Inviting travelers to the smorgasbord of sustainable urban transport: evidence from a MaaS field trial

The intended audience for the service was urban households judged to have sufficient access to the existing transport solutions (carsharing and PT in particular), and large enough travel needs for UbiGo to be economically competitive with their current solution. The ‘household’ could be comprised of any number of individuals including both adults and children, i.e. typically a family.

Participants were recruited using media, internet, social media, and targeted marketing in areas where the transport infrastructure was considered sufficient. The recruitment material targeted groups deemed to benefit from the service, that is households with a seldom-used car, households considering buying a car, or households already using many of the included services and interested in an easier way to handle them.

The resulting UbiGo participant group consisted of 83 customer subscriptions covering 195 persons: 173 adults and 22 children (under 18). Table 1 provides a socio-demographic breakdown of the sample, and comparison with the demographics of Gothenburg, including residence, occupation and travel behavior.


Data was collected via a mixed-methods approach, comprising questionnaires, personal or household interviews, focus groups, travel diaries, as well as use data, error reports and customer service issues. The basis for this paper is the questionnaire and interview data.

The participants were asked to fill out three online questionnaires: ex-ante, in-itinere, and ex-post (151 adults completed all three questionnaires). The ex-ante questionnaire concerned expectations, motivations to join and included a self-reported baseline for travel behavior. The in-itinere questionnaire addressed the participants’ experience of all aspects of the service and included a measurement of current behavior; and the ex-post questionnaire concerned changes in behavior and willingness to continue with the service.

A selected number of participants were invited to ex-post interviews; 14 individual and three household interviews were performed. The interviews allowed the participants to elaborate around topics from the questionnaires; they were probed to provide more in-depth information on the reasoning behind their opinions and to more thoroughly explain their experiences using the service. These audio-recorded interviews lasted 60–90 min.

Each interview was transcribed and summarized using a version of contact summary forms (Miles and Huberman 1994), in which statements regarding expectations, reasons, behavioral change, and reflections regarding travel and behavior were gathered. These summaries were comparatively analyzed and, from the pattern analysis, four subgroups of participants could be identified, as there were noteworthy differences between participants depending on their perception of the value the service would provide them combined with their expectations for behavioral change based on this service offer. The interview material was then revisited to explore behavior, reasoning, and reflections within these four categories to confirm the subgroups and identify underlying explanations.

The questionnaire data was first summarized for the entire group and subsequently analyzed based on the four identified subgroups. Statistical analyses of the questionnaire data were performed with the software SPSS.

From the ex-ante questionnaire, on the whole, participants joined mostly due to curiosity about the new concept and what it could offer them with 63% citing this as their main motivation, with convenience/flexibility, economy, and environment far behind (see Sochor et al. 2014). Some, but far from all, expected their mode choice and travel patterns to change during the trial (see Fig. 1a).

As mentioned, four different subgroups were identified based on the interview data. They were named Car shedders, Car accessors, Simplifiers, and Economizers based on the value that they perceived UbiGo would afford them. Table 2 describes these groups and their expectations of the service in-depth. The groups also differed sociodemographically, for example Economizers lived in a house (vs. an appartment) to a statistically significant greater extent than did all other subgroups, and they also had more adults per household and higher household income than did Car accessors. Degrees of car ownership and carsharing membership also differed. The subgroups also had different expectations of other types of change that would occur on the basis of the altered preconditions offered by the service (Fig. 1a). For example, the Car shedders expected to reduce their environmental impact the most as they also expected their mode use to change the most (away from car). Accessors on the other hand looked forward to more choice of modes and thus being able to adapt the mode to the trip, as well as a simplified way of handling travel expenses (similar to the Simplifiers).

At the end of the trial, the most frequently reported behavioral change was mode (42%), followed by the amount of pre-trip planning (34%), trip chaining (21%), destinations (21%), and travel time (21%). There were also those, 36% of participants, who reported no change (mainly Simplifiers and Car accessors).

Looking closer at reported mode use changes, the overall trends are that all subgroups used car less and walked more. The two car-owning subgroups reduced their private car use; 95% of Car shedders and 53% of Economizers reported using their car considerably more seldom than before the trial (see Fig. 1b). The Car shedders appear to have spread their former car trips across all other available modes, while the Economizers primarily used more PT (60%), which may be due to their joining to get access to cheaper PT. They also used more carsharing and walked more, but less private bicycling. A large proportion of the Car accessors increased their use of carsharing (78%) as well as rental cars (30%) but reduced private car use (37%), a result of not borrowing cars from friends and family anymore. At the same time many of them increased their use of PT (41%), walking (18%), bicycle-sharing (28%), and private bicycle (17%), even if some participants reported less use of these active modes. The fourth subgroup, Simplifiers, is the most split subgroup in the sense that they changed behavior but not in a uniform way; one share increased their use of most modes while another share decreased their use.

