1016/j.tra.2019.09.012. References Alta Bike Share, 2011. Melbourne Bike Share Survey. Alta Planning + Design, 2012. King County Bike Share Business Plan. The Bike Share Partnership, Jan. 27, 2012. Retrieved May 26, 2018 from http://mobilityworkspace.eu/wp-content/uploads/kcbs_business_plan_final.pdf. APTA, 2017. 2016 Public Transportation Fact Book. American Public Transportation Association. Retrieved May 26, 2018 from http://www.apta.com/resources/ statistics/Documents/FactBook/2016-APTA-Fact-Book.pdf. Bachand-Marleau, Julie, Lee, Brian, El-Geneidy, Ahmed, 2012. Better understanding of factors influencing likelihood of using shared bicycle systems and frequency of use. Transp. Res. Record: J. Transport. Res. Board 2314, 66–71. https://doi.org/10.3141/2314-09. Bachman, Rachel, 2015. Do bike helmet laws do more harm than good? Wall Street J. Oct. 12, 2015. Retrieved May 26, 2018 from https://www.wsj.com/articles/dobike-helmet-laws-do-more-harm-than-good-1444662837. Barber, E., Kopca, C., Starrett, R., 2018. Linking bikeshare trips to light rail use in the Minneapolis–St. Paul Region. TRB Paper No. 18-047112. Transportation Research Board 97th Annual Meeting. Barber, E., Starrett, R., 2018. Unraveling the relationship between bikeshare and rail transit use: a Chicago case study. TRB Paper No. 18-05682. Transportation Research Board 97th Annual Meeting. Bluebikes, 2019. Blue Bikes Trip Data. https://www.bluebikes.com/system-data/ (accessed June 14, 2019). Campbell, Kayleigh B., Candace, Brakewood, 2017. Sharing riders: How bikesharing impacts bus ridership in New York City. Transport. Res. Part A: Policy Practice 100, 264–282. Cervero, R., Duncan, M., 2003. Walking, bicycling, and urban landscapes: evidence from the San Francisco Bay Area. Am. J. Public Health 93 (9), 1478–1483. City of Seattle, 2018. Pronto Cycle Share Trip Data. Retrieved May 26, 2018 from https://data.seattle.gov/Community/Pronto-Cycle-Share-Trip-Data/tw7j-dfaw/. City of Seattle, 2019a. 2nd Ave Cycle Track North of Marion St. Retrieved June 1, 2019 from https://data.seattle.gov/Transportation/2nd-Ave-Cycle-Track-North-ofMarion-St/avwm-i8ym. City of Seattle, 2019b. Spokane St. Bridge Counter. Retrieved June 1, 2019 from https://data.seattle.gov/Transportation/Spokane-St-Bridge-Counter/upms-nr8w. City of Seattle, 2019c. Fremont Bridge Hourly Bicycle Counts by Month October 2012 - Present. Retrieved June 1, 2019 from https://data.seattle.gov/Transportation/ Fremont-Bridge-Hourly-Bicycle-Counts-by-Month-Octo/65db-xm6k. District Department of Transportation, n.d. Capital Bikeshare Station Costs. Retrieved June 14, 2019 from https://comp.ddot.dc.gov/Documents/Capital%20Bikeshare %20Station%20Flyer%20Costs.pdf. Fischer, C.M., Sanchez, C.E., Pittman, M., Milzman, D., Volz, K.A., Huang, H., et al., 2012. Prevalence of bicycle helmet use by users of public bikeshare programs. Ann. Emerg. Med. 60 (2), 228–231. Fishman, E., Washington, S., Haworth, N., 2013. Bike share: a synthesis of the literature. Trans. Rev. 33 (2), 148–165. https://doi.org/10.1080/01441647.2013. 