Home

Phase 01

Building (pre)MVP

Phase 02

From (pre)MVP to MVP

Final Design

Result

From 0 → 1

From 0 → 1

How I Validated Before Building:

A Complete Journey of Reinventing VAT Filing in 7 Months

How I Validated Before Building:

A Complete Journey of Reinventing VAT Filing in 7 Months

Project

Overview

Project

Overview

Most business owners file VAT only once or twice a year, and the official government tool is complex enough that first-timers really struggle with it. We saw an opportunity to make the whole thing far simpler, but the risk was real. VAT season only opens one launch window a year, so if we built the full product and got it wrong, we'd lose months with nothing to show for it. Rather than guess, I started by building a lightweight estimation tool, just enough to confirm whether people genuinely wanted this. They did.

That validation set up the harder challenge: two months to tax season, a single designer, and a product that had to handle 12 different filing situations, since people track their sales and expenses in so many different ways. To pull it off in time, I kept everything as simple as possible, building each feature in its most basic form and testing constantly with real users. Easy filings ran automatically, while complex ones were routed to real tax accountants. What follows is the full story of how it came together: navigating constant ambiguity to build the first mobile VAT filing service for SMB owners.

Most business owners file VAT only once or twice a year, and the official government tool is complex enough that first-timers really struggle with it. We saw an opportunity to make the whole thing far simpler, but the risk was real. VAT season only opens one launch window a year, so if we built the full product and got it wrong, we'd lose months with nothing to show for it. Rather than guess, I started by building a lightweight estimation tool, just enough to confirm whether people genuinely wanted this. They did.

That validation set up the harder challenge: two months to tax season, a single designer, and a product that had to handle 12 different filing situations, since people track their sales and expenses in so many different ways. To pull it off in time, I kept everything as simple as possible, building each feature in its most basic form and testing constantly with real users. Easy filings ran automatically, while complex ones were routed to real tax accountants. What follows is the full story of how it came together: navigating constant ambiguity to build the first mobile VAT filing service for SMB owners.

Role

Role

Sole product designer

Sole product designer

Led product design from early validation to MVP launch

Led product design from early validation to MVP launch

Team

Team

Product Managers

Product Managers

3

3

Tax Planners

Tax Planners

2

2

Marketer

Marketer

1

1

Front-end Developers

Front-end Developers

3

3

Back-end developers

Back-end developers

3

3

CS manager

CS manager

1

1

Timeline

Timeline

2022 July - 2023 January (7 months)

2022 July - 2023 January (7 months)

Key Skills Applied

Key Skills Applied

Product & Business Strategy

Product & Business Strategy

Led a design strategy balancing regulation, trust, and business viability

Led a design strategy balancing regulation, trust, and business viability

Lean UX & Rapid Testing

Lean UX & Rapid Testing

Ran hypothesis-driven experiments to confirm product-market signal

Ran hypothesis-driven experiments to confirm product-market signal

Project Management

Project Management

Shipped 3 complex products at once under tight timelines

Shipped 3 complex products at once under tight timelines

Impact

Impact

Validated Product-Market Fit

Validated Product-Market Fit

63,000

63,000

registered within one month of launch

registered within one month of launch

High user satisfaction

High user satisfaction

72.2

72.2

Net Promoter Score (industry avg: ~60)

Net Promoter Score (industry avg: ~60)

Background

3.3 (Three Point Three), South Korea’s leading

tax-tech platform with over 25 million users, planned to penetrate the growing SMB business market by offering a comprehensive tax solution for business owner by launching a VAT tax

filing service.

