Intro
Pantauan Barang (Product Monitoring) helps sellers improve product performance by providing strategic recommendations to increase sales and reduce refunds. As part of Bukalapak’s analytics suite, it enables sellers to monitor key product metrics and grow their business effectively. For the MVP, the feature targets the top 10 products with declining sales or low stock. Success is measured by the number of users clicking on the recommendations.
Understanding why sales drops & how to fix it
Sometimes sellers (Pelapak) experience a drop in sales without understanding the cause. They often respond by blindly bidding on ads, changing product images, or offering discounts. However, without identifying the root issue, these efforts don’t effectively boost sales or improve product visibility.
Refunded transactions are also a significant problem. Currently, about 7.39% of Bukalapak Marketplace transactions are refunded, resulting in an estimated 68 billion Rupiah monthly loss in GMV. Nearly half (48.8%) of these refunds are due to seller-related issues, with 69.11% caused by reasons like “store closed,” “no stock,” and “old price.” To address this, we aim to help sellers monitor their product stock and prevent these avoidable refunds.
Learning from our competitors
For our initial research, we conducted a competitor analysis to evaluate the strengths and weaknesses of our Pantauan Barang feature compared to direct competitors Shopee and Tokopedia. While Shopee offers a similar product monitoring feature, Tokopedia currently does not—though we identified several useful insights from their platform for future improvements.
Key findings from the analysis include:
Shopee provides detailed, in-depth analytics, but the interface can be difficult to understand. However, as a market leader, many sellers are already familiar with it.
Shopee’s filter and search functions make it easy for sellers to find specific products and assess their performance.
All recommendations must be actionable to truly support sellers in improving sales.
Design ideation
To generate ideas and tackle potential challenges, I facilitated a design ideation session with the Analytics and Management squad, including UX researchers, UX writers, project managers, developers, and data scientists. During the session, we voted on the best ideas and assessed their feasibility based on project scope, timeline, and development effort. From this ideation, we uncovered key insights:
Evaluate performance not only at the product level but also at the store level. Educating sellers about the product funnel is crucial because issues in the upper funnel can limit the impact of product-level recommendations.
Provide actionable, personalized recommendations. Once sellers understand what’s wrong, straightforward guidance helps them improve their products effectively.
Minimize clicks. Displaying action items directly on the screen without extra steps is essential, as research shows more steps reduce user engagement and click-through rates.
Ideation with the Analytics and Management squad
Personalized recommendation using decision tree
Collaborating with the data and product teams, we developed a decision tree to personalize product-level recommendations. Each branch of the tree addresses a specific issue, with threshold values guiding the recommendations:
Low product score: Improve product details through the Product Detail Score page.
Low product impressions: Use Promoted Push or Promoted Keyword campaigns to boost visibility.
Low click-through rate to Product Detail Page: Enhance product details via the Product Detail Score page.
Low conversion from Add to Cart to Checkout: Send targeted promotions to buyers who visited the product or store using Bukalapak’s CRM.
Low conversion in Checkout: Remind buyers who added products to their cart but didn’t complete checkout through CRM broadcasts.
Low product stock: Send reminders to update stock levels.
decision tree that will personalize the recommendation on a product level
Design fast iterate fast
Before creating high-fidelity designs, I started with wireframes to quickly focus on the user experience without investing too much time. These wireframes were later used for concept testing.
The core idea of this feature is to highlight products with declining sales between two periods and provide personalized recommendations based on each product’s issues. The main challenge was making this information easy for sellers to understand. Our success metric was the number of users who clicked on the recommendations.
After testing the concept with four users, our UX researcher gathered key insights:
Terms like “Tingkatkan Performa” and “Reco” were confusing to users.
Sellers have different definitions of low conversion rate; some consider 3% high, others see 6% as low.
Low stock thresholds vary by product category — for example, 50 units is low for food but not for electronics.
Wireframes made for concept testing
Iterating Toward a Usable MVP
Based on insights from concept testing, I iterated the design while moving from low-fidelity to high-fidelity. Collaboration with developers also uncovered technical constraints that required further adjustments:
Product details with statistics won’t be included in the MVP due to development effort; instead, we provide a link to the existing Product Statistics page.
The Sales Funnel (Informasi Penjualan graph) will be prioritized for future improvements, as it requires significant backend work.
The recommendation flow was simplified from 2 clicks to 1 click to improve usability.
We integrated Pantauan Barang with existing features by adding an entry point on the Daftar Barang (Product List) page.
In the final design for development, there are several important changes. First, seller can get personalized product recommendation for each product in just one click. Second, there is no product detail anymore due to development timeline.
Final design