In this presentation, I seek to improve the retention of hotel seekers on Yelp with a multi-level sorting based on qualitative attributes to provide trust for results and speed up the decision-making process. I go through user personas, pain points, use cases, potential solutions, prioritization, validation, metrics, and design.
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Yelp Product Challenge
1. Improving
Retention of
Hotel Seekers
0- Problem & Executive Summary
1- User Personas
2- Journey Map
3- Search: Use Cases & Solutions
4- Review: Use Cases & Solutions
5- Compare: Use Cases & Solutions
6- Action: Use Cases & Solutions
7- Prioritization
8- Final Solution & Metrics
9- Design
10- Calculations
Hisham Radwan
2. Problem & Executive Summary
The hotel industry is experiencing major demographic and technological changes. International leisure has
reached a record high in 2018 and tech savvy millennials continue to occupy a greater portion of travellers
(50% by 2025). Moreover, hotels are increasingly adopting seamless connectivity across platforms and
devices.
A significant portion Yelp users in the US comprised of high income individuals (72% earn 60k+), the segment
of the population that travels the most (75% of aforementioned income bracket identify as occasional or
frequent travellers) .The potential revenue from hotel seekers alone is over $100 000 000 annually. The ROI
of a 1% improvement in retention is $1 148 000. This suggests that Yelp should improve retention among
other business objectives to achieve the best product/market fit for hotel seekers. (calculations in last slide)
This goal of this proposal is to increase retention of hotel seekers by 10%. To improve retention of hotel
seekers by enhancing the value proposition of Travel and Hotel to users. The goal of the user is to efficiently
find and book the most convenient hotels. In this presentation, I identify the user needs, investigate and
prioritize potential solutions to best meet this user goal.
I propose a multi-level sorting based on qualitative attributes to provide trust for results and speed up the
decision making process.
3. User Personas
Michael Fleming
Michael Fleming regularly goes on business trips across
the United States. During his trips, Michael must be
within reasonable distance of this conference or
meeting. He also needs internet access to remain
connected to his clients and colleagues and to maintain
constant access to important files and emails. In the
mornings, Michael appreciates amenities like breakfast,
coffee machine, iron and ironing board to be ready as
he prepares for an important day at work.
● Age: 37
● Status: Single
● Context of Use: Laptop at work for travel
research.
Corporate Executive
Sarah Sterling
Sarah Sterling is currently planning a family trip to Paris
for the first time this summer. As a middle income
family, finding a reasonably priced hotel is important.
Recommendations from family, friends and online
reviews are influential in her choice of hotels. Sarah and
her family always seek to enhance their travel
experience with hotel packages. She is picky about her
hotel’s facilities which should include evening
entertainment, kids clubs and a pool.
● Age: 32
● Status: Married
● Context of Use: Home desktop for travel
research.
Public School Teacher
4. Journey Map
A Hotel Seeker’s Journey to Make a reservation on Yelp
Potential Activities
“Which hotels offer services that
I am looking for”
“I would like to see a summary
of the top results”
“I want to search based on my
standards”
Pain Points:
- Search results not tailored to
specific needs of different
travellers (ex: facilities,
amenities)
- Overwhelming number of
search results
“I want to see the best & worst
reviews”
“I want to find reviews of similar
hotels or from similar people”
“I want to see current discounts
and packages”
- Star ratings are subjective
- Reviews of different
customers value different
qualities and attributes (sleep
quality vs food)
- Baised and contradictory
reviews on travel sites
“I want to rank hotels based on
my criteria”
“I want to shortlist & compare
hotels instantly”
- Unable to compare different
hotels at once
- Unable to compare different
hotels based on desired
attributes that user values
most
“I want to make a reservation”
“I want to get a quote or
estimate”
- Quote may not be readily
available at time of comparison
- Unable to make a reservation
on the review site
5. Search: Use Cases & Potential Solutions
Use Cases
1. User finds hotels based on specific
services and features provided:
● Amenities
● Facilities
● Internet access
● Location
● Packages
2.User finds hotels based on specific
qualities:
● Sleep quality
● Cleanliness
● Food
● Customer service
Potential Solutions
1. Video reviews of 10-20 seconds along with
text instead of reading long reviews, (use case 1)
2. Filter by type of service such as amenities and
facilities in addition to location (use case 1)
3. Affinity-based search recommendations
based on user desired standards & quality (use
case 2)
4. Search results based on quality of service
(cleanliness, sleep etc) that users care about..
