How a Dutch Developer Built a $3 Million Solo Business From a Google Spreadsheet He Posted on X/Twitter

In 2014, a 27-year-old Dutch developer was sitting alone in a cheap hotel room in Chiang Mai, Thailand.

He had sold everything he owned the year before. His apartment. His furniture. His music equipment. He had packed a single backpack, bought a one-way ticket to Asia, and told himself this was the beginning of something.

Twelve months later it looked more like the end of something.

His YouTube channel, which had been earning him $2,000 to $3,000 a month from music tutorials, had dried up. His savings were almost gone. He was making a few hundred dollars a month. He had no job. No team. No plan. No investors waiting for his pitch deck.

He was staring at the ceiling of a hotel room in Thailand thinking, as he later told Lex Fridman, "I'm 27. I'm a loser."

What happened next built a $3 million per year business.

With no employees. No co-founders. No office. No funding.

Just a laptop, a Google Spreadsheet, and a problem he was trying to solve for himself.

That is the whole story. Except it is not. Because the how is what nobody shows you. And the how is everything.

The Setup

Pieter Levels was born in Amsterdam on July 11, 1987.

He grew up in the Netherlands tinkering with computers from a young age. As a teenager he taught himself MS-DOS Batch from a book called Windows for Kids. He got into music. Drum and bass. Electronic production. He studied at the Conservatorium van Amsterdam, learning music production seriously. He put tutorials on YouTube under a channel called Panda Mix Show, teaching people how to produce tracks on a Digital Audio Workstation called Reason 3.

The channel grew to 139,000 subscribers. By the time he graduated from Rotterdam School of Management with a business degree in 2012, it was making him $2,000 to $3,000 a month. Enough to live on.

But he was restless. His friends were getting office jobs. Taking the track. He could not make himself want it.

In April 2013 he sold everything he owned and flew to Asia. Twenty-six years old. One backpack. One laptop.

His first taste of life outside Europe had been a semester abroad in South Korea during university. That had done something to him. Shown him the size of the world. Made the Netherlands feel small in a way he could not ignore once he had seen it.

He traveled through Thailand, Bali, Hong Kong, Bangkok, Seoul, Chiang Mai. Living cheaply. Trying to build things online. Experiencing the life he had imagined when he sold his furniture.

And then the YouTube money stopped. The algorithm shifted. The income dried up.

He found himself back in the Netherlands briefly. Then returned to Asia. Broke. Anxious. Questioning everything. Alone in cheap hotel rooms with a laptop and no clear answer to the question of what came next.

His father gave him the advice that changed everything.

Get active. Do something. Stop lying there.

So he did.

The Constraint

Here is what makes Levels different from every other solo founder story you have read.

He did not have a flash of inspiration and build the perfect product. He built eleven things that mostly did not work and then built the twelfth one from the problem directly in front of him.

The constraint was not money. It was not time. It was not technical skill.

The constraint was survival. He was running out of money in Asia and needed to build something that generated revenue fast enough to keep him there. Not in ten years. Not after raising a seed round. Right now. This month.

That urgency produces a completely different kind of decision-making than the kind that happens in an office on a whiteboard.

He launched the 12 Startups in 12 Months challenge in early 2014. Documented on his blog from the beginning. Inspired by Jennifer Dewalt's 180 Websites in 180 Days project, but with a different problem. Dewalt was learning to code from scratch. Levels could already build things. His problem was finishing them and shipping them before he talked himself out of it.

One product a month. Each one with a real website and a real Stripe payment button. No exceptions.

He was not building for investors. He was not building for a dream exit. He was building to survive long enough to find the one thing that worked.

The products he shipped across those twelve months ranged from interesting to forgettable. Play My Inbox, a tool that turned music emailed to you into playlists, got coverage from MTV and Lifehacker. Tubelytics, a YouTube analytics tool, got picked up by major media. Go Fucking Do It, a goal-tracking app that charged you money if you failed to hit your targets, went viral on TheNextWeb and WIRED.

None of them made real money.

He kept going.

"We creatives have one common problem: finishing things. From musicians to writers to developers, we are perfectionists and projects simply never are just done."

The challenge was never about the twelve products. It was about building the habit of shipping until something connected.

The seventh one connected.

The Opportunity

This is the part that sounds simple in hindsight and was anything but obvious at the time.

