Toss Makers Conference: Design Day Review

Creating the Worst Experience at Toss

This session shared Toss’s advertising UX improvement case study, focusing on how to transform ‘unavoidable features’ into better experiences.
Ads are an unpleasant element for most users, but from a service perspective, they can’t be easily removed.
The presenter shared their approach to creating not just ’non-intrusive experiences,’ but advertising UX that users can accept or even view positively.

What users hated most were two things:

1. Ads that appear suddenly
2. Ads that interrupt important information

Structures that pop up without warning or block user flow caused such aversion that users would close the app, and there were actual cases of related inquiries to customer service.

How was this solved?

Simply adding a phrase that notifies in advance that it’s an ad significantly reduced feedback, and when ads were moved to the periphery of information, click rates actually increased. Through small experiments and iterations, a baseline for acceptable UX emerged, and within those standards, a structure was created where business metrics didn’t collapse.
Next came attempts to transform into ‘good experiences’: The presenter says they moved beyond “being as unobtrusive as possible” to considering how to make users willingly view ads.

Three directions were introduced for this:

Necessary ads: Providing ads that match user context and situation
→ Building B2B advertising platform, securing ad diversity
Entertaining ads: Even without creating directly, building foundation through idea sessions
→ Later expanding to various formats like quiz-type, game-type
Rewarding ads: Experimenting with various reward forms and intensities
→ Diverse attempts including cash, gift cards, paid rewards

This design actually increased both usability and revenue without user churn, bringing both internal team perception changes about ad UX and external attention as a case study.

Building Products That Cut Work Time by 70%

Background

This session covered Toss Bank’s inheritance/fraud compensation processing.
These two tasks, rare experiences for most users, had such slow and complex structures that average processing time took over 2-4 weeks.

The problem started with documents that needed to be prepared in advance.
Users didn’t know what documents were needed, and staff had to explain everything over the phone. Submitted documents were equally analog.
When incorrect applications were added on top, users only accumulated frustration asking “Why is this taking so long?”

Improvement Direction

First, they changed the user application flow. (Unprocessable cases decreased by nearly 30%, and staff could focus on truly core tasks.)

Before	After
Existing large, expensive kiosks were burdensome	Kiosk function using terminal screen, handling over 50% of orders
Thought regulars were about half, but actual data showed 20%	Marketing consent rate reached 90%, intuitive visualization and automatic coupon features
Card payment was always ‘what the owner pulls’	Customers choose payment method themselves, added card company discount info and seasonal events

Next, they fixed the internal admin.
The existing admin had only minimal functions, with most processing done manually. The new direction was:

Change	Effect
Organize work by status	Clarify current to-dos in table format
Configure screens by stage	Clear what information to review
PDF viewer on left, controls on right	Review without window switching
Document verification checklist	Reduce omissions, increase review accuracy ↑
Link field information	Make it easy to judge “is this correct”

As a result of these changes, from the user’s perspective, the wait of “I applied but why no contact?” decreased, and for staff, an environment was created where they could immediately immerse themselves in work without complex manuals.

Research That Grew 10x Despite Rejection, Transforming Products

This session started not from ‘how to recruit interview subjects,’ but from ‘why business owners refuse interviews.’ The Toss Place team learned through research with self-employed business owners that B2B user research must be designed differently from the approach itself.

Contacts work, but interviews don’t?

There were many contacts, but most were unresponsive. Brand awareness was low, so many didn’t even know what ‘Toss Place’ was. It wasn’t a problem that could be solved simply by segmenting target filters or automation. Ultimately, what mattered wasn’t the approach method, but ’trust.’

So they redesigned from scratch

They set the goal to make the research experience itself a better product, experimenting with the most responsive time slots, channels, and tone & manner by industry. They send from official Toss accounts, clearly convey brand identity, and use the persuasion point “your story will actually be reflected” rather than emphasizing compensation.

Results?

Business owners who prepared five A4 pages of feedback in advance, users who filmed and sent actual in-store customer usage scenes, people who voluntarily participated in interviews 2, 3 times. Along with the message that B2B or offline customers aren’t persuaded by conventional user research methods alone, came the realization that what they need isn’t ‘research participation’ but ‘co-creation experience’ design.

Why Aren’t They Using It? Uncovering the Real Problem

In this session, a Toss Securities UX researcher explained in three stages what role researchers should play in product development. The core message was ’the more research becomes something anyone can do, the more professional researchers’ role should be asking fundamental questions.’

Initial Planning Stage – Problem Definition and Direction Setting
In the early product stage, even ‘what to build’ isn’t clear yet. At this point, researchers, like someone fitting puzzle edges, define the problem scope from the user perspective and set product direction. For example, when preparing an AI-based investment information service, users wanted contextually connected interpretation rather than simple news alerts, and expressed ‘information fatigue.’ Based on this, value elements like information search practicality, recognizability, and psychological stability were derived and incorporated into the product structure to establish initial direction.

Feature Improvement Stage – Looking at Roots, Not Just Surface
In situations where features exist but need improvement, rather than just polishing screens, whether users’ goals themselves are properly set must be re-examined first. A representative case was the ‘market calendar’ feature. Initially it just listed schedules, but users actually wanted to gain confidence and prepare for investment decisions through this feature. So the state of ‘using well’ was redefined as stages of recognition–understanding–preparation, and the design structure was changed accordingly.