Even if many of the changes described are in line with the participants’ expected changes, the interviews showed that they did not fully match. The in-itinere questionnaire confirmed this; most participants did not experience a modal shift or travel pattern change to the degree expected. For some this meant changing more than expected, and for others it meant changing less (see summary Table 2). Some reinforced their existing behaviors but in ways they did not envisage, depending on their ex-ante goals. The two main identified mismatches are lower than expected car use, and higher than expected PT and active mode use.

The UbiGo customers/participants were not chosen at random; no randomized allocation to different groups was undertaken, and no control group was created. The choice was instead strategic; the target group was travelers who (apart from stating an interest in participating) were considered to potentially benefit from the service, i.e. for whom the service could provide an added value or relative benefit. This is essential from both the innovation dissemination and commercial perspectives, as users’ adoption of an innovation relies on it offering a perceived relative benefit compared to the existing solution (product and/or behavior) (Rogers 1995); and the relative benefit will depend on the value offering, which is related to the MaaS level (Sochor et al. 2017). In the case of UbiGo, MaaS level 3, this meant targeting urban households with reasonable distances to PT and carsharing, and a service design that offered perceived added value (see below). The approach can thus be compared with a market segmentation into different customer groups with the purpose of creating, in this case, a service that could meet their specific needs rather than a solution that would ‘fit all’. According to Anable and colleagues (2006), there is a general consensus in literature that a “…staged and targeted strategy of travel behavior change” is likely to be more effective, but the most common segmentations appear to be based on behavior, i.e. car-users and PT users, or on attitudes (cf. Anable 2005), in order to deliver targeted information rather than on actual, contextualized transport need in order to design a product adapted to this need.

A comparison between the demographics of Gothenburg municipality and the subscribers of UbiGo show some important differences that must be considered in assessing the potential of MaaS (Karlsson et al. 2017). For instance, UbiGo was more attractive for those living in the city centre than those living in suburban areas (49% compared to 23% when considering all Gothenburg citizens), the share of retired subscribers was very low compared to Gothenburg overall (1%) and the share of students slightly lower than the municipality (11% compared to 15%). In addition, the distribution of participants across different location areas (i.e. city centre, southwest, southeast, northeast, and north) differed from the municipality but mirror fairly well access to carsharing facilities.

The FOT participants’ evaluation shows that the UbiGo service was considered to solve their transport need and that participants were more satisfied with the transport system using the service. This illustrates the potential of MaaS, but there are other important design elements, deeper than the general MaaS concept of ‘bundling’ or having everything in ‘one spot’ (e.g. a marketplace à la MaaS levels 1 and 2) that are important to customers and that influence behavior (Sochor et al. 2016). For example, the way in which one pays (e.g. increasing or decreasing lock-in effects), the perceived freedom to choose, and the way the service presents potential choices are very important if users are to feel both satisfied with the service and make sustainable choices. Other crucial elements relate to how well the service both practically and psychologically supports the (potential) change process (Strömberg et al. 2016), as people are not well aware of their travel needs and behaviors going into a new type of service like MaaS and therefore have difficulties anticipating potential changes (further discussed below). Open evaluations of additional MaaS implementations need to be performed to further understand the relationships between design elements (and business models) of successful value offerings, customer satisfaction, and sustainability impacts as related to different types of MaaS (e.g. levels 0–4 Sochor et al. 2017).

UbiGo intended to support more sustainable urban mobility without privately owned cars. However, a distinctive characteristic of the service was meeting the household’s full mobility needs—even if this included car use. Although estimates suggest that there are viable options for 40% of car journeys, and with small investments an additional 40% (Brög and John 2001, in Chapman 2007) some trips may be impossible to accomplish, or only with considerable effort, without a car. Certain trips were also difficult within the UbiGo model, which could be an issue potentially affecting acceptance of such a service in the long run (Sochor et al. 2015). Earlier studies have shown that the investment in a private car can be determined by the perceived need shaped by very infrequent, out of the ordinary trips. The participants in the trial had therefore the opportunity to choose a car—if and when it was considered most useful. This flexibility enabled participants to choose the mode most suitable to the trip at hand, instead of being locked into a specific mode. This may appear strange as the service did not actually offer any new modes but several participants mentioned how certain lock-in effects had stopped them from choosing among available modes. Owning a car costs a considerable amount of money even if one does not use it and therefore one might as well use it; a PT pass creates similar effects. UbiGo offered instead an integrated multimodal service where one paid for what one used, which further enhanced the flexibility of the service. Nevertheless, some lock-in effects could be observed during the FOT, as participants tended to use PT more if they had already paid for the day and less if they had not. However, in UbiGo, the lock-in effects were short-term (i.e. a trip or a day) rather than long-term (month, year, or longer); a step forward.