Levels was sitting in Chiang Mai in the middle of his 12 Startups year. He was part of a growing, informal community of people doing the same thing he was doing. Remote workers. Freelancers. Developers. A small but real group of people who were trying to live and work from anywhere.

They all had the same problem he had. Where do you actually go? Not just which city looks good in photos. Which city has fast internet? Which one is safe? Which one is affordable enough that you can actually work and live there without burning through your savings in three months?

He opened Google Sheets. Started entering cities. Cost of living. Internet speed. Weather. Safety. The information was scattered across dozens of websites and forums and had never been put together in one place.

He shared it on Twitter. Not as a launch. Just asking people to help him fill it in.

"Help me build the definitive list of locations for digital nomads and remote workers, edit with me."

He posted that on June 24, 2014.

Then something went wrong. Or rather, something went accidentally right.

The spreadsheet settings were not locked. It was publicly editable. Hundreds of people found it and started adding data. City after city. Data point after data point. The spreadsheet filled up faster than he could manage.

The accident became the product.

He could see something real was happening. This was not a polite response from a small online forum. This was genuine hunger from people who had the exact problem he had and had never found a single place that answered it.

He built a website from the spreadsheet data. Basic. No fancy features. Just a list with a map. Filters for cost of living, internet speed, safety, weather.

He called it Nomad List. He launched it officially on July 29, 2014.

On its first day it made $600.

That number is easy to skip past. Do not skip past it.

He had been grinding through eleven products across months of work, watching each one fail to generate meaningful revenue. The spreadsheet he had shared as a casual tweet asking for help made $600 in a single day.

That is not a coincidence. That is the market telling you something as clearly as it knows how.

The Playbook

He built the product for himself and let that be the validation.

This is the part of the Levels playbook that sounds obvious but almost nobody does.

He was not theorising about a market. He was not conducting user interviews for a persona he had constructed. He was personally trying to solve the problem of where to live and work remotely every single week of his life. He needed the information Nomad List provided before anyone else did.

When you build for a problem you personally have, you know the answer when you find it. You do not have to conduct surveys to know whether the solution works. You experience it yourself first. And the people who find it next will find someone who genuinely understands their problem because that person lived it.

Every product Levels has built that succeeded started from his own life. Nomad List for where to go. Remote OK for how to find work that pays for the lifestyle. Interior AI for designing spaces while constantly moving. Photo AI for generating professional images without paying a photographer in every new city.

The market is in your own problem. Not in someone else's research document.

He charged from the first day and never apologised for it.

Nomad List made $600 on day one. That was not from enterprise contracts or a complex monetisation strategy. He had a simple paid membership for community access and premium features.

He did not spend six months building the product before charging. He did not offer it free and hope to figure out revenue later. He put a Stripe button on the page from the beginning because his 12 Startups rule required it. Every product had to be able to earn revenue from day one.

That rule turned out to be one of the most important decisions he made. Not as a business strategy. As a test of whether the idea was real. If people will not pay for the early rough version, they are probably not going to pay for the polished one either.

$600 on day one said they would pay. Everything else followed from that confirmation.

He hit the front pages and used the momentum without getting distracted by it.

Nomad List launched to the number one spot on Product Hunt. It went to the front page of Hacker News. Media coverage followed. Wired. BBC. CNN. The story of digital nomads was emerging as a cultural trend and Nomad List was sitting at the exact centre of it at the exact right moment.

Most builders in that position would have pivoted to a full-time PR operation. Levels kept building.

He shipped updates constantly. Added data points. Improved the filters. Listened to what members were asking for. The community started contributing reviews and photos. Members began organising meetups in Chiang Mai and Amsterdam. He went to some of them. Met his users in real life.

Revenue hit $120,000 annually from subscriptions alone.

He was making $120,000 a year. Solo. With a laptop. Traveling the world.

He did not hire anyone.

He built the second product by listening to the question everyone was already asking.

By early 2015 Nomad List members were asking the same question constantly. "This is great. But how do I actually find remote work to fund all this?"

He heard that question enough times that the answer became obvious.

In February 2015 he launched Remote OK. A job board specifically for remote positions. Companies paid $199 to post a job. Simple model. Immediate revenue. No complexity.

Remote OK became one of the largest remote job boards in the world. By 2021, during the pandemic-driven remote work boom, it was generating $140,000 per month.