25mil+

user base

Blue Ocean Market

6.9m+

5.6m

2023

2018

Number of Small Businesses

Even before we built anything,

several challenges made this project

genuinely risky

Challenge 01

A basic MVP needs a complex tax system

VAT filing demands heavy upfront investment in integrations, tools, and support

Challenge 02

Product-market fit was unproven

Despite real pain points, owners could always fall back on the free government service

Challenge 03

User personas were highly fragmented

SMB owners are so diverse it was unclear how to deliver value for everyone

Challenge 04

Iteration was limited to tax season

We could only gather real feedback once or twice a year, creating a one-shot validation window

Strategy

Build fast, fail (learn) fast

through (pre)MVP

With high uncertainty and little room to iterate, I skipped long upfront research and validated our riskiest assumptions through a fast (pre)MVP, buying us one more iteration before release

2 weeks

2.5 months

3 months

Scale to MVP development

Research

(pre)MVP build & test

Early iteration with (pre)MVP

1 month

MVP release

Challenge

At its core, this was a

one-shot, high-cost launch

A failed launch would result in months of sunk cost with no reliable signal.
Reducing risk, therefore, became the core design challenge. How might we reduce risk in a one-shot, high-cost launch?

2 months

4 months

1 month

MVP release

Design & development

Research

High cost MVP, and no chance to test

So we split the project into two phases:

Phase 01

Building (pre)MVP to gather early results

Phase 02

MVP development for the real launch

Everyone had a different picture of our target user,

so I aligned the team with proto personas.

Different types of user confused the team and it became clear everyone had a different picture of our target user and what we should build. With limited time, I skipped heavy persona research and used proto personas instead, a fast, hypothesis-driven way to align the team on who we were designing for. It helped us quickly agree on our target users and turn early signals into concrete problem statements.

Early Research Signals

Building (pre)MVP

1

From research, I realized SMB owners were

incredibly diverse. One solution wouldn't fit them all.

Online and offline businesses approach VAT differently

To see how different SMB owner groups approach VAT filing,

we ran a quick survey (n=56) to spot key behavioral patterns.

S1

Finding 1

Online businesses report lower annual sales than offline ones

Annual sales comparison:

online vs offline business

online

offline

< 30M KRW

43%

90%

> 30M KRW

57%

10%

S2

Finding 2

Offline businesses lean on tax accountants; online ones lean on Hometax

Tax filing method preferences

online

offline

37%

72%

Used tax accountant

31%

69%

Used Hometax (gov site)

S3

Finding 3

Offline owners are generally older than online ones


Age comparison: online vs offline business owners

online

offline

38%

15%

20-30yr. old

85%

62%

> 30yr. old

User interviews revealed diverse needs and pain points

We ran 14 phone interviews in one week. A synthesis workshop surfaced five key insights about our potential users.

PHASE 01

Building (pre)MVP to test.

The (pre)MVP lets us learn what matters most to users, and where the real value is, before shipping an MVP

Building (pre)MVP

We had too many assumptions, so I prioritized what to build and how to measure success.

2

With many shaky assumptions and little time before tax season, we needed a structured way to decide what not to build. I led a hypothesis-driven workshop to surface the product's key assumptions, then ranked them by risk and user value using Jeff Gothelf's Hypothesis Prioritization Canvas. For the top hypotheses, we set success metrics and validation methods, then turned them into the minimum features needed to test them in the (pre)MVP.

1

Making Hypotheses

Step 01

2

Prioritizing Hypotheses

Step 02

3

Defining Requirements

Step 03

Prioritizing Hypotheses

Which assumptions carry the most risk and value, and are worth testing first?

H1

H2

Ship & measure

H5

Don’t test & build later

H3

Discard

H4

H6

Test

High risk

Low risk

High value

Low value

Building (pre)MVP

3

We had to know if demand was real before building, so I turned the (pre)MVP into a market experiment.

Instead of launching a full filing system right away, I reframed the (pre)MVP as a market experiment to test our riskiest assumptions before the full build. I proposed turning it into a VAT estimation tool, so SMB owners could quickly check what they might owe based on their sales and expenses. That gave them immediate value while we validated key assumptions. I defined the minimum requirements to test the hypotheses and designed the (pre)MVP around them.

churn rate

CTR

Sign up rate

CTR

CTR

CTR

Sharing rate

(pre)MVP Experiment: VAT Estimation Tool

Experiment Result 01

Hypothesis 1 & 2

confirmed real market need for a mobile VAT reporting solution.

