Disclaimer | This article may contain affiliate links, this means that at no cost to you, we may receive a small commission for qualifying purchases.
Bringing Bridj to Kansas City seemed like a no-brainer to transit officials. For just $1.50, anyone could use an app to summon a ride downtown in van that would follow a route calculated on the fly by an algorithm. No one within the service area was ever more than a 10 minute walk from a stop, and as an added incentive, your first 10 rides were free.
It flopped. Just 1,480 people rode on a Bridj van, a laughably small figure in a city of 2 million people. The city launched the program with the Boston mobility startup in March 2016, and in the past six months just one-third of riders took more than 10 rides. The one-year, $1.3 million project ended Friday. You might call it a failure.
Government officials and transit researchers call it a success.
“I’ll be honest: The ridership was not the top priority,” says Jameson Auten, who leads the innovation division of the Kansas City Area Transportation Authority. “The top priority for us was learning who uses on demand. Really, the big goal for us was learning itself.”
Transit agencies nationwide hope to learn from it, too. Many of them think on-demand, app-driven transportation services could make public transit cheaper, more accessible, and more convenient. The typical rider can reach just 30 percent of local jobs on mass transit, and rides can last 90 minutes. So Uber drivers are getting folks in Summit, New Jersey, to the train station. Lyft provides residents of Centennial, Colorado, with lifts to light rail. Both companies work with the city of Boston to serve those with disabilities.
The Bridj project was a bit different, in that it used unionized transit employees driving American Disabilities Act-compliant vehicles. “I think this was a bridge to inspiring a lot of transit agencies to start looking at public-private partnership,” says Susan Shaheen, a UC-Berkeley civil engineer who studies mobility innovation. The results, however, prove the model needs revising, and a lot more data. […]