Why I Left My Job as an 8-Year Data Analyst to Start DataPopcorn

I spent close to eight years working with data at SOCAR, TADA, Zigzag and Coupang, then left to start DataPopcorn. The first picture I had in mind was a "DataPopcorn Universe": working analysts writing content and sharing courses together. A year in, though, roughly 80% of revenue (by my own estimate) came from lectures I taught in person, and once the lectures dried up, DataPopcorn had nothing it could do on its own. This post is a portfolio of everything I think about DataPopcorn: why I started it, what actually happened in the first year, what students taught me, the plan for year two, what I learned running a company on my own money for a year and a half, and why I now want to go back to working at a company.

A year spent getting ready to move beyond growth built on lectures and outsourced projects, and to leap into a scalable business built on DataPopcorn's own platform.

That is how I summed up DataPopcorn's first year in one sentence.

I was a data analyst

Before I left, I worked in data analysis and data engineering. Here is where I worked.

Industry	Company
Mobility	SOCAR (Korea's largest car-sharing service), TADA (VCNC, ride hailing)
E-commerce	Kakao Style (Zigzag, a fashion shopping app), Coupang (Korea's largest e-commerce company)
OTT	Coupang Play

The domain kept changing, from mobility to commerce to OTT to advertising, but the core of the job stayed the same: using the company's data to solve the problems the company had chosen.

I left for three reasons

What I wanted to do wasn't on the company's priority list

Inside a company, the data team moves within the company's products and roadmap. That's how it should be. But most of what I wanted to try sat outside that, and inside the company its turn never came. The gap between what I was good at and what I wanted to do kept growing.

I wanted to really use AI in my work

As AI and automation tools started pouring out, I wanted to see for myself what changes when you plug them into real work. Not a meeting that ends with "let's consider adopting it," but building something with my own hands, running it, and teaching other people how it works.

I could see myself a few years out

At year eight, I could picture myself a few years ahead pretty clearly: someone who does the current job a little better. Not a bad future, but I didn't want that to be all there was. And it felt like leaving would only get harder with time.

The first picture: the DataPopcorn Universe

The company I wanted to build

When I started DataPopcorn, the company I imagined had three faces.

• An education company that teaches practical data and AI automation skills
• A group of experts who raise companies' productivity with no-code tools like n8n and AI
• A consultant who lowers the barrier to technology for individuals and companies

What I really wanted was an MCN

There was a bigger ambition inside that. What I really wanted to build was less a company and more a kind of MCN (multi-channel network). I wanted my own crew, so to speak.

• Recruit working analysts to contribute data content
• Turn those contributions into videos for YouTube and Instagram
• Train junior instructors and pass lecture requests on to them when they came in

The idea was not just me, but many analysts creating content and courses together, and I called it the "DataPopcorn Universe."

It's a loop: the crew in the middle writes content and takes on lectures, the people we meet through content and lectures gather in the community, and from there come new people to work with. The universe splits into three pillars.

Pillar	What it does
Education · consulting · mentoring	Corporate training, courses for outside education providers, work automation consulting, mentoring (online and offline)
Creators	Publishing data and AI automation content on YouTube and Instagram
Community	Wagu Wagu

How I first planned to make money

The revenue model was simple.

Category	Initial plan
Lectures	B2B corporate training, teaching for outside education providers (Learning Spoons, Codeit, Team Sparta and others, all Korean online education companies)
Consulting	Custom work automation setups for each company
Services	Planning stage
Products	Planning stage

I left services and products as "to be planned later." At the time I had no idea how that decision would come back a year later.

What I actually did in a year

Lectures, consulting and small experiments

Most of the first year was lectures and consulting. Here's a summary.

Category	What I did	Revenue contribution	What was hard	Would I do it again?
Lectures	Government-funded training (Woongjin KDT advanced course; KDT is Korea's government-funded K-Digital Training program), corporate B2B data and AI automation tracks	High	Time spent on in-person training and resources spread thin	Yes
Consulting	Work automation diagnosis and adoption guides for early-stage companies and students	Medium	The effort of proposing and building from scratch every time	Needs to be packaged after setup
Service experiments	Planning our own platform, running an n8n book on WikiDocs (a Korean online publishing platform)	Medium	No immediate revenue while still in planning	Yes (will grow it)
Products / Smart Store (Naver's online storefront platform)	None	Low	-	-

Some of the organizations I worked with: LG Electronics, Upstage, Mirae Asset Securities, Lotte Card, LG Uplus, Incheon Sejong Hospital, Kakao Style and Coupang Play.