Implication

These results gave us the confidence to move into MVP development and invest in richer features.

H1

Hypothesis 1

Mobile service for complex sales/expenses drives engagement

Strong Demand

of SMB users want to use mobile service to file

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

landing page conversion rate

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

H2

Hypothesis 2

Satisfied users will sign up for early access and share

Early Traction

of the users apply to have early access to the product

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

avg. early signup rate

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

Experiment Result 02

Two early assumptions turned out to be wrong.

Users were happy to sync their data, and strongly preferred manual input.

Implication

These findings gave us confidence to support manual input and deeper data syncing, expanding

the MVP to serve

more SMB users.

H4

Hypothesis 4

Web users will avoid syncing their business data due to privacy issues

Low Privacy Concern

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

of users successfully sync their business data through simplified certification process

H6

Hypothesis 6

Manual sales input won’t boost usage in target users

Manual Input Preferred

of users want to input their sales data

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

of users manually input their sales as miscellaneous sales last year

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

Experiment Result 03

Key drop-offs in

the funnel showed us where to optimize.

Implication

These findings let us prioritize conversion optimization over other MVP work and run A/B tests on the critical drop-off points.

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

PHASE 02

From (pre)MVP to MVP.

MVP

Technical Feasibility

UX
Feasibility

Business
Feasibility

How do we cover diverse tax filing scenarios?

How do we convert users into revenue?

How do we help users understand and trust this unfamiliar service?

The (pre)MVP proved demand, but building a real product brought new complexity. With just two months before tax season, the MVP had to balance technical, UX, and business feasibility. I led the design decisions across all three to shape it.

With demand validated, the challenge shifted to building a viable product.

User Side | Mobile Service

Growing the (pre)MVP into
a real filing service in 8 weeks.

Weeks 1–2

Sales & expense input (12 methods)

Weeks 3-4

Flow iteration to support tax logic

Weeks 5-6

Payment & filing confirmation

Weeks 7-8

QA, polish, edge cases

The next challenge was fitting complex VAT filing features, including 12 filing methods, into an 8-week build window. As the sole designer,

I reduced each feature to its leanest workable form, keeping only the essential actions required for standard VAT filing.

(pre)MVP flow

Added MVP flow

Figma developers handoff compilation file

Internal tool

To process thousands of filings in weeks, attorneys, support,
and users had to be in sync. I designed that ecosystem.

User

Customer support

Users input sales and expenses to calculate their VAT

Payment and filling out personal info details

Waiting for final review

Submit additional docs when requested

Waiting for VAT filing submission

Tax Attorney

Review the tax case

Request additional information

Filing the tax case to the government

Supporting users & tax attorneys in every step and answering their questions

Main mobile service

Consulting service with tax attorney

Tax reviewing tool

Admin tool

Main mobile & consulting service

Direct consultation channel connecting users with tax attorneys for case-specific guidance

Tax reviewing tool

Internal workspace for tax attorneys to review cases, request documents, and finalize tax filings

Admin tool

Admin console for customer support to track cases, manage inquiries, and support all stakeholders

1/3

Technical Feasibility

Building The Tax

Review System

We couldn't automate every case, so I designed a hybrid system

where tax attorneys took the complex ones.

With demand validated by the (pre)MVP, the next challenge was enabling real VAT filing. But filing is full of complex edge cases and legal risk, hard to build in a 2-month timeline. With the PM, I weighed several infrastructure approaches to find the most viable one for the MVP.

OUR PICK

Covers a wide range of user cases with relatively lower development cost

High operational cost and requires building internal tools and workflows

Option 02

Hybrid model

Combine automation with human support, routing complex cases

to tax attorneys.