The second product was not a pivot. It was not a new idea. It was the next logical answer to the question the first product's users were already asking. He just listened.

That sequence, build for your own problem, charge immediately, listen for the next question your users have, repeat, is the entire Levels playbook in four steps.

He scaled with automation instead of people and stayed the only employee intentionally.

When Remote OK was making $50,000 a month and Nomad List was making $30,000 a month, every startup playbook in existence would have told him to hire. Scale the team. Raise a round. Grow faster.

He did not.

He built over 2,000 automation scripts instead. Customer onboarding. Payment processing. Content moderation. SEO systems. Everything that a team of five would have handled, he automated. Not because he could not afford to hire. Because he decided that every employee makes a company slower and he had no interest in becoming slower.

"Taking VC money is like selling your soul. Most startups don't need to scale. They need to be profitable."

That decision shaped the entire character of what he built. No investors in the room. No board. No salary overhead. No management meetings. Just the products, the users, and the systems he had built to connect them.

He saw the AI wave before it was a wave and shipped fast enough to catch it.

In September 2022, Stable Diffusion was released. The first open-source generative image model that could run on a MacBook.

Levels downloaded it the same week and started experimenting. Not with a plan. Just playing. He tried generating houses. Generated interiors. Thought about what happened if someone uploaded a photo of their own room and got it redesigned by AI.

He built Interior AI and launched it. Within a week it was making $10,000 to $20,000 per month. It later stabilised around $40,000 to $50,000 monthly recurring revenue.

Then he discovered that the same model could generate photographs of people. He uploaded photos of himself. The outputs were, in his own words, so bad. Grainy. Distorted. Clearly not real.

He shipped it anyway as Avatar AI. Because his rule had not changed since 2014. Build it fast. Put it in front of real users. Improve based on what they actually do with it, not what you imagine they might want.

A few weeks later a company called Lensa AI launched the same concept. They had a team, better marketing, and millions in funding. They dominated the market. Avatar AI revenue crashed.

He did not close the product. He watched what Lensa was doing and watched what his own users were asking for. They did not want cartoon avatars. They wanted realistic professional photographs. The kind you would hire a photographer for.

He rebuilt it as Photo AI. Launched it in February 2023.

It made $100,000 in its first ten days. By September 2024 it was generating $100,000 every month. By November 2025 it was at $132,000 to $138,000 per month, running at an 87% profit margin.

By 2025, across Nomad List, Remote OK, Photo AI, Interior AI, and several smaller products, Levels was generating approximately $250,000 to $300,000 per month. Over $3 million per year.

Zero employees. Zero funding. Zero office.

A backpack. A laptop. And over 70 products, of which exactly four made real money.

Tools Used

PHP, HTML, and CSS. His entire portfolio runs on a stack most developers would refuse to use professionally. He learned PHP first and never switched because the tool that ships is worth more than the correct tool you are still learning.

Google Sheets for the original Nomad List prototype. Not as a proof of concept. As the actual product, until he had enough data to build the website version.

Stripe from day one. Every product he has ever shipped has had a payment button live at launch. No exceptions.

Twitter as his primary distribution channel. He shares revenue numbers publicly. Stripe dashboard screenshots. Everything. That transparency built a following of over 400,000 people that now makes every new product launch dramatically faster than the one before.

Replicate for AI model hosting once the AI products started. He did not build his own infrastructure. He used the existing platform and focused on the product layer.

2,000+ automation scripts instead of a team. Not one hire. Every process that could be automated was automated.

Timeline

1987, July 11: Born in Amsterdam, Netherlands.

2007: Starts uploading music tutorials to YouTube. Channel grows to 139,000 subscribers. Making $2,000 to $3,000 per month by 2012.

2012: Graduates from Rotterdam School of Management with a business degree.

2013, April: Sells everything. Flies to Asia. Begins the digital nomad lifestyle. YouTube income eventually dries up. Returns briefly to Netherlands. Faces anxiety and depression.

2014, early: His father tells him to get active and do something. He launches the 12 Startups in 12 Months challenge. Documents it publicly on his blog.

2014, June 24: Posts the Google Spreadsheet to Twitter asking for help building a list of cities for digital nomads. Spreadsheet settings accidentally left public. Hundreds of people add data immediately.

