Page 4 of 22
launching the Pilot program, SFMTA updated its specifications for the Taxi API data by requiring
the submission of the types of data necessary for tracking the Pilot’s performance, such as
upfront or metered fare payment method and the trip application origin (Taxi E-Hail or Third-
Party). For Third-Party Trips, the relevant data is transmitted to SFMTA on a bi-weekly basis by
Flywheel, currently the only taxi company permitted to service the trips — the SFMTA does not
receive any data directly from third-party entities.
As the SFMTA receives data from the industry, the Data Analytics team applies automated and
manual validation procedures to check the quality of the data in each record received. The
validation process has identified numerous data quality issues since the start of the Pilot, such
as regular taxi trip records mislabeled as Taxi Pilot Trips and inaccurate meter fare estimations
for Pilot trips. When an issue is identified through the validation process, the Data Analytics
Team conducts an independent investigation and then collaborates with industry partners to
identify a solution. SFMTA hosts regular technical troubleshooting sessions with each of the
participating taxi companies individually to address data quality issues in a timely fashion. The
data presented in this report has been vetted by the Data Analytics Team and is therefore
reliable and accurate to the best of our knowledge. However, since this is a new program that
involves significant technical updates and challenges, the validation process is ongoing, and all
industry data remains subject to future corrections and updates.
Based on the validated industry data, SFMTA tracks key metrics to measure the actual impact
of the Pilot on the industry based on the goals of the program. To measure that impact, the
metrics compare the data received since the start of the Pilot to historical data from before the
Pilot began. The historical data has undergone a similar validation process, and also remains
subject to future corrections and updates. The Data Analytics Team primarily uses year-over-
year comparisons to account for strong seasonality in the taxi industry but relies on short-term
baseline comparisons to a period immediately preceding the pilot if deficiencies in the long-term
historical data make it necessary. For example, the taxi industry did not consistently report how
a passenger requested a trip (i.e. hail type) prior to October 2022. Therefore, when analyzing
hail type during the Pilot, due to historical deficiencies, October 2022 is used as the baseline
period for comparison purposes.
Staff has also analyzed the Q2 data as compared to Q1 data, understanding that the seasonal
differences between those two quarters impact results, but that the analyses are still important
to understand trends over the course of the Pilot term.
In addition, SFMTA acknowledges that the impact of the Pilot on the industry may be
confounded by other contextual factors such as the easing of COVID-era restrictions or other
SFMTA efforts to benefit the industry like increasing the taxi meter rate. We take these factors