Shared bikes, e-bikes, and scooters can pull short car trips off crowded streets. They don't do it automatically.
Can a shared fleet actually shorten commute times in your city? Evidence says yes in one well-studied U.S. case. It says almost nothing solid for the country as a whole.
Atlanta is the case people cite. In 2019 the city used geofencing to shut down e-scooters and e-bikes during late hours, and Georgia Tech researchers found average commute times rose by about 10%. The result is worth your time. It isn't a national congestion formula.
If you operate a rental or shared fleet, or you help a city decide whether to permit one, treat congestion relief as a local test. Don't sell it as a guarantee.
What counts as micromobility
Micromobility covers small, low-speed devices people ride for short urban trips, including pedal bikes, e-bikes, scooters, and e-scooters that cities allow on streets and paths. Powered models in the SAE class used in federal briefings sit under 30 mph and 500 pounds.
Shared trip volume in the U.S.
Scale is the first filter. A handful of scooters can't change a freeway.
U.S. shared bike and scooter systems logged 157 million trips in 2023, an all-time high after a 2019 peak of 147 million and the 2020 collapse, according to NACTO's 2023 Shared Micromobility report as summarized in the ITS micromobility briefing. SafeTREC's recap of that same report puts 13-year ridership in the United States and Canada at 887 million trips.
Those numbers show demand. They don't prove car trips vanished.
SafeTREC notes that 40 percent of dockless e-scooter trips happened in ten major U.S. cities. Dense cores absorb most of the rides. A small town shouldn't expect an Atlanta-sized traffic shift from a thin fleet.
Thing is, congestion only eases if the new trips replace driving, not walking, transit, or a trip that would never have happened. That substitution is the whole game.
Atlanta's 2019 shutdown
Atlanta did not run a polite opt-in pause. Devices went dark through geofencing, so compliance was effectively complete, which is rare for this kind of rule.
Georgia Tech treated that blackout as a natural experiment on whether micromobility displaces cars. Average commute times increased by about 10% after e-scooters and e-bikes were banned. For evening travelers, the extra delay added up to about 784,000 hours a year in traffic, and that count starts at 9 p.m., not the morning rush. The morning peak was not the test.
Omar Asensio, the study's principal investigator, confirmed that a moratorium during peak rush hour would cause even more congestion. NSF funded the work through a CAREER award. The paper is titled "Impacts of Micromobility on Car Displacement With Evidence From a Natural Experiment and Geofencing Policy."
Read that sequence slowly. The city was trying to cut nighttime crashes, riders lost the devices, and drive times still stretched, which is the whole point of a natural experiment you almost never get when a city just asks operators to please cooperate and hopes people listen.
This is the cleanest U.S. before-and-after operators can point to. Use it as evidence that availability can affect car travel.
What the evidence does not show
NACTO and federal micromobility briefings still don't publish a U.S.-wide figure for vehicle miles avoided. City results do not transfer cleanly.
A dockless scooter downtown is not the same service as a docked e-bike near a rail station. Hills, weather, parking rules, and who is allowed to ride all change whether someone leaves the car at home. Those details decide the outcome.
To be honest, a lot of fleet pitch decks skip that substitution question and jump straight to trip counts.
How to test congestion impact before you scale
You need a pilot you can defend. Guesswork wastes fleet and political capital.
- Confirm local rules first. Hours, sidewalks, speed caps, parking, and geofence zones decide what you can operate. Atlanta showed a remote shutdown can erase availability in a night.
- Record a baseline. Grab commute times, corridor speeds, and any VMT or traffic counts from the city or state DOT. Name the streets and hours you will watch.
- Place a small fleet on short, painful car trips. Stay inside the SAE micromobility class. Skip the edge of town if nobody will ride those miles.
- Instrument every vehicle. Log trips, idle time, rebalancing, and utilization so you can see movement versus sidewalk clutter. Fleet software should tie a booking to a specific bike or scooter and keep one live status: available, reserved, out, overdue, or in maintenance.
- Measure mode shift, not vanity ridership. Ask riders what they would have done instead: drive, rideshare, transit, walk, use a bike they own, or skip the trip. Watch the same corridors for 6 to 12 months.
- Set fail conditions before launch. If utilization stays low or surveys show you replaced walks, shrink or relocate. Scale only if car replacement shows up in both surveys and travel times.
A simple scorecard for the pilot
Track these items for the full pilot window, not just launch week.
| Metric | Why you collect it |
|---|---|
| Trips by zone and vehicle type | Volume has to matter on a real corridor |
| Utilization per device per day | Idle fleets don't take cars off the street |
| Stated replacement mode | Shows driving vs walking vs transit |
| Corridor travel time at the same hours | Direct congestion check |
| Rebalancing miles and parking complaints | Costs cities notice first |
| Downtime from battery, repairs, or geofence lockouts | Availability is what Atlanta removed |
First-mile and last-mile connections to transit belong on that list if that is part of the city's pitch. Federal micromobility briefings treat that use case as a core job, not a bonus slide.
Rules, parking, and overnight shutdowns
Cities permit these systems, restrict them, and sometimes kill them. Hours, helmets, sidewalk riding, and equity zones differ by jurisdiction. There is no single U.S. rulebook.
Turns out the enforcement tool matters as much as fleet size. Geofencing gave Atlanta near-total compliance. Older moratoriums that asked people to cooperate did not.
Parking remains the quiet stall. Riders won't adopt a system if they can't leave the device without a ticket. Stations or marked corrals fix some of that. They also cost space.
Start with two corridors
Pull the latest shared-trip counts for your market from NACTO or the local program operator. Pick two or three congested corridors you already watch. Compare those counts with travel times, then run a small instrumented pilot under the rules you already have before you buy a bigger fleet. Keep the first fleet small on purpose.