2014, July 29: Nomad List website officially launches. Makes $600 on the first day.

2014, late: Remote OK begins as a jobs page inside Nomad List.

2015, February: Remote OK spins off as a standalone job board. Companies pay $199 per listing. Grows rapidly.

2016: Nomad List and Remote OK together generating over $600,000 annually. Still zero employees.

2017: Annual revenue reaches $120,000 from Nomad List subscriptions alone.

2021: Remote OK peaks at $140,000 per month during the pandemic remote work boom. Then the market cools and it drops to roughly $10,000 per month before recovering to approximately $40,000 monthly.

2022, September: Stable Diffusion releases. Levels experiments immediately. Builds Interior AI. Profitable within a week. Launches Avatar AI. Competition from Lensa AI kills it. Rebuilds as Photo AI.

2023, February: Photo AI launches officially. Makes $100,000 in the first ten days.

2024, September: Photo AI crosses $100,000 in monthly recurring revenue.

2025, November: Photo AI at $132,000 to $138,000 per month. 87% profit margin. Full portfolio generating approximately $250,000 to $300,000 per month. Over $3 million per year.

Over 70 products built total. Four made real money.

Mistakes and Lessons

The Avatar AI story is the honest version of the Levels narrative that most breakdowns skip.

He built it fast. Shipped it when the quality was, by his own description, terrible. Watched Lensa AI arrive with a team and funding and dominate the exact market he had just entered.

A different builder folds at that point. Cuts the product. Moves on. The market has spoken.

Levels did something more interesting. He looked at why Lensa was winning. Better quality output. Better marketing. And then he looked at what his own users were actually asking for. Not avatars. Not cartoon versions of themselves. Real-looking professional photographs. The kind that cost $500 to $1,500 to produce with an actual photographer.

He rebuilt the product around that gap. Photo AI.

The lesson is not about pivoting fast. It is about the willingness to keep watching after you have lost. Most people turn away from failure because watching it is uncomfortable. Levels watched it carefully enough that he found the real product hiding inside the failed one.

The second mistake is one he has acknowledged directly in interviews. Too many projects with not enough focus in the early years. He built over 70 products. Four worked. That 95% failure rate is not a bug in his system. It is the system.

But the cost is real. Each failed product represents weeks of building and launching and watching nothing happen. The question is whether that cost produces returns that justify it. In his case, clearly yes. One Nomad List more than pays for eleven things that did not work.

The deeper lesson: the volume approach works when you build fast and charge early so that each failed bet costs weeks, not years. The moment a product becomes a long development cycle without revenue confirmation, the entire model breaks down.

Ship fast. Charge immediately. Watch the numbers. Move on if they do not appear. This only works if the bets are genuinely small.

The Psychology

Three things about how Levels thinks that explain results his technical skills alone cannot.

He treats failure as data, not verdict.

He has said this clearly. Only four out of over 70 products he built ever made real money. His hit rate is roughly 5%.

Most people build one thing. It does not work. They conclude they are not a builder. Levels built 70 things and concluded he needed to see more evidence before drawing any conclusion. That reframe is not optimism. It is statistical thinking applied to product development.

If your hit rate is 5%, you need volume to win. And if each bet is small enough that a miss does not end your experiment, then 70 tries produces four winners and that is exactly enough.

He has never confused freedom with comfort.

He was making $2,000 to $3,000 a month from YouTube. Comfortable. Free time. No boss.

He sold everything and flew to Asia with a backpack.

He was making $30,000 a month from Nomad List in 2016. Very comfortable. No employees. Complete autonomy.

He kept building. Added Remote OK. Added AI products. Kept shipping.

The goal was never a number that he would reach and then stop. The goal was a life that felt like his. That goal does not have a finish line. It has a direction. He has been walking in that direction for over a decade without stopping to ask whether he has arrived.

He built his transparency into the product itself.

Sharing revenue numbers publicly was not a marketing decision. It started as personal processing. He needed to track what was working. He made that visible.

But what it built over time was something money cannot buy. A following of 400,000 people who had watched him fail and recover and build and fail again across more than a decade. By the time Photo AI launched in February 2023, that audience had been watching him since 2014. They did not need a launch campaign. They were already invested.

His transparency is the compounding asset every other number is built on top of.