Reason

Mixing automation with human support let us test demand across the whole market, including complex cases, without hitting technical limits

Lower risk and cost, allowing us to focus on product quality

Excludes a large portion of potential users (around 70% of the market)

Option 01

Limited scope

Support only the users

the (pre)MVP already covers,

while we polish the UX.

Dealbreaker

An MVP should test real demand,

not polish UX

Most scalable and could significantly reduce operational cost

High risk of errors, leading to legal consequences or loss of user trust

Option 03

Full automation

Fully automate every flow so

even edge cases are covered

within the 2-month timeline.

Dealbreaker

Full automation is too complex now

and risks incorrect filings that

could break user trust

2/3

UX Feasibility

Refining the UX Through Constant Testing

Reinventing how people file taxes risked confusing users,

so I tested constantly to keep it intuitive.

Since the service introduced a new self-reporting workflow for SMB owners, usability risk was high. I ran iterative usability testing throughout development to keep the product aligned with users' mental models and intuitive as it grew more complex.

Development

Iteration

Tax season

Pilot UT

UT 1

UT 2

Pre-development UT

Check the flow fits users' mental models and expectations

Using Figma Prototype

3 Users

Post-Development UT

Find and fix critical usability issues before launch

Using STG Environment

2 Users

After Release UT

Learn how real users use the product during tax season

Using MVP in Production

1 User

UT recording for analysis

Usability testing script

Usability testing session

Usability Testing Results

Testing confirmed the flow was intuitive, caught issues

before launch, and revealed users need control to feel confident.

Read More

01

Pre-development UT

The flow is intuitive and matches users' mental models with no trouble

02

After release UT

Users may prefer transparent, verifiable results over simplified automation, to keep confidence and control

03

Post-development UT

Found 5 key usability issues

3/3

Business Feasibility

Finding What Users

Truly Valued

Every drop-off was a question about what users valued, so

I tested each one to prove the business was feasible.

In the (pre)MVP funnel, users dropped off heavily before the paywall. Since the MVP's goal was to measure real demand, conversion was the most critical signal. I focused on cutting friction at key decision points and ran targeted A/B tests to learn what users valued most.

1

Calculation Result

In the (pre)MVP, users dropped off sharply right after the calculation result. Many checked their estimated tax but never reviewed sales and expense details, sparking debate over whether they preferred speed to detailed verification. To settle it, I proposed a flow preference test.

2

Paywall

Since most users drop off here, I proposed A/B tests to find which value propositions cut hesitation and convert better, aligning our messaging with what users value and guiding future direction.

Flow Preference Test | Calculation Result

Do Users Want Speed or Control When Filing Taxes?

Read More

Flow 1

Conversion Rate

26.6%

Flow 2

Conversion Rate

41.1%

Winner

Drop-offs at the calculation result sparked a debate: should users review sales and expenses step-by-step for accuracy, or skip to the final result for speed? To answer it, I tested a detailed review flow against an express filing flow. Express won by a wide margin (41.1% vs 26.6%), showing most users valued speed over verification.


So we made express filing the default

and added a confirmation step

to keep users in control.

Value Proposition A/B Testing | Paywall

What Drives Users to Pay?

Read more

A

Conversion Rate

58.6%

B

Conversion Rate

63.6%

Winner

At the paywall, the key question became: what value proposition would motivate SMB users to proceed with payment? To answer this, I proposed an A/B test comparing expert review before submission with pricing tailored

for small businesses.


A boosted confidence in a small, high-revenue segment, but B won on overall conversion (63.6% vs. 58.6%), confirming that fair pricing was the main driver for our core small medium business owner users.

1

Comprehensive Sales and Expense Review

To prevent mistakes, users can check that their sales and expense data is complete before submitting their VAT report.

Expenses details page

Sales details page

2

Guided
Manual Input

Surveys and usability testing showed most SMB owners knew little about VAT filing, so we added a step-by-step input flow to guide first-timers and cut reporting errors.

