Interactive A/B Testing Dashboard
Interactive dashboard for a store enabling A/B testing between different recommendation algorithms, with real-time analysis and comparison of algorithm performance to maximize profit.

The challenge
A retail store couldn't tell which recommendation algorithm drove the most profit.
What we built
Data pipeline, A/B testing framework, and interactive performance dashboard.
Impact
Clear insight into best-performing algorithms, seasonal trends, and product combination patterns.
Read more
We have experience building interactive dashboards connected to databases and capable of handling large datasets. In one of our projects, we developed a dashboard for a store that enabled A/B testing between different recommendation algorithms. This allowed us to analyse and compare the performance of each algorithm in real time, helping the store optimally recommend products and maximise profit. The data we received included information about the products sold by the store, transactions, prices, and times of sale.
Approach
First, all store data was consolidated into a database, linking transactions, product information, and sales timestamps. Next, systematic A/B tests were conducted between algorithms (e.g., ItemKNN versus Popularity). Comparisons were made based on various metrics. Analyses were extended to patterns such as product combinations, seasonal trends, and algorithm efficiency over time. Results were presented on an interactive dashboard, allowing the store to easily gain insights into the best algorithms, top-selling products per period, and other relevant metrics.
Result
This project demonstrates how data analysis and AI recommendation systems can directly create value for businesses. The dashboard provides the store with clear insights into which strategies work best, when products are sold most, and how product combinations influence purchasing behavior. The result is a fully deployable system that supports real-time analyses and enables strategic optimization of sales processes.
Stack