Uber: Applying Machine Learning To Improve Customer Experience
Uber: Applying Machine Learning To Improve Customer Experience
Uber: Applying Machine Learning To Improve Customer Experience
Uber started as a luxury car service that catered to high-end customers and from there it evolved to
a mobility service that caters to every person who desires to go from place to place. Since Uber
targets all the market segments it has to create different personas or archetypes according to the
services used and the reason of their use based on their frequency to effectively formulate ideas to
personalize the experience which will best suit the customers and minimize or remove their pain
points.
The customer persona for Uber is divided into 5 broad categories each with its expectations and pain
points from the service. The personas are;
Premium – The riders who desire of elite transportation service that a mobility service can offer are
in this category. Premium customers are willing to pay a huge sum for premium services that are
tailored according to their needs the customers mostly include professionals who are traveling out
of their business obligation and their traveling expenses are covered by their company.
Unavailability of cabs: since the premium segment of Uber has the least number of cars, the
unavailability of cabs can be a big displeasing factor for the rider.
ETA vs ATA: Since we are talking about working professionals the difference between the
estimated time of arrival and the actual time of arrival is an important parameter that can
put off their schedule.
The number of steps walked by the rider: if the rendezvous point makes them walk their
discomfort will grow with every step as they are paying a hefty sum and so they wish to be
treated in a personalized manner. Anything apart from the elite will lead to discomfort and
hence discontent.
High-Frequency riders (HFR): These are frequent users of the service who typically avail the services
several times a day. Generally, they take a short ride in densely populated urban areas.
The possible pain points for these types of riders can be:
Event and real-time diversions and congestion: Densely populated areas can have
congestion along with a potential hotspot or some type of construction work or lane closure
which might make it difficult to navigate and track sometimes for the driver and the
ambiguity of the situation will irritate the rider.
Location Uncertainty: In high traffic areas pickup can become difficult as the driver might
end up on the opposite side of the traffic due to inaccuracy of the map. This will waste the
time and can add up to the rider’s frustration.
ETA jump: The driver’s ETA might change from what was mentioned initially due to traffic,
this might leave the riders to feel confused and discontent with the service.
Commuters: The riders who use the service every day to go to work and then return from it are
commuters. They are identified by the same origin and destination for most of the trips.
Waiting time: Commuters use the service to get to work and they would not like to be late
especially waiting for the cab to arrive, this might be the biggest pain point for the
commuters.
ETA vs ATA: For Commuters whose reporting time is fixed the difference between ETA and
ATA can be very frustrating.
Repetition of instructions: Since the commuters are booking can from the same location
every day, it would be a pain for the rider if they have to instruct the same things to the
driver every day, for example how to get to their place if they are living in a densely
populated area.
Social: The riders who use Uber for leisure and social activities falls under this category, the rider
uses the service to go to restaurant, concerts, or bars. These users are most active during evening
hours and weekends.
Surge Price: The price is usually high during odd hours or hours of high demand, especially in
a social pocket of an event like concerts, pubs, or theater after a movie. Such high cost of
commute can be rather off-putting for the riders
Event and real-time diversions and congestion: The navigation and commute can be difficult
along an area of congestion, even the pickup is difficult in an area where some event has just
ended, as a theatre after a movie or pick up at a concert location after the concert has
ended; it is confusing and irritating for both rider and the driver.
Driver cancellation rate: it is seen that drivers prefer to pick riders who will go in the same
direction as the driver’s home in late-night when a driver is planning to retire for the day. If
the rider is going in a different direction the driver sometimes cancels the ride, if it happens
multiple times the rider feels very frustrated.
Travelers: These are the riders who travel to airports or railway stations. They can be business
persons, family on vacation, or a tourist.
ETA vs ATA: the riders who are waiting to pick up from the airport would not like to wait for
too much or the rider who has to go to an airport will have this anxiety of being late, this will
add up to negative customer point
Pickup location ambiguity: often the riders are confused about the pickup location at an
airport or railway station, as the cab has a designated place till where it can come and also
there are numerous terminals which add up to the confusion, the communication gap
between the driver and rider can make it even worse.
Luggage handling: Most of the time this kind of rider travels with lots of luggage, it can be
harrowing to load and unload the luggage by themselves and driver’s help is something they
will appreciate
Competitor analysis:
SWOT ANALYSIS
Strength
1. Data is available in abundance. This historic and telemetric data can be used for making
machine learning algorithms to develop ML models.
2. They collaborate with city and state administrators to ensure Uber drivers can efficiently use
roads and follow traffic and parking rules.
3. Uber collaborates with cities to provide anonymized data which can contribute to decision-
making for the administrators to develop existing infrastructure, minimize congestion, and
invest in efficient transportation.
4. Uber provides different modalities to different segments of customers. In addition to this,
Uber also provides different experiences to a different segment of consumers. This ensures
repeated use of Uber.
5. Uber provides Insurance to the rider, driver, and car. This will ensure a safe environment for
them.
6. They are constantly working on analyzing the customer pain points and improving the pickup
experience through automation.
7. Uber has entered into the food delivery market as UberEats. This can allow them to cross-
sell their services giving them a competitive advantage.
Weakness
1. Every stakeholder involved (rider and driver) are irrational. They make the decision based on
heuristics and their decision depends on their behavior. This can hamper the pickup
experience and reduce pickup quality metrics.
2. Their services can be easily imitated. The advent of high-end technologies like GPS, ML, etc.
along with the high penetration of smartphones has given Uber’s competitors like Yandex (in
Russia), OLA (in India) to challenge a global player like Uber.
3. In densely populated parts, the accuracy of GPS signals can be degraded because of tall
buildings and walls. This will give a marginal error of 50m or more when pinpointing the
location.
Opportunity
1. The advancement of new technologies like the wide acceptance of GPS navigation systems,
and the huge penetration of smartphones has increased the usage of Uber in developing
countries like India.
Threats
1. The rise of competition in developing countries like India, China, and Southeast Asian
countries has challenged Uber from acquiring a bigger chunk of market share.
2. Extensive market research is needed if Uber wants to enter a new market because of
cultural differences in other countries.
Issues
Uber drivers can park a little farther from the actual destination due to drives phone’s GPS
not being accurate enough to pinpoint the exact location
Drivers may reach the pickup point only to find the location obstructed due to it being in a
gated community, on a busy street, or at the airport forcing them to park at a distance
For customers using rides frequently or for social activities, a major pain point would be
surge charging and overall higher price in general
Wait-time while picking up or delay during commuting is also one of the major pain points
for high-frequency riders and commuters
Another major pain point in line with premium customers would be the level of comfort
during the rides from baggage handling to the accuracy of a pickup even if it costed money
Solutions
One of the solutions suggested would be using automation to refine pickup heuristics to
accurately pinpoint the location
They also used GPS locations and manual entry by the rider to increase accuracy
Using metrics like quality score and pickup happiness to determine the quality of ride and
enhance the experience accordingly
They utilized historical data of customer’s pickup location and destination to enhance
customer’s experience via Machine learning
Assigning spots at the airport and stations to ensure customers and riders don’t have issues
while deciding or locating a pick-up point
UberPool, providing offers and riding passes for those customers who wanted affordable
price