Another key feature was the design of the app, which appears to have played an important, intervening role in the participants’ travel behavior, not only because it was the ‘interface’ by which trips were accessed, but as it presented different modes as equally feasible options every time a trip was to be undertaken. This list of options made participants reflect on mode choice, hereby acting as a kind of habit intervenor, and over time made them more conscious of the ‘smorgasbord’ of transport choices. The possibility to choose different modes, in combination with the app, can thus be argued to have contributed to a change in participants’ perception of what was feasible, i.e. a change in perceived action space (cf. Strömberg 2015). Having a continuous, built-in reminder of the available options may have different, habit-breaking effects than one-time eye-openers that characterize information campaigns.

Despite the encouraging results from UbiGo, it is difficult to draw any general conclusions regarding how many people will change behavior due to a service of this type and how big the changes will be. The findings show that behavioral changes vary with the household’s initial preconditions and motivations/reasons. The findings also show that the participants were unaware of their preconditions, travel needs, and behavior when joining the service, and thereby could not predict how their behavior would change. In adopting the service, participants spent much time figuring out those preconditions and needs. The flexibility of the service played an important role in supporting this change process, as it showed the options and allowed participants to alter their subscription when they had figured out that their initial assumptions about their own travel behavior were incorrect. The behavior change process observed in the study needs to be further explored to be able to model demand and design effective MaaS but indicates both that such services need to contain flexibility to allow people to change within the service, and that the service should support people in their exploration of their own preconditions, and thus which new behaviors they can adopt (Strömberg et al. 2016).

Although the design of the service is argued as fundamental to support the reported behavioral changes, the social and technological context in which the service was introduced must also be acknowledged. The UbiGo service was, as already mentioned, based on existing transport services and infrastructures, and thus potential customers were those deemed to have sufficient access to the existing services and infrastructures. Although the UbiGo service ‘only’ repackaged already available services, the process utilized: (1) the dissemination of new mobile technology, i.e. smartphones, even though access differs across age groups; and (2) shifting consumer attitudes and behaviors towards sharing and servicing with less emphasis on personal ownership (Hamari et al. 2016).

Just as there is currently no definitive version of ‘MaaS’, there also does not exist any ‘one size fits all’ MaaS for all contexts. Noting that MaaS is largely an organizational approach, any MaaS service needs to be tailored to the local geography, culture, and stakeholder ecosystem, as well as personalized to the individual/household. The tailoring of the service should also be coupled with a cost-effectiveness analysis for the case at hand.

The case of UbiGo illustrates the potential for an, overall, ‘virtuous cycle’ between the service and its impact (e.g. behavioral change, sustainability). The need-based approach for the entire household and design elements such as the ‘transportation smorgasbord’, simplicity, improved access and flexibility, economy, and trialability (Sochor et al. 2016) facilitated the use of the service in a personalized way. It met participants’ mobility needs (almost; see below) by offering a full range of choices in a flexible manner. The pricing structure encouraged mode choice for each trip and reduced lock-in effects, and the app design made the choices apparent, which enabled the customer to freely choose and experiment within their adaptable, personalized subscription. This led to both anticipated and unanticipated behavioral changes, which meant that participants both gained insight into that they did not previously fully understand their needs and preconditions, and re-evaluated their choices, options, and perceptions of convenience. And in the end, users generally shifted to more sustainable transport choices.

Overall, the results from the trial demonstrate the potential and relevance of new and innovative measures such as ‘Mobility as a Service’ solutions. Although it does not yet exist, the larger vision is to achieve MaaS level 4 in which local policies are integrated into the service, whereby MaaS becomes more strongly connected to city governance and planning (Sochor et al. 2017). In the end, fundamentally new, different and targeted transport service offers are needed, based on a more holistic approach and the notion of a co-operative and interconnected transport system (including services, infrastructure, information, payment, etc.), where boundaries between not only transport modes are blurred but also between public and private actors.