(use case 2)
6. Review: Use Cases & Potential Solutions
Use Cases
3. User wants to see the best and worst
reviews to make an objective evaluation
4. User wants to see hotel rating or read
reviews regarding attributes they care most
about etc:
● Sleep quality
● Rooms
● Cleanliness
● Location
● Food
Potential Solutions
5. Show hotel’s best and worst reviews side by
side to increase credibility. (use case 3)
6. Define quality of service for each star to
standardize evaluation. (use case 3 & 4)
7. Create tags out of frequently used words used
by reviewers to describe hotels. User can filter
for reviews that mention these words. (use case
4)
8. Provide labels for the most used description in
reviews per hotel. (use case 4)
9. Star ratings per attribute to allow users to
review based on attributes preferred (use case
4)
7. Compare: Use Cases & Potential Solutions
Use Cases
5. The user may want to compare selected
hotels side by side instantly,
6. The user wants to specify certain
qualities to rank providers and place
varying weights on selected attributes
Potential Solutions
10. Enable users to select providers and
compare them side by side. (use case 5)
11. Enable the user to choose 2 or 3
attributes to sort hotels by, in prioritized
order. This allows the user to compare
hotels by order of attributes they care
about most. For ex: sort hotels by sleep
quality first and then by food. (use case 6)
8. Action: Use Cases & Potential Solutions
Use Cases
7. The user may want to make a reservation
8. The user may want to get a quote or
estimate at the instance of comparison to
make a timely decision
Potential Solutions
12. Use chatbots to enable users to request
a quote or make a reservation. This helps
users close the deal.. (use case 7 & 8)
9. Prioritize the solutions based on impact to the user’s goal
and complexity of development. The user goal To efficiently
find and book the most convenient hotels
Short list of solutions that best meet user goal:
- Feature 4: Results of providers based on qualitative
attributes, provides a more reliable way for users to rank
providers than only using stars
- Feature 11: Multi-level sorting, makes it faster to choose the
right provider because it sorts providers based on attributes
that are important to the user
- Feature 12: Chatbots, provide immediate feedback to users,
which helps them make tangible progress in getting the
provider they want and scheduling the work.
- Feature 2: Filter by type of service such as amenities and
facilities in addition to location
- Feature 9: Star ratings per attribute to allow users to review
based on their preferred attributes
Prioritization
Solution Impact to user goal Difficulty of Implementation
Results of
hotels
based on
qualitativ
e
attributes
High: provides
customized results
based on user’s
values
Low: text mining to identify
important attributes used to
describe quality of service
provided. Also, machine learning
to find matches based on user
and reviewers’ attributes.
Multi
Level
sorting
High: reduces
decision making time.
(hardest part)
Low: multi-level sorting
algorithm
Chatbots Medium:helps users
close deal
High: chatbot for each domains
of service, and integration with
hotel’s database
Star
ratings
per
attribute
Medium: provides
assessment based on
attributes most
valued by user
Medium: front end development
to create stars for each attribute,
database to store ratings and
APIs to return average rating
Filter by
type of
service
Medium: narrows
down search to
hotels with specified
services
Low: sorting algorithm
10. Final Solution & Metrics
Final Solution:
Qualitative Attributes Based Search and Multi-level Sorting provide the highest impact for the
lowest effort.
The user specifies qualitative attributes during a first level search. Then the user applies multi-
level sorting to rank search results of providers based on specified attributes.
This solution provides a quantitative way of filtering search results as opposed to a purely
feature based search (47% of customers consider the sentiment of reviews a factor in their
decision making as opposed to the overall star rating). This provides higher degree of credibility
and trust in search results and speeds search process .
Key metric:
65% of Americans (from all income brackets) consider themself occasional travellers. Thus,
Repeated usage of new feature per user should be measured annually.
This metric should be compared between the new and the current feature.
11. Design
A new category, “Qualities”, should added below
“Sort By” filters to multi-sort the results (to
maintain consistency in design).
After clicking on “Qualities”, the user should be
prompted to a second screen. Here, the user can
type and specify qualitative attributes.
After specifying his/her attributes on the second
screen, three drop down lists enable users to
prioritize selected attributes as 1st, 2nd and 3rd
priorities.
The result is a list of providers sorted and ranked
based on attributes selected by the user.
Qualities