The 2026 Builder Translation

Nomad List started as a Google Spreadsheet in 2014. It grew into a platform making over a million dollars a year. That specific path is harder to replicate now because the market for digital nomad information is no longer empty.

But the principles underneath Nomad List are not market-specific. They are structural. And in 2026 they are arguably more powerful than they were in 2014.

Here is what the same playbook looks like if you started it today.

Start with your own problem. Not someone else's research.

Levels did not study the digital nomad market and conclude there was an opportunity. He was the digital nomad. He had the problem. He built the spreadsheet for himself and then realised other people needed it.

In 2026 you are surrounded by problems you personally experience because of your specific job, your specific industry, your specific life. The gap between how something works and how it should work is your product idea. Not the one from a trends report. The one from Tuesday morning when something frustrated you so specifically that you opened a new tab to search for a solution and found nothing useful.

Write that down. That is the start.

The spreadsheet is still the fastest MVP available.

In 2014 Levels posted a Google Sheet and called it a product.

In 2026 the same move is available. A public Notion database. An Airtable. A Google Sheet with a form attached. A simple table of information that answers a question people in your industry are searching for constantly.

You do not need a Next.js app with Supabase auth to validate that a market exists. You need a URL people can visit and data they cannot find anywhere else. Build that first. Charge for the deeper version later.

AI has made the gap between idea and shipped product smaller than it has ever been.

Levels built his entire product portfolio in PHP because it was what he already knew. He did not wait to learn the correct technology. He shipped with what he had.

In 2026 you have AI-assisted development tools that compress the distance from idea to working prototype dramatically. A builder in 2026 with Cursor or Claude can build in days what would have taken Levels weeks in 2014. The excuse that you cannot build fast enough has become genuinely difficult to justify.

The bottleneck is not technical. It is the willingness to ship something imperfect and watch what real users do with it.

Build in public before you have anything to sell.

Levels has 400,000 followers on X because he has been sharing real numbers and real failures since 2014. Photo AI launched into that audience and made $100,000 in ten days. Not because the product was better than everything else available. Because the people watching already trusted him before the product existed.

You can start building that trust today. Not after you have a product. Not after you have validated the idea. Now. By sharing what you are learning in your industry. What problems you are noticing. What you are trying to build and why.

The audience you build before the product is worth more than any launch campaign after it.

The hit rate matters less than the bet size.

Levels has a 5% hit rate across 70 products. That sounds like a terrible record until you understand the structure underneath it.

Each bet was small. Each product was built in weeks, not years. Each one had revenue confirmation attempted from day one. The failures cost weeks. The wins paid for decades.

In 2026 that structure is available to anyone. Not the 70 products. The principle. Make the individual bet small enough that you can afford to be wrong. Then make enough bets that the law of averages works in your favour.

One product that takes three years to build and fails is a tragedy. Ten products that each take three weeks to build and one of which works is a portfolio.

Modern Opportunity Radar

If this playbook is making you think about what to build, here are three real opportunity spaces in 2026 that share the same structural DNA as Nomad List's origin.

Crowdsourced data products for underserved professional communities.

Nomad List was a spreadsheet that aggregated information scattered across the internet and made it filterable. That gap exists in every professional community right now. Healthcare workers comparing contracts. Logistics managers mapping carrier rates by region. Teachers tracking curriculum resources by state. The information exists somewhere. Nobody has assembled it into one filterable place. A public spreadsheet asking for contributions is still the fastest way to find out if the gap is real.

AI tools that make expensive professional services accessible to individuals.

Photo AI replaced a $500 to $1,500 photographer with a $29 monthly subscription. Interior AI replaced a $2,000 design consultation with an upload and a click. That substitution pattern is available in dozens of professional services that are currently priced out of reach for individuals and small businesses. Legal document drafting. Financial modelling. Medical second opinions. Architectural rendering. In every case the question is the same. What does a professional charge for something that an AI tool could deliver at a fraction of the cost and who is the specific person who would pay for the accessible version.

Niche job boards for categories that the general boards ignore badly.

Remote OK succeeded because the general job boards were terrible at surfacing remote work specifically. That same dynamic exists today in dozens of categories. AI-specific roles. Climate tech. Regenerative agriculture. Web3 development. Roles in emerging markets. Any category where the people hiring and the people looking for work have specific needs that Indeed and LinkedIn are too broad to serve well. A focused job board with clear positioning and a simple $199 posting fee can generate meaningful revenue from day one with very little infrastructure.

