AeronauticsARTIFICIAL INTELLIGENCE
Predictive AI makes selling and buying used bicycles dynamic and competitive
ALTEN’s VMO, specialized in software services, applied machine learning to help an online dealer of used bicycles develop a high-performance, predictive analytics platform. The platform automates bicycle value calculations, taking the guesswork out of pricing and ensuring a dynamic, rapidly adapting offer.
In today’s market, the huge variety and number of products available – new and used – make it essential to be competitive, both in pricing and in the ability to respond quickly and efficiently. This is particularly important in the used goods market, where customers are looking for online platforms that offer an optimum price-value relationship and make buying easy, fast and reliable. The leading platform in the US for selling and trading used bicycles engaged VMO holdings to help it respond to these needs.
Challenge: Develop an agile, dynamic platform for used bicycle trading to optimize the buying and selling experience
Solutions: A high-performance, predictive analytics platform that applies automated machine learning to accurately report on bicycle values
Benefits:
- Accurate price predictions
- Optimized user experience
- Streamlined sales and purchases
- Scalability
- Customer satisfaction
Key performance indicators
- 29% lower average prediction error compared to previous version
- 7% lower mean squared error compared with previous version
- <0.2 seconds AI response time
- 22.7 requests per second
Dynamic pricing
VMO drew on its expertise in software development and AI to create a platform that processes historical data to offer dynamic bike pricing suggestions and recommendations. Using the Value Guide (VG4) algorithm, the platform draws on data from the sales of tens of thousands of bicycles over the past 20 years and across millions of transactions. Machine learning and predictive analytics make it possible to offer precise bike value predictions, minimizing pricing errors. A searchable bicycle database facilitates easy comparisons and recommendations.
VMO’s VG4 algorithm is the latest in a series and is the first to be deployed in the production environment. In addition to the machine learning model for dynamic pricing of used bikes based on historical transactions, when the bicycle in question does not exist in the system or the owner has modified or replaced some components of the original bike, it recommends similar bikes. In addition, using explainable machine learning it provides a score for similar bikes and an explanation of their similarities in terms of components.
Satisfied customers
The VG4 platform simplifies the buying and selling process by recommending the best prices for both buyers and sellers. It can handle thousands of concurrent transactions, making it suitable for large-scale e-commerce. Thanks to automation, it reduces effort and improves the efficiency of estimating the value of the used bikes, thereby streamlining the process. The platform enhances user satisfaction by offering personalized features for a seamless, e-marketplace experience.
The power of AI has enabled this company to become the largest e-commerce platform for used bicycles in the US.
