Professional Documents
Culture Documents
Jam Session
Magical Experience at Events
Fulvio Cozzi
About me
● DoorDash (2021-2023)
○ Feed ranking
○ ETAs
○ Growth
● Amazon (2014-2021)
○ Inventory guidance for Sellers
○ Inventory guidance for Amazon Retail
○ Inventory health
○ Middle mile network topology
○ Middle mile transportation scheduling
Fun fact about me
● Last time I gave a presentation was 10 years ago, explaining network topology,
without transportation background, on my first 30 days at Amazon….
Agenda
- Context
- Is this a problem worth solving?
- Users
- Problems
- Solutions
- Roll out / Metrics
Why?
Context
Uber
Ridesharing
- 2-sided marketplace of Riders and Drivers (+
Advertisers)
- Riders: seek reliable and affordable ways to move
from A to B
- Drivers: seek ways to generate dependable
earnings
Uber
Marketplace efficiency
- An efficient marketplace minimizes wait time for all sides
- Riders: get faster from A to B, at a more affordable price, with higher
certainty
- Drivers: spend more time earning (vs ‘open’: waiting for dispatch or
‘en-route’: driving for a pickup)
- Wait times are low when:
- Marketplace participation is high: there are many riders and drivers, so they are more
likely to be close to each other and are more efficiently matched
- Healthy Balance of riders and drivers: for a given time-geo, the number of ‘open’ cars is
comparable to the trip requests, so matches and pickups happen quickly
Uber
Marketplace efficiency for Events (1/3)
- Events might create a market imbalance between demand and supply
- Riders: request trips for the same place at the same time, usually close
to the event start / end time
- Drivers: receive a large number of requests close to the event start /
end time, receive few or no requests in between
Uber
Marketplace efficiency for Events (2/3)
- Some signs of a market imbalance:
- Higher requests per time period and geo location
- Higher ETAs
- Higher Prices
- Lower trip completion rate
Uber
Marketplace efficiency for Events (3/3)
- Rebalancing mechanisms (surge, Boost+, Quest) are somewhat effective
in delivering reliability and increasing supply
- Many challenges with events still remain unsolved:
- Congestion
- Road blocks
- Hard to reach designated pick up/drop off areas
- Riders and drivers unable to find each other
- Riders and drivers cancel on each other
- Riders can’t secure trips, drivers don’t feel compensated enough
Uber
North Star for this case study
- Product Goal: Create a magical experience at events for Riders and
Drivers
- Measured as: Wait time reduction for the network as a whole
Uber
Why?
Uber
Mission
- Magical experience for riders and drivers at
events is aligned to Uber’s mission to reimagine the
way the world moves for the better
Uber
Market attractiveness
Events have an attractive $4.5B Total Addressable Market (TAM) (23.3%
Uber mobility), TAM is expected to grow 5.6x by 2030 (data)
Uber
Right to win
Uber is well positioned to tackle the mobility challenge
- 74% rideshare market (as of: Aug-2023, source: Bloomberg)
- $4.9B in cash and cash equivalents
- $4.9B in mobility revenue
- 137M in Monthly Active Platform Consumers (MAPCs) (as of: Q2-2023, source:
Uber earnings report)
- In addition:
- World class demand and supply of transportation
- Smart matching, advanced routing and pricing
- Strong brand and safety record
Uber
Recommendation
Invest in improving Events
Uber
Why?
Users
Uber
Users
Riders
- Use case: sport events, concerts, festivals
- Who they ride with: solo, significant other, in group
- Engagement level: Uber One vs Classic
- Demographic: teenagers, adults, senior citizens
Drivers
- Use case: sport events, concerts, festivals
- Rating: Highly rated vs not
- Lifetime trips: 50+, 150+, 250+, 500+
- Type of service: UberX vs Uber Black
Uber
Why?
Problems
Uber
User Research: Drivers (1/2)
Uber Drivers hate working for Events
Uber
User Research: Drivers (2/2)
Uber
User Research: Riders (1/2)
Uber Riders don’t enjoy riding to or from Events
Uber
User Research: Riders (2/2)
Uber
User journey
Riding to a sport event:
- Request
- Pickup
- Ride to event
- Drop off at event
Uber
User journey prioritization
User journey: prioritize riding from a sport event (request + pickup)
Uber
Problems
Riding from a sport event:
- Request from event
- ETAs are too long
- Route is unfeasible
- Fare is too high (rider) too low (driver)
- Rider and driver not getting matched
- Received a cancellation
- Pickup from event
- ETAs are too long
- Route is unfeasible
- Can't find rider/driver
- Received a cancellation
- Rider/Driver unresponsive
- No/bad phone service
Uber
Problems prioritization
Problems: prioritize
- Request from event
- ETAs are too long
- Fare is too high / too low
- Rider and driver not getting
matched
- Received a cancellation
- Pickup from event
- ETAs are too long
- Received a cancellation
Uber
Why?
Solutions
Uber
Why this is a hard problem to solve
In the best case scenario, we need to facilitate 20 pickups per minute or 1
every 3 seconds, to absorb 15% of an event demand within 45 minutes.