775612. Fishman, E., et al., 2014. Barriers to bikesharing: an analysis from Melbourne and Brisbane. J. Transp. Geogr., Retrieved May 26, 2018 from https://doi.org/10.1016/j. jtrangeo.2014.08.005. Ford GoBike, 2018 (accessed 26 May 2018). Fucoloro, T., 2012. The plan for puget sound bike share. Seattle Bike Blog, Aug. 9, 2012. Retrieved May 26, 2018 from www.seattlebikeblog.com/2012/08/09/theplan-for-puget-sound-bike-share/. Fucoloro, T., 2013. Alta selected to build and operate Seattle bike share system. Seattle Bike Blog, Apr. 23, 2013. Retrieved May 26, 2018 from www.seattlebikeblog. com/2013/04/23/alta-selected-to-build-and-operate-seattle-bike-share-program/. Fucoloro, T., 2014. Pronto Cycle Share Launches: What You Need to Know. Seattle Bike Blog, Oct. 14, 2014. Retrieved May 26, 2018 from www.seattlebikeblog.com/ 2014/10/13/pronto-cycle-share-launches-what-you-need-to-know-updates/. Fucoloro, T., 2015a. City Proposes Massive Pronto Cycle Share Expansion. Seattle Bike Blog, June 8, 2015. Retrieved May 26, 2018 from https://www.seattlebikeblog. com/2015/06/08/city-proposes-massive-pronto-cycle-share-expansion/. Fucoloro, T., 2015b. Happy First Birthday Pronto. Seattle Bike Blog, Oct. 13, 2015. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/2015/10/13/ happy-first-birthday-pronto-a-look-at-use-and-how-the-bike-share-system-can-grow/. Fucoloro, T., 2015c. No federal money for northgate bikewalk bridge pronto expansion. Seattle Bike Blog, October 27, 2015. Retrieved May 26, 2018 from https:// www.seattlebikeblog.com/2015/10/27/no-federal-money-for-northgate-bikewalk-bridge-pronto-expansion/. Fucoloro, T., 2016a. Pronto needs city buyout before end of March. Seattle Bike Blog, Feb. 4, 2016. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/ 2016/02/04/pronto-needs-city-buyout-before-end-of-march-how-did-we-get-here/. Fucoloro, T., 2016b. City council decides the fate of pronto cycle share. Seattle Bike Blog, Mar. 14, 2016. Retrieved May 26, 2018 from https://www.seattlebikeblog. com/2016/03/14/city-council-decides-the-fate-of-pronto-cycle-share. Fucoloro, T., 2016c. Pronto buyout draws ethics investigation against SDOT director. Seattle Bike Blog, Mar. 23, 2016. Retrieved May 26, 2018 from https://www. seattlebikeblog.com/2016/03/23/pronto-buyout-draws-ethics-investigation-against-sdot-director/. Fucoloro, T., 2016d. After strong week of sales pronto plans first ever ride free day. Seattle Bike Blog, May 13, 2016. Retrieved May 26, 2018 from https://www. seattlebikeblog.com/2016/05/13/after-week-of-strong-sales-pronto-plans-first-ever-ride-free-day/. Fucoloro, T., 2016e. Motivate is out city picks young Quebec Company for New E-Assist Bike Share System. Seattle Bike Blog, Oct. 10, 2016. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/2016/10/10/motivate-is-out-city-picks-young-quebec-company-for-new-e-assist-bike-share-system/. Fucoloro, T., 2016f. Council sets pronto shutdown deadline. Seattle Bike Blog, Nov. 23, 2016. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/2016/ 11/23/council-sets-pronto-shutdown-deadline-keeps-bike-share-expansion-funds-for-now/. Fucoloro, T., 2017a. Mayor murray cancels bike share expansion. Seattle Bike Blog, Jan. 13, 2017. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/ 2017/01/13/mayor-murray-cancels-bike-share-expansion-will-shut-down-pronto-march-31/. Fucoloro, T., 2017b. By killing Pronto Seattle could become the center of private bike share innovation. Seattle Bike Blog, Apr. 28, 2017. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/2017/04/28/by-killing-pronto-seattle-could-become-the-center-of-private-bike-share-innovation/.
