← Work
Chotot · Carousell Group · Senior Product Designer

Helping marketplace sellers understand what they're paying for

Company
Chotot · Carousell Group
Timeline
6 months · Jun-Nov 2023
Scope
Product Design • Data Analysis • AB Testing
Chotot POS checkout redesign
Overview
In 2023, Chotot set out to grow its ad-listing revenue - specifically through the point-of-sale services sellers buy at checkout.

Chotot is Vietnam's marketplace for second-hand goods, vehicles, and real estate. I'm a lead designer at Nhatot, its real-estate arm, where sellers pay listing fees and paid boosts to get their ads seen.

Problem
After the Lunar New Year spike in early 2022, sellers' ad spend fell steadily all year.

Premium features dropped worst of all - down nearly 58% - and the slide was still running when we pulled the report.

Month-on-month revenue decline across order types
Objective
Lift checkout conversion on listing fees and premium features for real estate.

Three metrics to move:

  • Conversion from service selection → payment
  • New paid users (2nd listing onward - the first is free)
  • Services per order
Solution
Unified by design, tailored by journey.
  • Streamlined the service cards to cut confusion and help sellers grasp a service's value before acting.
  • Segmented the POS experience by journey, fully separating Listing Fee from Premium Features.
  • Promoted overall usage by recommending POS Bundles for a compounded effect at a cheaper price.
Final POS design 1Final POS design 2Final POS design 3
Audit
Too many things at once.
  • Relevancy - too many sections that don't support the seller's need; sellers just want their ads online fast.
  • Usability - multiple layouts hinder seller expectations, losing confidence to complete their actions.
  • Value - few descriptions of how services help, and no scannable visuals to translate their value.
The old POS: listing fees and every premium feature on one page
Research
Sellers hadn't stopped spending - they wanted to spend smarter.

I interviewed 8 sellers - licensed brokers and private sellers. Three findings stood out:

Finding 1 - provided budget was restricted
Finding 2 - I only use Single Bump
Finding 3 - cautious about my listings

The data backed it up: 5%+ of orders already bundled multiple services, and 31% of listing-fee orders also bought Bump - pointing to an opening: upsell bundles right after listing, with Bump as the hook.

Bundle-services breakdown: share of orders by service combination
Redesign
Design for comprehension over promotion.

The core insight: sellers couldn't make sense of some services. So every service card now carries a name, a plain description, a price/promo, and a visual cue.

The four essentials of a service card

Iterating on service selection.

Iteration 1
Step-by-step selection tested badly - sellers didn't connect the two sections.
Iteration 2
Merging duration into the card (with a checkbox) fixed grouping, but hid price comparisons.
Iteration 3
Final: bring offers back, add a 'Save up to' label. A/B test: +8.93% conversion (28.86% → 37.79%).

Splitting the flow to cut overload - listing fee at posting, premium features in the dashboard:

POS - Listing Fee flow
POS - Premium Feature flow
After launch
Conversion jumped - but we'd been accidentally upselling listing features.

A month in, listing-fee conversion rose (39.6% → 52.9%), yet some premium features dipped. Investigating, we found two issues to address quickly:

  • We'd accidentally been upselling premium features. With the old POS laying everything out at once, a fair amount of revenue came from sellers mistakenly adding services to their order - they assumed inserting the ad wouldn't complete until they'd bought something.
  • Some features needed more than a text explanation. Sellers hesitated on complicated services, doubting the presumed benefit from the description alone.
Premium-feature upsell in the listing-fee flow
We added a premium-feature upsell to the POS-LF flow, using the post-launch lift in listing-fee conversion to promote more PF usage.Visually it's a supplement to the listing fee - images and descriptions stripped out to keep it as unobtrusive as possible.
Animated preview in the more-info popup
In POS-PF, we added an animated preview inside the more-info popup for sellers who still found the text hard to comprehend.For a continuous experience, changes made in the popup carry through to the final order.
Outcome
A significant 31% increase in CR within six months of full release.
Conversion growth after launch
  • Reached a 4.9% uplift in ads with bundles. In Phase 2 we shifted seller behaviour toward bundles by introducing more cost-effective bundles with compounded effects.
  • Expanded the POS treatment to two more categories - Goods and Vehicles - on the strength of a proven record.
Reflection
Two things I'd carry forward.
  • Design for economy of scale. With many sub-brands, solutions for key products have to flex across use cases - doubters thought the revamp wouldn't transfer to other categories; it did.
  • Owning a product line is a documentation marathon. Extending the POS to each category meant reworking offerings and edge cases, and making sure new requirements never broke live mechanics.