How You Can Replicate This

You are employed. You have a specific domain of knowledge your industry has given you. You have access to problems people in your field experience every single week.

Here is the sequence that works.

Write down the question people in your industry search for constantly and never find a good answer to. Not a general question. The specific one. The one your colleagues ask on Slack. The one that comes up in every team meeting. The one you have personally Googled seven times and found nothing useful.

Open Google Sheets. Start building the answer to that question as a public spreadsheet. Add every data point you already know. Share it on Twitter and LinkedIn with a direct ask for others in your industry to help fill it in.

Watch what happens in the first 48 hours. If people add data, you have something real. If nobody touches it, you have learned something important for free.

If people engage, build a simple website from the data. Price access to the premium version between $9 and $49 per month. Share the revenue numbers publicly from the first day.

That is the whole thing. The spreadsheet is the prototype. The website is the product. The revenue number you share publicly is the distribution engine.

You do not need a co-founder. You do not need investors. You do not need a team.

You need a problem your industry has and the willingness to ship the rough version before you have finished building the good one.

Related Playbooks

The Markus Persson playbook covers how a Swedish developer built the best-selling video game in history during lunch breaks and evenings at a day job, using community as the entire distribution engine. Levels did the same thing. The digital nomad community spread Nomad List without any marketing because the people using it felt like they were part of building it.

The Elston playbook covers how a JP Morgan employee found the one audience inside an impossibly crowded market that nobody else was talking to. Levels did the same with digital nomads in 2014. Not a new category. An existing category nobody had served specifically yet.

Every playbook in this series returns to the same question. Not what should I build. Who is currently being failed by everything that exists and what is the smallest possible thing I could ship this week to find out if they would pay me to solve it.

Premium Insights

Here is the detail that most Pieter Levels breakdowns skip entirely.

The accident that made Nomad List work was not the spreadsheet going viral. It was what he did the moment he saw it happening.

Most people in that situation would have cleaned up the spreadsheet. Locked the editing permissions. Built a proper product before showing anyone else. Done it correctly.

Levels did the opposite. He watched hundreds of strangers add data to his publicly editable spreadsheet and instead of trying to control it, he accelerated it. He asked for more. He built the website from the crowdsourced data while the crowdsourcing was still happening.

He let other people build the product with him before he had finished building it alone. That is not a strategy that appears in any startup playbook. It is what happens when you care more about whether the idea is real than about whether you get full credit for building it.

The community became the product because he let them in before it was ready.

The second insight is about the number 70.

Levels has built over 70 products. Four made real money. Most people read that and think the lesson is about persistence or volume.

The actual lesson is about what happens to your judgment after the first five failures.

After five things that do not work, most people stop shipping. They conclude that the market is wrong, or that they need more preparation, or that the timing was off, or that they need a co-founder, or that they need funding. They find a reason to stop.

Levels built eleven things that did not work before Nomad List. Eleven. And he shipped Nomad List the same way he shipped the first ten, without any special confidence that this one was going to be different. He just kept applying the same method to the next problem in front of him.

The insight is not about the product. It is about the ability to keep your judgment clean after repeated failure. To not let the previous results contaminate the next attempt. To look at the problem in front of you with the same fresh curiosity you brought to the first one, even after the tenth one did not work.

That is genuinely hard. Most people cannot do it.

The ones who can are the ones building $3 million businesses from hotel rooms in Thailand.

Your Move This Week

Write down the single question that people in your industry, your team, or your professional community search for constantly and never find a clean answer to. Open Google Sheets. Spend two hours building the start of that answer as a public spreadsheet. Share it on X and LinkedIn with a direct ask for others in your field to add what they know. Do not build a website first. Do not design a logo. Do not write a business plan. Just share the spreadsheet this week and watch what happens in the first 48 hours. The market will tell you whether the idea is real. It told Levels on June 24, 2014. It can tell you this week.

* * *

Levels built a $3 million business with a 5% hit rate across 70 products.

That is not a story about luck or talent.

It is a story about what happens when you keep your bets small enough that failure teaches you something instead of ending everything.

The spreadsheet is still available. The problem in front of you is real.

The only question is whether you share it before it is ready.