225
Transportation Research Part A 130 (2019) 208–226
L. Peters and D. MacKenzie
Fucoloro, T., 2017c. City releases draft bike share pilot permit. Seattle Bike Blog, June 9, 2017. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/2017/ 06/09/city-releases-draft-bike-share-pilot-permit-list-of-interested-companies-grows-to-ten/. Fucoloro, T., 2017d. Bike share is now live. Seattle Bike Blog, Jul. 17, 2017. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/2017/07/17/bike-shareis-now-live-a-handy-guide-to-the-new-1-bikes/. Fucoloro, T., 2017e. Spin smashes pronto ridership in week one. Seattle Bike Blog, Jul. 25, 2017. Retrieved May 26, 2018 from https://www.seattlebikeblog.com/ 2017/07/25/spin-smashes-pronto-ridership-in-week-one-announces-improved-bikes/. Fucoloro, T., 2017f. Bike Share Giant Ofo Announces Thursday Launch. Seattle Bike Blog, Aug. 15, 2017. Retrieved May 26, 2018 from https://www.seattlebikeblog. com/2017/08/15/bike-share-giant-ofo-announces-thursday-launch/. Fucoloro, T., 2017g. Seattle Bike Share Guide. Seattle Bike Blog, Aug. 20, 2017. Retrieved May 26, 2018 from www.seattlebikeblog.com/seattle-bike-share-guide/. Fucoloro, T., 2017h. Bike Share Pilot’s Daily Ridership Blows Past Pronto’s Lifetime Totals. Seattle Bike Blog, Dec. 15, 2017. Retrieved May 26, 2018 from www. seattlebikeblog.com/2017/12/15/bike-share-pilots-daily-ridership-blows-past-prontos-lifetime-totals-rivals-both-streetcars-combined/. Gutman, D., 2017. Helmets may be Seattle law, but many bike-share riders don’t wear them. The Seattle Times, August 19, 2017. Retrieved June 4, 2019 from https:// www.seattletimes.com/seattle-news/transportation/helmets-may-be-seattle-law-but-many-bike-share-riders-dont-wear-them/. Healthy Ride Pittsburgh, 2018. Healthy Rides Data. Retrieved May 26, 2018 from https://healthyridepgh.com/data/. ITDP, 2013. Bike share planning guide. Institute for Transportation and Development Policy. New York, NY. Retrieved May 26, 2018 from https://www.itdp.org/wpcontent/uploads/2014/07/ITDP_Bike_Share_Planning_Guide.pdf. LDA Consulting, 2012. Capital Bikeshare 2011 Member Survey Report. Jun. 14, 2012. Retrieved May 26, 2018 from https://d21xlh2maitm24.cloudfront.net/wdc/ Capital-Bikeshare-SurveyReport-Final.pdf?mtime=20161206135935. Lewin, A., 2011. Temporal and weather impacts on bicycle volumes. In: Proceedings of 90th 23 Transportation Research Board Annual Meeting, Washington, D.C. Midwest Bikeshare, Inc., 2018. Bublr Bikes Data. Retrieved May 26, 2018 from https://bublrbikes.org/media. Miranda-Moreno, L.F., Nosal, T., 2011. Weather or not to Cycle; Whether or not Cyclist Ridership has 14 Grown: a Look at Weather's Impact on Cycling Facilities and Temporal Trends in an Urban 15 Environment. In: Proceedings of 90th Transportation Research Board Annual Meeting, Washington, 16 D.C. Motivate International, Inc., 2018a. Citi Bike Monthly Operating Reports. Citi Bike NYC. Retrieved May 26, 2018 from https://www.citibikenyc.com/system-data/ operating-reports. Motivate International, Inc., 2018b. Divvy System Data. Divvy Bikes. Retrieved May 26, 2018 from https://www.divvybikes.com/system-data. Motivate International, Inc., 2018c. Capital Bikeshare System Data. Capital Bike Share. Retrieved May 26, 2018 from https://www.capitalbikeshare.com/system-data. NACTO, 2015. Walkable Station Spacing is Key to Successful, Equitable Bike Share. National Association of