Along the way I co-authored the book n8n Does It All (《n8n이 다 해줌》) with Hanbit Media (a Korean tech publisher), I run the Korean n8n community as an n8n Global Seoul Ambassador, and I spoke on stage at Google I/O Seoul. This is also when I started building up content on WikiDocs and YouTube.

Best-selling topics vs. best-received topics

The three topics that sold the most lectures and consulting were:

• n8n basics, prerequisites and tool setup
• Prompt engineering and applying AI to real work
• Data-driven work automation

What got the best reaction was a bit different: hands-on training where people automate their own real work with no-code (n8n), rather than theory, and workflows for unattended blog and social media posting. Intro courses sold the most, but the moment students were most engaged was when they saw their own work actually running on its own.

Year one in numbers

80% of revenue was lectures

Looking back on the year, I wrote down my rough sense of the revenue split.

Item	Share	Details
Lectures	80%	Learning Spoons, short- and long-term corporate B2B training, etc. All taught by me
Consulting	15%	Follow-up consulting after training and individual requests
Services / automation	5%	Early tests and building our own workflows
Products / Smart Store	0%	Plan to expand into selling packages and workflows

What made the most money, and why

The two things that brought in the most revenue and were the most efficient:

1. Long-term teaching for outside education providers (KDT and others) and B2B clients
2. Short corporate seminars and AI literacy and automation training for employees

There were reasons it made money.

• Customers were in a hurry. Following the AI trend, demand for retraining employees exploded at company after company.
• It was something I could do well. Practical data and n8n expertise, plus the teaching experience I had built up, matched that demand.
• The rates were good. B2B lectures funded by corporate and government training budgets paid well.

But this model couldn't last

It made money, but I felt it would be hard to keep going on this model alone, because most revenue came from in-person training that I, the CEO, taught myself. There were four reasons.

1. It took far too much of one person's time.
2. It depended heavily on me. If I didn't teach, revenue was zero.
3. It wasn't recurring revenue. When a course ended, the next month started from zero again.
4. No platform, no asset, was building up inside the company.

What turned out differently than I expected

Three things surprised me.

First, the business was tied directly to my time and stamina as the instructor. The more lectures I took on, the more revenue grew, but the less time I had to prepare the next one and the less room to try anything new.

Second, customers wanted working tools, not training. Demand for workflows and automation tools that actually run was far bigger than demand for one-off training.

Third, teaching how to use the tool wasn't enough. Students wanted tangible results beyond how to use n8n, such as monetization and automation for their specific role.

What I learned from students

The three questions I kept getting

Almost every question I got in lectures and consulting came down to these three.

"How do I set up n8n? I got an error, how do I fix it?"

"How can I use what I learned in my role (PM, HR, marketing) right away?"

"Can I use this to automate Shorts or a blog and make money from it?"

What students really wanted

Putting those three questions side by side, I saw what they had in common. Not one of them asked me to explain the underlying principles more. What students wanted was:

• How to use AI in their own work
• How to save time with automation
• Real tools, templates and ready-to-use workflows

In one sentence: not studying principles, but finished workflows that cut their working hours right now, plus maintenance support.

Problems worth paying for

So I separately listed the problems customers would actually pay to solve.

Problem	Customer	Likelihood to pay	Potential as a service
Adopting n8n workflows tailored to each role	Solo businesses, startups	High	High
Setting up AI automation for monetization (blogs, Shorts)	Side hustlers, marketers	High	High
n8n troubleshooting and after-sales support	Practitioners, company staff in charge	Medium	High (sold as a package)

What went well in year one

Results

• I generated revenue. Outside lectures such as KDT and B2B training created stable cash flow.
• I built teaching and consulting skills. After many lectures, I turned the curriculum into a structured path from beginner to practitioner.
• I built brand trust. "n8n no-code automation = DataPopcorn" became a thing, and it led to a book contract with Hanbit Media.
• I set a direction for AI and automation. I finished a concrete roadmap for moving beyond plain education into selling n8n workflows and micro-services.

