The problem

Sellers were losing momentum at critical moments

Mercari’s original “Offer-to-Likers” feature allowed sellers to promote their listings by sending a discounted offer to people who liked an item. However, the system had two rigid constraints:

• Offers could only be sent once every 24 hours

• Each offer required a minimum 10% price drop

These limitations created a frustrating scenario: sellers couldn’t reach newly interested buyers who liked an item after the original offer had been sent.

New offer logic

To support sellers in reaching new likers efficiently, I worked with PM, content designer to brainstorm different ways to reimagine offer logic. From our workshop, we co-created this new offer logic using a Parent Offer model:

• An offer could now automatically extend to new likers within a 24-hour window

• Sellers no longer needed to lower prices again to reach new interest

• Only when choosing to re-engage all likers would a price drop be required

This logic aligned seller needs with platform goals - more reach, less friction.

My initial approach was to insert a clear “fork in the road” inside the offer modal: Extend the current offer to new likers or create a new offer for all likers.

At a system-logic level, this made sense. However, placing this “fork in the road” inside the offer modal created unnecessary friction. By this point, sellers had already chosen to promote their item - introducing another strategic decision felt out of place.

Reframe the problem

I quickly gather feedback internally and with

• New likers appear continuously

• Sellers rarely want to segment who does and doesn’t get an offer

• The system - not the seller - should handle timing and delivery

This reframed the problem: Sellers don’t want to manually manage branching logic. They want the platform to take care of it, as long as the behaviour is predictable and transparent.

Once I realized the “fork in the road” shouldn’t be a manual decision, the next question became:

How much control should sellers have over an automated system?

I explored three design variants that gave sellers varying levels of control over auto-sending offers to new likers:

• V1: Toggle

• V2: Opt-out unchecked

• V3: Opt-in checked

• V4: No opt-out control

  To validate these directions, I partnered with UXR to run a quick usability sessions with 3 casual sellers and 3 power sellers.

Here’s what we found:

• ✅ All 6 participants understood the concept of auto-send offers and saw it as helpful

• ✅ Power sellers strongly preferred default automation, wanting speed and efficiency

• ✅ Casual sellers wanted to feel more in control, but none opted out when presented with V3

• ❌ Toggle and checkbox make offer progress slower for power sellers

• ❌ With the checkbox, casual sellers felt unsure if they’d miss opportunities

After synthesizing usability feedback, I aligned with PM and Eng to evaluate whether offering seller-facing controls was still necessary. Together, we weighed the trade-offs across transparency, technical complexity, and long0term system behavior. We aligned that maintaining automation would create a more predictable experience and reduce repeated manual steps for sellers.

This led us to move forward with V4 (no opt-out control) and focus instead on educating sellers up front about how the logic works through contextual onboarding and in-product messaging.

Because sellers would no longer configure auto-send themselves, transparency became important in the sellers journey.

Instead of adding persistent UI controls or complex settings, I designed a lightweight, contextual onboarding flow:

Solution: Contextual onboarding flow

• One-time modal for returning sellers, clearly explaining how offers now reach new likers

• Inline cues on offer UI to reinforce logic subtly over time

• “Learn more” link for deeper understanding without blocking actions