City Transportation Officials. April 2015. Retrieved May 26, 2018 from https://nacto.org/wp-content/uploads/2015/09/NACTO_Walkable-Station-Spacing-Is-Key-For-Bike-Share_Sc.pdf. NACTO, 2019. Shared Micromobility in the U.S.: 2018. National Association of City Transportation Officials. April, 2019. Retrieved May 21, 2019 from https://nacto. org/shared-micromobility-2018/. Nice Ride Minnesota, 2010. Nice Ride Minnesota Survey November 2010. Retrieved June 13 2019 from https://projectadvisoryteam.files.wordpress.com/2011/01/ nice-ride-subscriber-survey-summary-report-nov_1_2010-nice-ride-fall-subscriber-survey.pdf. Pierce, J., Kolden, C., 2015. The Hilliness of US Cities. Geographical Review, Oct. 2015. PSRC, 2017. Household travel survey program: Spring 2017 travel survey. Puget Sound Regional Council. Spring, 2017. Retrieved June 1, 2019 from https://www. psrc.org/household-travel-survey-program. Recode, 2017. Full transcript: LimeBike President and co-founder Brad Bao answers bike-sharing questions on Too Embarrassed to Ask. Retrieved on June 14, 2019 from https://www.vox.com/2017/8/23/16189474/transcript-limebike-president-brad-bao-bike-sharing-questions-gps-too-embarrassed-to-ask. Peters, L., MacKenzie, D., 2019. Seattle bikeshare survey. Mendeley Data V1. https://doi.org/10.17632/9zfrh6r4p9.1. Rogers, E.M., 1962. Diffusion of innovations. Free Press of Glencoe, New York. Rose, G., Ahmed, F., Figliozzi, M., Jakob, C., 2011. Quantifying and comparing the effects of weather 18 on bicycle demand in Melbourne (Australia) and Portland (USA). In: Proceedings of 90th 19 Transportation Research Board Annual Meeting, Washington, D.C. SDOT, 2018a. Bike share by the numbers. May 2, 2018. Retrieved May 26, 2018 from sdotblog.seattle.gov/2018/05/02/bike-share-by-the-numbers/. SDOT, 2018b. 2017 Free-Floating Bike Share Pilot Evaluation Report. August, 2018. Retrieved June 13, 2019 from https://www.seattle.gov/Documents/Departments/ SDOT/BikeProgram/2017_BikeShare_Evaluation_Report_113018.pdf. SDOT, 2019a. Monthly Status Report - December 2018, Seattle Free-Floating Bike Share Program. February, 5 2019. Retrieved June 13, 2019 from https://www. seattle.gov/Documents/Departments/SDOT/BikeProgram/BikeShare_Summary_Report_Dec_Final%20(2-5-2019).pdf. SDOT, 2019b. Monthly Status Report - January 2019, Seattle Free-Floating Bike Share Program. March 2019. Retrieved June 13, 2019 from https://www.seattle.gov/ Documents/Departments/SDOT/BikeProgram/BikeShare_Summary_Report_Jan19_Revised_03262019.pdf. SDOT, 2019c. Monthly Status Report - February 2019, Seattle Free-Floating Bike Share Program. Retrieved June 13, 2019 from https://www.seattle.gov/Documents/ Departments/SDOT/BikeProgram/BikeShare_Summary_Report_Feb19.pdf. SDOT, 2019d. Monthly Status Report - March 2019, Seattle Free-Floating Bike Share Program. Retrieved June 13, 2019 from https://www.seattle.gov/Documents/ Departments/SDOT/BikeProgram/BikeShare_Summary_Report_Mar19.pdf. SDOT, 2019e. Monthly Status Report - April 2019, Seattle Free-Floating Bike Share Program. Retrieved June 13, 2019 from https://www.seattle.gov/Documents/ Departments/SDOT/BikeProgram/BikeShare_Summary_Report_Apr19.pdf. Shaheen, S.A., Martin, E.W., Cohen, A.P., Chan, N.D., Pogodzinski, M., 2014. Public Bikesharing in North America During a Period of Rapid Expansion: Understanding Business Models, Industry Trends & User Impacts, MTI Report 12-29.
226