Lisbon put on-demand transport in the middle of a capital city
Shotl's first deployment in a dense urban core, run with Carris in the Avenidas area — where the challenge is not empty streets but full ones.
- Setting
- Dense city centre, ~3 km²
- Service model
- On-demand transport
- Operator
- Carris
- Hours
- Mon–Fri, 8:00–20:00

Results
Reached in the first three months of operation in the central district.
From request to pickup, in traffic, across a 3 km² zone.
Pooled journeys within the centre, door to door.
The challenge
On-demand where the problem is density, not distance
Carris did not start from a gap in the network. They started from a question — where should on-demand transport go in a city like Lisbon, and who would actually use it? Rather than guess, they chose the most central, most mixed district they could: Avenidas Novas with Campolide and São Domingos de Benfica, where elderly residents, schools, universities and IT offices all sit within the same few square kilometres.
The relationship began with an earlier pilot for people with reduced mobility. This second deployment was chosen partly to prove geolocation precision: many of these avenues carry two central lanes with side lanes either side, so a pickup only works if the vehicle stops on the correct carriageway.
What we deployed
Carris’s own service, running on Shotl underneath
- 0161 virtual stops
Placed across a 3 km² zone, lane by lane — no new infrastructure built.
- 02Xbus white label
Booking app and live vehicle map published under Carris’s own brand.
- 03Weekday service
Monday to Friday, 8:00 to 20:00, alongside the existing fixed network.
- 04Multi-GNSS positioning
GPS, Galileo, GLONASS and BeiDou combined for lane-level accuracy.
Stopping on the right side of the avenue
Positioning runs on multi-GNSS — GPS, Galileo, GLONASS and BeiDou together — so a virtual stop on a central lane is not confused with the side lane a few metres away. In a grid of divided avenues that precision is what makes the pickup work.
How virtual stops are placedOn site
João VieiraDirector for Strategy and Innovation, Carris“Our focus is not so much on understanding which services we want to explore, but more on the people. What kind of people use it? What do they need? And whether we can use this technology to actually meet their needs and make public transport more competitive.”
Similar challenges
Proof that the model is not only for thin demand
Lisbon was Shotl’s first launch in the centre of a large capital, and it was run as a learning exercise. Carris wanted to know which riders on-demand actually serves before deciding where to take it next — low-demand neighbourhood lines connecting points within one district, or night services, are both candidates.
The answer holds on two conditions: the service must complement the fixed network rather than shadow it, and the stops must be accurate to the lane. Any dense district with a diverse user base and streets a large vehicle handles badly meets the same test.
Multimodal bookings4.5/5Kawasaki
JapanA feeder service with Odakyū linking residential streets to the rail network.
Demand responsiveRider-set timesSant Cugat
SpainRidership rose on suburban routes once riders could choose their own departure.
Demand responsive−90% waitTorre Baró
SpainA steep, disconnected neighbourhood reached a 4.7/5 rider rating.