In concrete terms:

1. Winning and successfully running major outside training contracts, such as the Woongjin KDT advanced course
2. Finishing plans for building our own platform in 2026 and selling VOD courses and packages
3. Preparing the Hanbit Media n8n book n8n Does It All for publication
4. Starting to build content assets on YouTube, WikiDocs and elsewhere

What sets DataPopcorn apart

I think this is why customers chose DataPopcorn: we explain things in a way that fits real work, we can connect AI, data and automation, and there is trust in Inyoung Lee's personal expertise.

If I pick three core strengths:

1. Practical automation skills that combine n8n with AI tools (Claude and others). I'm confident this is among the best in Korea.
2. Teaching that explains complex technology at a non-developer's level. I think this is where I'm strongest by a wide margin.
3. Execution. I can build micro-services and workflows myself with vibe coding.

What I regret about year one

I chased revenue and neglected branding

Looking back after the first half, this is what I regret most. I started fast, but I was so focused on making DataPopcorn profitable that I never built a proper brand.

Most lectures went through outside education providers and corporate training, and DataPopcorn had almost no channel of its own for meeting customers directly. While things were going well, I didn't see the problem. But the moment the lectures stopped, there was nothing DataPopcorn could do on its own through its own platform.

I also wrote down how customers see DataPopcorn:

• A company built on Inyoung Lee's personal brand
• An n8n and automation speaking company
• A practical data education company

Of the three, the first was the most accurate. The universe I first pictured had stalled exactly where it was. Far from gathering a crew of analysts, the whole thing only ran if I taught every lecture myself.

The three most fragile parts of the structure

1. Revenue proportional to time, which doesn't move unless CEO Inyoung Lee teaches in person
2. No platform of our own for meeting customers directly and keeping those relationships
3. Course materials and workflows that get used once and used up, instead of being sold again

Heavy dependence on lecture and consulting revenue, dependence on the CEO's time, rebuilding proposals and operating materials from scratch every time, and not enough recurring revenue. In the end it was all the same problem.

The community failed once, too

Students wanted to stay connected after the lectures, so I opened KakaoTalk (Korea's dominant messaging app) open chat rooms and a Discord server, which at one point reached 950 and 300 members. But I was always the only one starting conversations, and whenever lectures piled up, the community was the first thing to slip. Both eventually went quiet.

Operations were the same. I handled customer management and student support by answering one by one in KakaoTalk open chat rooms, and posting content to Instagram, Threads, LinkedIn and the tech blog was all manual.

That's why I started Wagu Wagu Club

After those failures, I started Wagu Wagu Club in Q4.

This time, instead of a room of hundreds of people, I designed a meetup where around 10 people gather offline on a set topic each weekday and spend three hours actually building something. The goal is to build, on the spot, the "results I can use in my own work right away" that students wanted most, such as video automation, repetitive-task automation and building an AI assistant.

An always-open chat room was a structure that wore me out, but a set time where people gather to build together is the thing I do best. And Wagu Wagu Club is the first place where DataPopcorn recruits, takes payment and meets people directly, not through an outside platform. I hope that someday, members of the universe I first pictured will come from the people I meet here.

DataPopcorn in year two

What to cut back, keep and grow

Cut back

1. One-off work that leaves nothing behind in the company once it's done
2. Manual customer support and repetitive guidance that can't be turned into a system
3. Simple time-for-money work that doesn't lead to recurring revenue

I'll keep doing lectures and consulting, but change them so each one leaves behind reusable assets.

Keep

1. Keep updating WikiDocs and offering high-quality free information (for inbound leads)
2. Regular YouTube and podcast content
3. Strategic partnerships with major outside channels such as JoCoding (a popular Korean coding YouTube channel)

Grow

1. DataPopcorn's own platform. A channel for selling VOD courses and workflow packages
2. Offline community. Small, cohort-based study groups and regular meetups
3. Share of product revenue. Revenue models beyond lectures, B2C and B2B micro-services

Courses to turn into packages

Package candidate	Audience	Core content	Follow-on service
n8n intro course (2026 revised edition)	Beginners, non-developers	n8n setup and basic logic	Workflow sales, hands-on course
n8n hands-on course (monetization, by role)	Marketers, PMs, HR	Monetizing Shorts and blogs, role-specific work automation	Maintenance package, meetup invitation
Claude Code & Antigravity course	Intermediate users	Using the latest AI tools beyond n8n	Advanced offline study groups and meetups

I also listed products I can offer after a course ends:

• Building and selling n8n workflows I've made
• Building AI work tools for each department (introducing micro-services)
• One-month PoC consulting (tied to an after-sales support package)

What to turn into services

Here are work tools I can build quickly with vibe coding:

• Internal automation dashboards built on n8n and vibe coding
• Unattended content generators and publishers for specific domains such as blogs and YouTube
• Custom chatbots and automated customer support systems

As service candidates:

Service candidate	For whom	Problem it solves	MVP feasibility
n8n workflows & support package	Course students, small and solo businesses	Apply it right away without complex setup	High
DataPopcorn VOD subscription/purchase platform	Our own students	Watch scattered courses in one place	High
n8n + vibe coding micro-services	B2B, B2C	Companies and individuals without automation skills	Medium

The first experiment

The first service I'll push is selling n8n workflows and after-sales support packages online.

• Why: It's the surest pipeline for turning the proven demand from lectures straight into product sales, in other words recurring revenue.
• First customer candidates: Existing students from outside training, WikiDocs and YouTube visitors
• MVP to build within two weeks: A product checkout page on our own platform (web) and one package uploaded (for example, an auto-posting workflow)

Automating DataPopcorn's own repetitive work first

A company that automates other people's work was doing its own work by hand. So I'm starting with DataPopcorn's internal repetitive tasks.

Task	Now	New approach	Priority
Multi-platform social publishing	Manual uploads	One-source, multi-use automated publishing with n8n	High
Support and student inquiries	Manual replies in open chat rooms	First-line responses with n8n + an AI chatbot	Medium
Meetup and webinar announcements	Emails and texts sent by hand	Notifications sent by automated workflows	Medium

Year-two goals

DataPopcorn's three most important goals for year two:

1. Build and launch our own platform. Make the ecosystem independent with a platform that ties together course and workflow sales.
2. Publish the book and make the related events a success. Claim a position in the industry through the book, webinars and online/offline meetups.
3. Plan and start building micro-services (n8n + vibe coding). Create product-based recurring revenue through workflow package sales.

2026 business plan

Quarter	Theme	To do
Q1	Planning and setup	Plan and build our own platform (web), plan the n8n intro course and workflow packages, plan webinars
Q2	Launch and expansion	Officially launch our platform and sell VODs (intro, monetization, etc.) and packages, regular offline meetups, ramp up YouTube and social media marketing
Q3	Content upgrade	Publish Hanbit Media's n8n Does It All and market the book, offline cohort study groups 1-4, new VODs such as role-specific n8n hands-on courses and Claude Code
Q4	Service diversification	Build and launch micro-services, offline study groups 5-6, meetups and webinars, plan for the next year

The structure I want by 2027

Revenue source	What it looks like in early 2027
Lectures	A larger share of passive income from VODs on our own platform
Consulting / services	Structured around n8n support package sales and premium B2B consulting
Products	Automated sales of workflow templates and packages
Operations automation	A fully automated pipeline from content planning (wiki) to multi-channel social distribution

The company I want to be seen as in year two

Right now DataPopcorn looks like a company built on Inyoung Lee's personal brand. In year two, I want it to look like this:

DataPopcorn is a technology platform company that hands anyone everything from A to Z, from courses to its own workflows and micro-services, so they can easily adopt AI and automation.

Rewritten in my own words:

DataPopcorn is an "AI automation solutions company" that doesn't stop at teaching principles, but provides automation tools (workflows and micro-services) you can use at work right away, cutting real time and cost.

Wrapping up the year

Year one in one line. A period of validating the market and cash flow through lectures and outside training.

Three biggest achievements

1. Securing revenue stability through major outside training (KDT and others)
2. Confirming demand for an n8n automation toolkit through lots of feedback from the classroom
3. Setting a concrete 2026 roadmap for moving to our own platform, a book and micro-services

Three biggest lessons

1. People open their wallets more readily for real time savings (workflows and tools) than for plain knowledge (lectures).
2. A revenue model that depends heavily on the CEO teaching in person has a fatal limit when it comes to scaling up.
3. You need your own platform (your own store) and community for traffic to stay with you as an asset.

