Background

Marketplace users face two major problems: too many offers and too few relevant products. Customization and personalization should help, but they often fail - either being too intrusive or offering no real value.


The goal of the project was to rethink personalization mechanics to:

• Adapt the interface to each user.

• Make the shopping process more convenient and faster.

• Increase conversion and retention by improving the overall experience.

Research and Key Hypotheses

Our research showed that:

• Users don’t notice customization options or don’t understand how they work.

• Customization feels complicated and unnecessary if there’s no clear explanation of its benefits.

• Users want to see tangible benefits immediately, not just a promise of a "better experience."

• Different product categories require different customization approaches - choosing pet food and selecting clothes are entirely different scenarios.

Solution: Dynamic Personalization and Modular Customization

We developed a flexible personalization system that adapts to user behavior instead of requiring manual adjustments.

What we implemented:

• Customization widget: unobtrusive yet accessible on the main screen.

• Interest selection during onboarding (similar to Spotify) to improve recommendations from day one.

• Customizable sections: users can prioritize the product categories they care about most.

• Personalized promotions and discounts: all offers are gathered in one place to show real savings.

• Complementary product recommendations: instead of just suggesting similar products, users see items that logically pair well with their chosen product.

What Testing Revealed

The initial testing iterations provided unexpected insights:

• Half of the users didn’t understand why customization was needed.

• The personalization questionnaire was too long, causing many users to abandon it halfway through.

• Users expect immediate benefits - customization is better received when it instantly highlights discounts or improved recommendations.

After refining the system:

• The customization process was shortened and made clearer, leading to higher completion rates.

• We added an introductory screen explaining the benefits before starting.

• Customization was first tested in the clothing section, where it proved to be particularly useful.

What did we do based on it

Redesigned the personalization questionnaire to be shorter and more engaging, reducing drop-off rates.


• Introduced an onboarding explainer screen to clarify the benefits of customization before users interact with it.


• Expanded A/B testing in different product categories to tailor personalization strategies for different shopping behaviors (e.g., clothing vs. electronics).


• Refined recommendation algorithms to prioritize frequently purchased or highly rated items, making suggestions more relevant.

• Implemented a dynamic customization widget that adapts in real-time based on user preferences and past interactions.

• Developed contextual tooltips and guidance (Coach Marks) to ensure users understand how to leverage personalization features effectively.