Facilitating Decision Making: Boosting CLV using data-based recommendations
Imagine being locked in a room with a group of aspiring designers, a ticking clock, and a product that needs your touch. That’s exactly what happened during the 48-hour design sprint I did with my peers.
The goal of the challenge was to put our problem-solving skills to the test and push ourselves to set new bars as designers. With a group of 6–9 people and a product to work on, each member was given a unique problem statement to tackle within the 48-hour time limit. It was a true test of our abilities and a chance to understand ourselves as designers.
Step back in time with me and get ready to experience the intensity, the adrenaline rush, and the camaraderie that brought the best out of us.
Product
Booking.com: Booking.com is an online travel agency that allows users to book accommodation, flights, and rental cars in destinations all over the world. The platform offers a wide range of options, from budget-friendly hotels to luxury resorts, and allows users to filter their search results based on specific criteria such as location, price, and amenities. Users can also read reviews from other travelers, view photos of the properties, and make secure payments directly through the website.
Problem Statement
Implementing User Data-Based Recommendation: The User Data Recommendations feature will use data from our user’s past bookings, searches, and reviews to provide personalised recommendations for their next trip.
Problem Brief
Context: Booking.com’s lack of personalised recommendations makes it challenging for users to confidently choose their next vacation stay or accommodation. The process of searching and exploring options can be time-consuming, yet users may still feel uncertain about their final decision.
Core Idea of Feature: The User Data Recommendations feature will use data from our users’ past bookings, searches, and reviews to provide personalised recommendations for their next trip. These recommendations will be based on their preferences and history with our platform and popular destinations and accommodations among similar users.
And for new users, this data will be shown as per their persona based on demographical information, age, and gender and relating them to the cluster of existing users with the same persona.
How it Helps Users: The User Data Recommendations feature will save users time and effort in planning their next trip. It will also increase their satisfaction with the booking process, as they will feel confident in their decision based on personalised recommendations.
Constraint: I need to ensure that the recommendations are accurate and relevant to the user, which will require a robust data analysis and recommendation algorithm. There may also be privacy concerns to consider when using user data in this way.
Goals
- Efficient Decision Making: The recommendation feature can save users time and effort in their decision-making process, as well as increase the chances of them finding a product or service that meets their needs.
- Increasing Customer Lifetime Value: Suggesting personalised and relevant accommodations and stays, leading to repeat bookings and satisfied customers.
Secondary Research (Desk Research)
Given the limitations of time and the responsibility of meeting deliverables, I initiated my research process by conducting desk research to gain valuable insights and gain a deeper understanding of the current user experience on booking.com.
Traffic Overview
Out of 1.45 billion users (in a span of 3 months), 63.5% of them access the website with mobile phones. And the interface for both the app and mobile-web version is quite similar except for the bottom navigation.
User Engagement Analysis
When we compare the average time spent on the site and the average number of pages visited by users, we see that people spend more time and visit more pages on Booking.com (on the right) compared to MakeMyTrip (on the left). This is generally a good thing, but it may not be the ideal situation while making decisions, as MakeMyTrip is a direct competitor to Booking.com in India.
There could be various reasons for this, such as users leaving the site before completing their task or the purpose of their visit. However, the increased time and page visits on Booking.com may be due to indecisive exploration.
Projected Growth among Top Competitors
A case study showcasing the current state of the market and future growth prospects of Indian travel space between 2018 to 2028 with booking.com as 2nd most preferred site.
Competitive Analysis
In order to gain insight into the various solutions being implemented by our competitors in the field of personalised recommendations, I conducted an analysis of the approaches utilised by similar apps in the same industry.
NOTE: The provided data for the designated applications has been accumulated as new registrations, indicating that there was no prior user information available concerning their habits and preferences.
MakeMyTrip: Although MakeMyTrip does have a dedicated option to explore locations, vacations, and trips, it does not send personalised recommendations based on user behaviour and data within the feed.
Agoda: Even Agoda has a discover option with some recommendations at the bottom of their homepage but no personalised discovery for new users.
Trip Advisor: TripAdvisor, as a “trip advisor,” does not provide personalised recommendations for new users.
User Journey
Moving forward I wanted to understand how users currently explore accommodations and how does the decision-making process looks like, so I mapped user journeys starting from opening the app to checking inside the hotel.
Hypothesis
After thoroughly conducting secondary research and gathering a significant amount of information, I felt that I had a comprehensive understanding of the context and was ready to move on to the next step in my research process — formulating a hypothesis.
- Personalised recommendations will lead to increased engagement with the product or service is recommended.
- Recommendations based on user data will result in higher conversion rates (e.g. more users making a purchase or taking some other desired action) compared to recommendations that are not personalised as it makes decision-making easy.
- Users will have a more positive experience with a product or service when they receive recommendations that are tailored to their interests and preferences.
- Recommendations based on user data will result in higher retention rates and will reduce users’ time scrolling rapidly for the preferred option.
- Users who receive recommendations based on their data will be more likely to try new products or services compared to those who do not receive personalised recommendations.
- Data-based recommendations will make the discovery experience delightful and less frustrating giving the user near to what they prefer.
Market Research
I began by identifying the most popular apps in a different category and downloaded them on my device to test them out. I paid attention to how these apps suggested content to users and how they tailored the recommendations to each individual user. Also, one important thing to study was when are they recommending a service/product to the user.
- Zomato’s restaurant page suggests food items based on user data before the menu begins.
- Spotify suggests artists, tracks, and albums based on the user’s past listening history and preferences.
- Amazon Shopping displays similar products based on the user’s search behaviour and product category below the listed item’s description.