Stagnant Growth Stage – Questioning the Product’s Place Again
When a product doesn’t grow for a while, researchers ask not just numbers but why this product should exist on this platform. Toss Securities’ PC web platform was representative. Functions were sufficiently implemented, but users didn’t feel a reason to use it on PC. At this point, research approached by asking ‘is this really necessary’ rather than ‘why aren’t they using it,’ and ultimately made the team rethink what environment and conditions are needed for the product to grow.

A presentation that re-reminded that researchers’ role is ‘the perspective of asking questions and how to frame problems within the team.’

How Toss Develops Interactions

Toss Interaction Team’s Collaboration Methods and Tool Design Story

This session introduced how Toss’s interaction team is refining workflow for designers and developers to implement motion quickly and accurately together. The points are ‘creating a language for communication’ and ‘designing tools directly’.

Rally’s Core Concepts

Rally: A structural language that allows designers and developers to express and understand motion in the same way

Pseudocode syntax that allows writing motion like code
Unified standard terms like playCount without ambiguity like “how many times does this repeat?”
Designed to maintain the same meaning across different environments like Sketch, After Effects, iOS, Android
Automated to actual code

Rally Main Systems

Rally Editor: Allows designers to easily write pseudocode
CodeGen System: Automation tool that converts to platform-specific code
Rally Viewer: Visualizes implemented motion and documentation for developers to compare at a glance

-> Designers can give feedback faster, and developers can reduce implementation mistakes.

Creating an Environment Without Interaction Designers

Toss’s proprietary design tool Deus’s features

Like Figma, motion can be specified with just keyboard shortcuts within the interface
Work while maintaining designer’s workflow without switching tools
Developers verify all parameters in Deus Inspector and reflect in code

Thanks to this, a structure became possible where interaction level is high but production difficulty is low.

Summary

Structured language is more important than words → Unified motion collaboration standards with Rally as common language
Directly created tools to solve problems → Designed collaboration tools centered on actual use like Rally Editor, CodeGen, Deus
Motion is implemented not by intuition, but by agreed specifications → Can implement quickly and accurately without rework between teams

Bonds: Unfamiliar Yet Familiar

This session addressed how to make ‘first-time financial products’ intuitively and easily understandable. The presenter shared trials and learnings experienced while designing bond services at Toss Securities.

Starting with ‘Simplicity’ Principle

Toss Securities always takes simplicity as a core strategy. Bonds were the same - the goal was to intuitively show only “how much you put in and how much you get” without complex concepts.

→ So they designed to attach a calculator to show expected returns at a glance. But user reactions differed from expectations. “Why is this graph growing like this?”, “How much do I actually receive?” etc. Even when information was connected, showing multiple at once made understanding difficult.

“Resonating” Over “Explaining”

Even when complex concepts were kindly explained with tooltips, they still felt unfamiliar and difficult to users.
Ultimately, the core wasn’t ‘amount of information’ but ‘resonance.’ It should feel intuitive without trying to understand to be ’truly simple’ UX.
So the presenter redesigned bonds by comparing them to ‘deposits,’ a familiar concept to users.
When they clearly emphasized “receive principal + interest after 2 months” and created a structure where you can see at a glance how much you put in and receive, user reactions changed. They understood with familiar thinking like “Oh, this is like a deposit?”

When they conducted A/B testing with the above UI improvement and the original, new user purchase conversion rate increased by 21%. The key point was not changing functions or reducing information, but changing to ‘a framework easy for users to understand.’

Three Years of Flipping Offline POS with UX

A session impressive for its detailed experiments to the point of “they even do this?” Shared how Toss team overhauled POS systems used by offline self-employed business owners, going into actual store sites to observe and gather feedback while refining UX.

Making Kiosks with Small Screens

The first experiment was turning existing POS terminals into ‘small kiosks.’ In situations where store operators had to handle ordering, serving, and payment alone, existing large and expensive kiosks were burdensome. So they added kiosk functionality using terminal screens and launched beta in 2 weeks for testing. By repeatedly experimenting with how users view screens and what UI configurations are efficient, usage rate rose to handle over 50% of actual orders.

Solving Customer Management with UX Too

Business owners thought ‘regulars are about half’ but actual data showed 20%. To raise marketing consent rates, they naturally induced phone number entry and accumulation before payment, and consent rate reached an impressive 90%. Based on customer data, they also introduced intuitive visualization and automatic coupon features. Simplified UX to a level anyone can use, reducing “this is complicated” feedback.

User-Led Payment

They changed the flow where card payment was always ‘what the owner pulls’ as natural. They designed the interface for customers to choose payment methods themselves, adding card company discount information and seasonal events (Santa appearance, etc.) to capture both experience fun and utility. Events with cuteness and interaction brought unexpected viral effects.

Business owners actually said things like “Now I don’t have to compare with the god of POS systems,” and internally the team gained confidence that “more natural experiences beat habits.” It was a case that stood out not for simply making MVP functions, but for meticulous site-based design to make truly beloved products (MLP).

POS Improvement Case Comparison

Before	After
Existing large, expensive kiosks were burdensome	Kiosk function using terminal screen, handling over 50% of orders
Thought regulars were about half, but actual data showed 20%	Marketing consent rate reached 90%, intuitive visualization and automatic coupon features
Card payment was always ‘what the owner pulls’	Customers choose payment method themselves, added card company discount info and seasonal events

Table of Contents