The Bay
Incentivizing user-generated reviews for increased customer conversion
Category
Collaborative Project/ Sprint
Duration
September 2022
1 week
Role
UX/UI Designer
Researcher
Tools
Figma
Overview
I collaborated with a team of Data Scientists, Web Developers and UX Designers to solve a problem posed by industry professionals from The Bay.
The Bay is focused on being a digital-first, purpose-driven retailer that “helps Canadians live their best style of life.”
Product designers from The Bay presented the following problem to our team,
“How might we alter our digital platform to further increase conversion, encourage repeat customers, and decrease returned items?”
To solve this problem we designed an updated review form that would incentivize users to provide a detailed first-hand review by offering discount codes. The level of detail in the review (assessed by written comments, photo and video uploads) would correspond to larger discounts received.
Our solution won the hackathon competition against five other teams - judged for collaboration, presentation and innovation.
My role involved leading an ideation workshop with two web developers and one data scientist alongside another UX designer, performing secondary research, sketching, wire-framing and designing the website screens.
Introduction
The Bay is focused on being a digital-first retailer and sells a range of items online, from clothes to furniture and everything in between.
Products like clothes are easier to buy online, and less likely to be returned, because of features like TrueFit offered by the Bay that give you confidence the product will meet your expectations. It’s far more difficult for big ticket items like furniture - you don’t know what that product will look like at home.
Problem Space
82% of shoppers surveyed in the UK believed in-store purchasing gave them more security when shopping for expensive items, knowing what a product looked like before committing.
According to the study, high value purchases such as cars, kitchens or furniture, could see drastically lower online sales implying lower conversion.
Furthermore, a mismatch between what the item looks like online vs in-person accounts for a significant portion of returns in the e-commerce space. It’s clear why people are nervous to make big-ticket purchases like furniture online, but there’s an opportunity to address this problem for the Bay and their customers.
Secondary Research
We completed a social listening exercise by reviewing social forums, like Reddit, for common challenges individuals face when purchasing furniture online.
Many posters shared they were hesitant to buy furniture online, and requested reviews from others who may have purchased the same item in the past.
Post purchase, customers were dissatisfied with items they received that didn’t match the online descriptions or photos on the website.
Shoppers who made successful online furniture purchases suggested looking at reviews with photos and in-depth comments from previous buyers.
Some say they only buy online if they can reference verified reviews with photos provided by customers, or they prefer to check out YouTube reviews of the furniture.
One poster even said they would only trust an item with an average 4/5 rating and more than 100 reviews.
This research led us to a refined problem space:
“How might we improve customers’ confidence in purchasing furniture online in order to increase conversion and customer satisfaction?
Persona
Based on our research, we developed the proto-persona - Sally Gournagour - a 28 year old operations manager living in Toronto in her tiny condo downtown.
Sally’s a busy girl and does a lot of online shopping. She recently moved and wants to buy furniture online but it’s so hard to envision how it’ll look in her space if she cant see it. Sally is frustrated that she might order furniture in the incorrect size and would have to return it. Imagine building the furniture just to box it up and send it back! Problematic for both Sally and the Bay.
Experience Map
We completed a mapping exercise to visualize Sally’s experience purchasing furniture online and identified opportunities to step in with a solution.
Solution
Guided by our HMW, and knowing who we were designing for, we started to brainstorm solutions. With less than a week to deliver a fully developed solution, we felt the pressure of the time crunch.
We initially considered a “furniture stylist” service. A Personal Stylist service is already offered by the Bay for clothing. Our hypothesis was that having personal, professional support to buy furniture that will fit well in your space would reduce refunds, encourage repeat customers and increase customer satisfaction. However, our developers highlighted the functionality that could not be built given the time constraints - chat features, booking functionality or even video call functionality.
With time constraints and technical constraints acting against us, we went back to the drawing board and continued to brainstorm together with the developers and data scientist to ensure that our idea was technically feasible.
Based on Sally’s experience and the research we completed, reviews appeared to be an evident driver for customer conversion.
We performed an audit of the review experience on 20 randomly selected furniture items on the Bay website. 14/20 had no reviews at all and only 2/20 had more than 8 reviews (although not a single one included pictures or video). That’s far from the 100 reviews necessary for some Redditors to feel comfortable making a purchase.
We realized there was an opportunity to incentivize buyers to leave detailed reviews to help other shoppers make an informed decision and reduce the chance they’d return their purchase.
We brainstormed many ideas, and given time and technical constraints, we designed a review form that incentivizes users to provide more detailed reviews using discount codes. If successful, this should reduce refunds from online shoppers who have access to user reviews, and encourage reviewers to become repeat customers to take advantage of their discount.
To be successful, we must assume that offering promo codes to all reviewers is financially feasible for the Bay. Financial analysis would be required to determine whether the increase in repeat customers outweighs the cost of the suggested discounts.
I sketched a simple user flow, demonstrating the actions a user would take to leave a review and receive their discount.
The other designer in my team and I split up the screens and started to design the experience within Figma. The developers collaborated with us within Figma and used our designs to begin developing the solution.
They used React and Tailwind CSS to build the interactive user interface. The back-end functionality of the review process was built using Node.js, Express and Prisma.
Prototype
The data scientist on our team requested that I design the framework upon which discount codes would be issued in order to use the logic for his model. I created a point system with which written reviews, and reviews supported with photos and/ or videos would be scored. Written reviews with enough detail would correspond with a 5% discount on a subsequent purchase.
This scoring system was brought to life using a “recurrent neural network” model built using Python. The model can learn the semantic meaning of the first 200 words in a comment and then score the reviews based on usefulness/helpfulness.
Written reviews supplemented with a photo would receive a 10% discount and written reviews supplemented with a video would receive a 15% discount.
EJS was used to generate the email containing the discount code, sent once the review is submitted.
Outcome
Our final developed solution was judged by three Product Designers from The Bay. Our team won the hackathon competition based on the demonstrated collaboration between each discipline to develop the final solution as well as the innovation behind the neural network powering our variable discount incentive.
The representatives from The Bay were particularly fond of the idea as they shared that a lack of good reviews is a business issue that the company has recognized and is actively seeking to solve!
Next Steps
As designers, we know that no project is complete without testing our solution with users. Ideally, we would complete user testing for our proposed solution to see if the incentives have the impact we expect and further iterate on our design.
As previously mentioned, we would also require financial analysis to determine whether the increase in repeat customers outweighs the cost of the suggested discounts.
Key Learnings
The main learning I took away from this project was about working within the technical constraints set by developers. I gained a real understanding of what is possible based on the technical abilities of the developer I’m working with, and the time constraints they’re working against.
A learning I will carry into future work is how to have an effective and productive discussion with stakeholders to arrive at a solution that works from both a design and development perspective.
