market share

Recommendation engine for the biggest price aggregator in SE Asia

Set Up

  • Online price aggregator
  • Importance of superior recommendation engine
  • Business goals: click-through rate and customer engagement
  • Product catalog with 500 mil items
  • Singapore, Malaysia, Indonesia, Philippines, Thailand, Vietnam and Hong Kong

Business Task

500 million items

With more than 500 million items in the product catalog, it became critical for our client to display the most relevant content first. We were asked to determine which items should be displayed on various landing pages in order to maximize customer engagement and the click-through rate.


90% new customers

Challenges we faced were given by the nature of the project. We could not use traditional ranking approaches since 90% of customers were first time users with limited information. We also did not have data about shown-but-not-clicked products. Moreover, presentation and position bias naturally occur in similar tasks. All of us have experienced that when searching on Google, we usually do not go above certain number of pages (presentation bias) and that we usually pay more attention to links listed in the beginning (position bias).

Business impact

20% increase in click-through rate

Our approach & Solution

Products' popularity prediction

We have developed state-of-the-art machine learning model to predict products’ popularity using extreme gradient boosting algorithm. We have also set up robust and scalable data infrastructure which allows to process the required volumes daily overnight. Our computation layer includes Spark and Python and we can directly observe the model performance via Tableau reports. According to the A/B test, we have achieved 20% average uplift in click-through rate.


  • Robust and scalable data architecture
  • State-of-the-art machine learning model
  • Daily updates and effective reporting framework
  • Python, Spark, Tableau

Contact us for more information about our work

Send a message

We Support

  • Sponsoring the education of those with less fortunate backgrounds through the CEDA project in Uganda
  • Lecturing through the Credit Risk in Banking course at Charles University in Prague
  • Supervising university students in their academic research

Contact us

Phone: +420 724 224 127 (Europe), +65 9052 1518 (Singapore)
WhatsApp: +420 607 209 443

Would you like to speak with us?

Let us know
  • Taran Advisory PTE. LTD., RN: 201902470R, 9 Raffles Place, Republic Plaza, #06-01, Singapore 048619
  • Taran Advisory s.r.o., IČ: 06935770, DIČ: CZ06935770, Na Perštýně 342/1, Staré Město, 110 00 Praha 1, sp.zn. C 291639 vedená u Městského soudu v Praze.