Ideation & Wireframing
With newfound clarity, I understood what actions were necessary to provide a better recommendation experience for the user. However, the challenge of determining the specific methods to achieve this goal remained. To overcome this ambiguity, I turned to my diary and began sketching as a part of my ideation process, searching for a clear path forward.
- Home Page: The homepage would consist of a very visible and discoverable section of recommendations labelled as “Pick For You” to help the user with decision-making rightly upon arrival.
- Search Result Page: In the SERP (search engine result page) page the cards will consist of a label over the image signifying that the stay is recommended, in a very subtle way without making the screen visually heavy.
- Search Result Map: With the map view available a specialised location pin icon would signify recommended location and upon clicking they got a pop-tray regarding the information from where the user can proceed to book.
- Property Overview Page: In case the user isn’t satisfied with the current property that they are viewing then by the end of the screen (which consists of info regarding the property) there is a section of recommended options as a nudge to pull them out of discovery frustration.
- Quiz Page: An additional idea to tackle the edge case of showing relevant information to new users (for which there is no user data) we allow them to participate in a low-effort 5-question “this or that” quiz format to have an idea of their liking along with their demographic data. Here we incentivise the user with a 10% discount on their first booking which currently booking.com is giving away for free.
After the ideation process I prioritised and decided (considering the user journey) to move with adding recommendations on the home page, a search result page and a property overview page as of now due to time constraints. And the quiz page and recommendation map could be worked upon as a part of future developments.
Optimizing UI through Usability Testing and Iteration
To speed up the UI development process, the team determined that it was necessary to create some custom design system for the booking.com project, as no suitable open-source option was available. This design system included style guides, an icon library, components, and established colour and text styles. Once this foundation was in place, I started with designing and prototyping the UI for usability testing and then iterating based on gathered feedback.
Usability Testing Guide
For the evaluation of the usability of the newly designed interfaces, I recruited three volunteer participants. These individuals were asked to observe each of the three screens on which the recommendation feature had been implemented, and verbally express their understanding and perceptions of the information presented on the screen.
Following this, I provided the participants with the opportunity to independently explore and interact with the application without any limitations or constraints. The objectives I aimed to accomplish during this remote usability session were:
- The user’s understanding of the recommendation section on the homepage, including their scanning and perception of the information presented in the cards.
- If they can grasp the distinction between suggested options and a general listing of properties on the search result page.
- Scrolling and decision-making behaviour on the property overview page.
- Exploring alternative options if the current property being viewed does not align with their desired preference.
Home Screen
The original version of the app did not display any recommendations on the home screen, even though this page typically receives the most traffic.
A new section has been added to the home screen to ensure that recommendations are visible as soon as the user arrives. The cards in this section provide a lot of important information that can help the user make a decision about where to stay. This includes information about the amenities available at the property, the ratings it has received, any discounted pricing that is currently available, its location, and an image of the property. There is also an option for the user to save the property to their list for future reference.
Feedback from Usability Testing: During usability testing, it was brought to my attention that the pricing format for the hotel booking was causing confusion among users. They were unable to discern if the pricing was for one day, one night or any other format.
Taking the user feedback into account, in the final version of the product, I made sure to clearly specify the format next to the price in order to improve understanding for our users. This change was made to ensure that users have a clear understanding of how the prices are presented, which will hopefully make it easier for them to make informed purchasing decisions.
Search Result Page
Scrolling through numerous properties and clicking on each of them to read a significant amount of information can be a tedious and time-consuming process. To make the process more efficient and save time, it would be beneficial to explore a limited number of recommended options based on behavioural patterns and data. This way, users can easily focus on properties that are more likely to meet their needs and preferences, rather than wasting time on properties that may not be a good fit.
A banner was added to the top of the image on the product card to signify that the card is specifically tailored for the user.
Feedback from Usability Testing: During usability testing, none of the users noticed the banner and was unable to differentiate between the recommended and generalised options. Additionally, when discussing this page, the users mentioned that they were being disturbed by the way other information in the cards appeared.
We considered the user feedback and realised that most of their visual and peripheral attention was focused on the informative area of the card. Therefore, I added a chip to this area and also incorporated a bright colour tone to make it more noticeable and appealing. These changes were made to draw attention towards the recommended chip.
Property Overview Page
The original property overview was lacking recommendations and had too much information. The goal of recommendations is to make decision-making easier, but this design does not allow the user to easily explore other options if they are unhappy with this property. The only way to do so is by starting a new search or navigating back to the search results page. It would be more convenient if the property overview included recommendations and was more streamlined.
A recommendation section similar to the one on the home page was added to the end of the property overview page. The purpose of this was to allow users to explore more options after they had learned all the information about the viewing property on the page.
Feedback from Usability Testing: During the testing session, it was found that users were hesitant to scroll down so far on the screen and the review and facility section seemed to be a saturation point, beyond which users were not inclined to scroll further. As a result, it may be necessary to reconsider the placement of the recommendation section in order to make it more accessible to users.
After conducting testing sessions to understand how users were behaving with the initial design, it was discovered that the recommendation section was not as easily accessible as it could be. Therefore, I decided to move this section below the facility listing section in order to make it easier for users to find and access. By placing the recommendation section in a more prominent location, I hope to encourage users to discover and utilise this feature more often, ultimately leading to efficient decision-making.
Prototype
End Note
On a personal level, it was an opportunity for me to learn more closely about user behaviour and to work with them in a hands-on way. Starting with a blank canvas and ending the 48-hour mark with an iterated UI and prototype gave me a huge sense of understanding of my skills as a designer. Lastly, I want to express my deepest gratitude to my teammates (Saurav, Prithvi, Kathryn, Garima, Travis, Anirudh and Swarupa) for pushing the whole group upwards throughout the challenge. Peace Out!
