Creating an AI-powered UI Design for a Furniture E-commerce Platform
Highly intuitive interface solutions and 3D visualization AI-based design technology provide photorealistic renders of interior and exterior settings for a global furniture e-commerce platform.
Industry:
E-commerce
Team:
1 PM, 4 Developers, 1 DevOps
Project State:
October 2018 - August 2024
Country:
Israel
AI Development
UX/UI Design
Web Application
Background
The client, a furniture-focused e-commerce company, tasked us with creating an AI-based interior design virtual stylist to transform online shopping. This innovative solution involved developing a user-friendly interface and an immersive virtual stylist for furniture selection.
We started with a questionnaire on the website, used its results to develop an algorithm for furniture selection, and then built a 3D room design with the chosen items. This comprehensive and challenging project aimed to enhance the shopping experience by providing necessary context for confident purchases.
Outcome
DBBS developed a user-friendly e-commerce interface featuring an innovative virtual stylist. This AI-powered solution significantly enhanced the shopping experience, leading to impressive results:
- Revenue Growth: Increased by 30%
- Average Order Value: Rose by 12%
- Conversion Rates: Improved by 16%
- Time on Site: 2x
These enhancements not only improved user engagement but also boosted client's key business metrics, reinforcing their market leadership in online furniture sales.
To explore more about the paramount importance of an optimized UI tailored to the needs of e-commerce businesses in the interior design space, look at Reimagining Interior Design E-Commerce: The Power of Intuitive UI.
Main Product Features
Easy and seamless integration with vendor sites is implemented through GA scripts.
Analytics from vendor's sites are collected using the Facebook Pixel approach.
Based on this analysis, user-oriented designs are created (different for each logged-in user).
The design creation algorithm prepares a list of recommended furniture items and up to seven alternatives for each of them based on the single selected item.
Business context
The client's company provides several interactive widget solutions for online furniture stores worldwide. Therefore, the client needed that:
- Each solution should have the ability to be integrated into the vendor's website
- Data for these widgets should be fetched from vendors in different ways.
Challenges
#1
Vendor Database Synchronization. One of the major hurdles we faced was synchronizing many furniture pieces with our customer stores' (vendors') databases. This required meticulous planning and execution to ensure a smooth integration process.
#2
Multi-Stage Client Integration. Integration with different clients, which have their stores, and various technologies are used. Therefore, the process was simplified in several stages.
#3
Complex Algorithm Optimization. The algorithm is quite complex and involves many queries to the database that are difficult to combine, and the requirements for the speed of building a design are 4-5 seconds.
Solution
DBB Software team was responsible for front- and back-end custom software
development utilizing Python, database management, deployment, feature integration, and system monitoring. Four products were developed:
- CTL: collage with images of furniture items.
- STR: photo of 3D-designed room with 3D-designed furniture.
- Virtual designer: a questionnaire that prepares designs based on users' answers (CTL or STR is configured on the vendor's side).
- 360-degree iFrame with one furniture item with the ability to rotate it.
The developed technologies made an online shopping platform more convenient and feature-rich, with benefits such as a wide furniture selection, interior stylist options, and a user-friendly interface.
Correlated Deep Tagging hyper-personalizes the shopping experience by making products more discoverable and improving the tagging accuracy with computer vision AI. The created methodology elevates and inspires online experiences, creating visually appealing interactions throughout the funnel.
It improves filtering based on the categories shoppers want to explore by adding context when making a match. The methodology delivers the most accurate results for small vendors and marketplaces with significant inventory sources.
This solution works on multiple visual discovery cues that shoppers can select, combine, or remove until they find precisely what they want. In a way, it turns a product catalog into a tailor-made one. Filtering out items that do not fit the exact criteria for a more accurate search alleviates shopper frustration.
It focuses solely on effective product discovery and personal style, enabling personalized style-based recommendations like similar items and completing a whole look. We provide the technology that bridges product discovery and purchase, providing the ultimate tools for increasing conversion and AOV and cementing brand loyalty.
Implementation
DBBS team followed a series of planned steps to develop a software solution tailored to our clients' needs. Here’s an overview of each stage:
An in-depth study and analysis were conducted to understand clients’ specific needs and requirements, providing a foundation for the development process.
The process is built on the agile methodology:
- Sprint is scheduled for two weeks
- Tasks are added to the monday.com board
- Tasks are divided into three main groups
a. current sprint
b. next sprint
c. backlog - Once every two weeks, the team estimates tasks from the backlog, and the product owner creates the next sprint. Twice a week, for 30-60 minutes, the team meets with the customer (sprint planning, retrospective, demo).
The team worked in an agile manner, developing and integrating the technologies. Practical tips for easy process navigation and optimal results when creating realistic room designs with 3D rendering can be found in Leveraging 3D Room Design and Cloud Platforms for Realistic Rendering.
The UI/UX team ensured the virtual design was translated into an intuitive and user-friendly interface. Python was used to integrate with the machine learning model. Our software development team leveraged a blend of cutting-edge technologies and frameworks to execute this project.
Developers conducted thorough reviews and moved changes to a staging environment for QA testing, pinpointing any problems.
After deploying the features, the team gathered user feedback to inform continuous improvement and refinement.
Ongoing stage:
- We are currently working on a new microservice architecture for an existing solution
- Training Chat GPT collects parameters for existing algorithms of design creation from users through on-site chat
- Using graph-oriented DB (Knowledge Graph) to optimize the performance of the design creation algorithm
- Preparing storybook for React component reuse across all features
Technologies
Explore More Cases
How can we help you?
Mina Morkos
Business Development Manager
“Harness our decade of expertise to save your money. Thanks to our DBBS platform's efficiency and ready-to-use components, we reduce development time by up to 50%.
Curious to see how it works? Fill out the form to book a free consultation!”
mina.morkos@dbbsoftware.com
Mina Morkos
2024 DBB Software. All rights reserved