Three core strategies for year two

1. Turn traffic into an asset by launching our own platform (online and offline combined)
2. Expand into a product business by selling n8n workflows and support packages
3. Build DataPopcorn's own automation community ecosystem connecting WikiDocs, the book, meetups and social media

DataPopcorn's first year was about creating market touchpoints and cash flow through lectures and consulting, and sharpening our best weapon, n8n automation. In year two, building on that experience, I'll reduce our dependence on outside parties and shift to an ecosystem that sells products (services and workflows) on our own platform. To get there, I'm focusing on launching our own VODs and packages, publishing the Hanbit Media book, and building an online and offline community and meetups.

What I learned running a company on my own money

I didn't take investment

DataPopcorn has run on my own money from day one. It's a self-funded company in the most literal sense.

I didn't take investment for two reasons. First, I didn't have an idea clear enough to raise money on. Second, I really wanted to do my own thing freely. Once other people's money comes in, you have to follow the speed and direction that money expects. Before that, I wanted to do what I wanted.

Because it was my money, I learned what money is worth

Running a company on my own money taught me a huge amount about what it takes to run a business. Above all, I learned how precious money is. Money that was a "budget" inside a company was, here, my money down to the last won. Every time I spent, I asked whether it would really come back, and that sense is hard to develop while you're employed.

I decided to try things even if they didn't make money

So I made up my mind to try businesses even if they wouldn't make money. I felt that if not now, I'd never get the chance. These were things I could never have tried while employed.

My 3D printing business was one. So were the Claude Code clicker and the NFC Bible keyring. These are products I make and sell myself, completely different from lectures.

I wasn't being greedy, either. I hoped these products would bring in a steady 1,000,000 won a month. But marketing was incredibly hard. Selling was much harder than making.

And why I want to go back to working at a company

DataPopcorn's goal was survival

DataPopcorn's biggest goal was one thing: could I survive without a company? I wasn't aiming for the life of a big-time entrepreneur. From the start, it was a challenge focused on survival.

I have my answer. Even now, I can live just fine without a company.

I proved a lot in a year and a half, and I have no regrets

In a year and a half, I achieved much of what I wanted to do. I published a book, taught at companies and bootcamps, ran a community as an n8n Ambassador, stood on the Google I/O stage, built several content channels run by AI agents, and made and sold my own 3D printed products and an NFC service.

I think the fact that I ran a company on my own for a year and a half, even with an employee, proves a lot by itself. I'm disappointed I didn't bring in more revenue. But I have no regrets.

That doesn't mean I've done everything I wanted. I still want to take on more challenges, and there's still work I want to do.

There are limits to what an individual, not an organization, can do

So why go back? Because there are clear limits to what an individual, rather than an organization, can do.

Even if DataPopcorn grows as far as it can, unless it produces a service or a breakthrough product, growth eventually stops. That's where the "structure that doesn't run unless the CEO teaches in person" I described above ends up. My own time and ability are the company's ceiling.

Building and selling were different skills

What I regret most is growth. I built products and services, but I failed to grow them.

In the end, building something well with AI and having the sales skills to sell it were different things. My AI skills grew enormously in a year and a half. But the skills of an entrepreneur who makes money are something I'm still developing.

Working alone showed me how much the people around you matter

The biggest thing I felt working alone for a year and a half was how much it matters to have an organization and people you can work with.

By "together" I mean less holding onto one task and building it side by side, and more everyone running toward a shared goal. You learn a lot from each other along the way.

Working alone also taught me about myself. I work alone, but I'm someone who feels lonely easily. I realized I like refining my thinking by asking people things and putting questions to them.

Being alone can be comfortable. But to go the distance, I think it's right to belong to a company.

I needed results beyond lectures

The AI education market has also changed a lot. The supply of AI instructors has grown sharply, and after more than 100 hours of teaching, I kept feeling that what students really need lies beyond the lecture. As I wrote above, what students wanted was results that work right now.

Now the human job is deciding what to build

With Opus 5.5 and GPT 6 Astra, model performance has become overwhelmingly good. I subscribe to both Claude Max x20 and Codex Pro x20, and lately I've even run short of tokens. That's how much more work I can hand off to AI.

I no longer need to study separately for things like prompt engineering or harness and loop engineering (skills). I think there's only one thing left for people to do: define clearly what to build and why.

Realizing that changed what I think about. Now I think less about how to teach and more about how to build well. That's why I want to go back to working at a company.

When I left, I wanted to do the work I wanted with my own hands, and for a year and a half, I did. I also confirmed I can survive without a company. Now I want to work among people running together toward a shared goal, deciding what to build and building it well.