Uber
Solutions (1/4)
Demand shaping (time)
- Riders are offered the option to schedule a ride through the Uber
app
- Trips can be scheduled when purchasing the ticket, as an upgrade,
for events where Uber partners with the Event Organizer
- Trips can also be scheduled when riders are in the event geofence
- Riders schedule trips through a time picker, the picker incentivizes
scheduling slots with low utilization
- ETAs and prices are used as incentives
- Larger incentives are offered for pickups before the event
completion
Uber
Solutions (2/4)
Demand shaping (space)
- Several designated pickup areas are created around the event in partnership with
Venues and Event Organizers
- Trips can be scheduled when purchasing the ticket, as an upgrade, for events where
Uber partners with the Event Organizer
- Trips can also be scheduled or booked on the spot when riders are in the event
geofence
- Trips are allocated to the area with the lowest utilization
- Riders receive walking directions to their pickup area through the app
- Directions are provided with richer details to the navigation basemap, including building
outlines, entrances and other features such as crosswalks
- Signs and on-site Uber operators facilitate the process
Uber
Solutions (3/4)
UberBus Shared
- UberBus Shared is a higher capacity vehicle service with fixed routes available the day of the
event only
- Trips can be scheduled ahead with the event tickets as an upgrade, or through the Uber
Explore tab
- Trips that have been scheduled ahead can be canceled free of charge up to 1 day before the
event
- Trips can be booked on the spot, through the Uber app, when there is available capacity and
the fixed route is compatible with trip origin/destination
- Trips can be booked for 1 or multiple people
- Trips pickup points have to be along the fixed route
- UberBus Shared uses higher capacity vehicles like vans (15 seats) and charter buses (56 seats)
to maximize asset utilization
- On the riders side, rides are 40% on average the price of an UberX Shared
- On the driver side, earnings are priced comparably to Uber Hourly Uber
Solutions (4/4)
UberPool Shared
- UberPool Shared brings online on the Uber platform all the supply of cars from event attendees
who aren’t registered as Uber drivers yet
- Car owners can sign up as drivers and start driving in 2 minutes or less through priority
onboarding
- Car owners can input their available seat capacity and drive as a one-off, getting paid through
their Uber rider payment method (if they are Uber riders) or Apple/Google Pay
- Riders and Drivers can decide to match with each other through the standard matching
algorithm or offline through QR codes
- The same safety features and upfront pricing will be available for these trips, irrespective of
how riders and drivers were matched
Uber
Solutions prioritization
Solutions: prioritize UberBus Shared
Uber
Why?
Rollout
Uber
Assumptions
Assumptions:
- Riders will be excited to take a higher capacity vehicle on a fixed route
to travel to and from the venue (riskiest)
- We will succeed in procuring sufficient supply of such vehicles (riskiest)
- Drivers will be excited to give up some flexibility in exchange for a
steady hourly income the day of the event
- We will succeed in operating the fleet
- Fewer vehicles on the road will decrease congestion, decreasing wait
time across the network (riskiest)
Uber
Phase 1
Riskiest assumption testing:
- Fewer vehicles on the road will decrease congestion, decreasing wait
time across the network (riskiest)
Success:
- ETAs across the network <0
- Trip completion rate >=0
- Demand >=0
- Supply >=0
- Price <=0
Uber
Phase 2
Riskiest assumption testing:
- Riders will be excited to take a higher capacity vehicle on a fixed route to travel to and from the
venue (riskiest)
- We will succeed in procuring sufficient supply of such vehicles (riskiest)
Methodology: ops test
Ops test Instructions:
- Pick one basketball event in a smaller suburban venue in the US
- Design a fixed route from a high density departure point like a train station
- Partner with Event organizers to offer a $5 flat fare one way $7 round trip for the event
- Send marketing emails, riders to opt-in through a Google form
- Uber Ops teams on-site to facilitate pickups and dropoffs on the day of the event
- Partner with local institutions to hire school buses for the event
- School buses are attractive because their demand seasonality (weekdays mornings and early afternoons) is the
opposite of events (weekend late evenings and nights)
- Post-experience Customer Satisfaction Score (CSAT) and 30-min qual assessment with both riders and drivers
Uber
Testing Goals
- Find Product Market Fit
(PMF) with both Riders and
Drivers
- Monitor entire funnel while
focusing on:
- Adoption
- # users purchasing
seats
- # users completing
their first ride
- Retention
- D28/D1
- D84/D7 Uber
Growth
Principles:
- Grow slowly, while retaining a large number of users
- Grow from one basketball event into more basketball events for the
same venue
- Grow into all sport events for the same venue
- Grow into more venues
- Grow into more use cases (concerts, festivals)
- Grow into more flexible mobility options
- Routes with multiple spots, pickups along the route
- More than one route, dynamically created based on demand and supply
- Dynamic routes, so pickups can happen when Riders request vs fixed points
- Integration with Uber transit, Uber Explore
- Ability for Uber to buy capacity and lease it to Drivers, help them grow Uber
Risks
Cannibalize the existing Uber mobility experience
- Riders might rely more on UberBus Shared and less on other mobility options
- Drivers might receive fewer trip requests
→ Not concerning, in line with Expanding Access and serving anyone’s need for mobility
Slower service
- Riders might have to wait for the bus to fill up
- Buses might be slower than other mobility options
- Drop offs might be more limited than other mobility options
→ Not concerning, it might be true for isolated cases, but not at a network level in the long run Uber
Why?
Thank you!
Uber