SMB

VAT

Quick access to detailed sales and expense breakdown for meticulous users

Because users fear filing mistakes, the full VAT calculation is exposed for transparency so they can verify the result with confidence

Sales

₩25,000,000

Expenses

₩17,000,000

VAT season starts July 1!

Final Design & Main Features

Introducing Simplified VAT Filing Service for SMB Owners

The first mobile VAT filing service for SMB owners of every type, from meticulous reviewers to busy operators, for fast, accurate filing.

3

Bringing Web-Level Sales Management to Mobile

One of the 12 filing methods, giving users web-level control on mobile by syncing tax system data while still allowing edits and additions.

Adding the details

Selecting business partner

Hometax data synced

Usability testing showed many users don't finish filing in one session. To fit different speeds and workflows, they can resume where they left off, with the home screen guiding their next step.

File at your own pace.

4

Key Results

The launch proved our early validation true:

promising product-market fit.

(pre)MVP

Conversion rate

MVP

Conversion rate

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
%

48.6%

40.7%

38.9%

19.4%

21.7%

12%

7.78%

4.64%

Scraping

Estimation
Page

Result
Page

Payment
page (MVP)

Application page (preMVP)

Funnel conversion rate

MVP

(pre)MVP

Key screens conversion rate

(pre)MVP

MVP

Scraping

66.2%

73.8%

Estimation Page

61.6%

55.9%

Result page

88.7%

67.8%

Early access sign up

43.7%

-

Payment page

-

52.8%

First Opening Revenue

USD

2023/01/17~26

In less than 2 years the revenue will be 10X

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
k

Net Promoter Score

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9

Similar service average NPS: 60

Average User Rating

0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9

Out of 5

Retrospective

What This Project Taught Me

The Challenge

VAT season comes once a year and there's only one launch window with no room to test. If we shipped without validation, we'd burn months on an unproven product.

The Strategy

Build a (pre)MVP first and validate demand before committing to the full build.

Phase 01

Building (pre)MVP

Instead of building a full filing system, I reframed the (pre)MVP as a VAT estimation tool, letting users check what they'd owe based on their sales and expenses.

It delivered immediate value while testing our riskiest assumptions.

Results

82.4% wanted a mobile filing service

62.9% landing page conversion

12% early signup rate

From (pre)MVP to MVP

Phase 02

With demand validated, the challenge shifted: build a real filing product in 8 weeks, covering 12 filing methods.

Technical feasibility

I proposed a hybrid model, automating simple cases and routing complex ones to tax attorneys to cover the full market without the legal risk of full automation.

UX feasibility

I designed and ran 3 rounds of usability testing (pre-dev, post-dev, post-launch) to keep the flow aligned with users' mental models as complexity grew.

Business feasibility

I identified the two biggest drop-off points and led A/B tests to uncover what actually drove users to convert.

Final Design & Main Features

Introducing Simplified VAT Filing Service for SMB Owners

The first mobile VAT filing service designed for SMB owners of all types from meticulous reviewers to busy operators, enabling fast and accurate VAT filing.

Comprehensive Sales and
Expense Review

1

To reduce filing mistakes, users can review whether their sales and expense data is complete before submitting their VAT report.

Sales

₩25,000,000

Expenses

₩17,000,000

VAT season starts July 1!

Guided Manual Input

2

Surveys and usability testing revealed most SMB owners had little VAT filing knowledge, so we introduced a step-by-step input flow to guide first-time filers and reduce reporting errors.

Bringing Web-Level
Sales Management to Mobile

3

One of the 12 filing methods designed to give users web-level control on mobile, syncing tax system data while allowing edits and additions when needed.

4

File at your own pace.

Usability testing revealed many users don’t complete VAT filing in one session. To support different filing speeds and workflows, users can resume where they left off, with the home screen guiding their next step

For the full process,

please view on desktop.