Shutting down startup: lessons learned

Shutting down startup: lessons learned

In 2022 I left my job to explore a new venture: I would teach English to the laptop class in Latin America, and connect them with employers in the US.

This week I made the difficult choice to shut things down.

This post shares some of my theses, process, and lesssons learned.

Why Latin America

Some notable stats:

  • An engineer in Latin America who jumps to the US market will make more than $1 Million more over the course of their career.
  • There are 4.3 Million engineers in Argentina, Colombia, and Brazil, but only 149k with a LinkedIn profile in English. This means every hiring manager in the US is effectively competing for 4% of the market.
  • Argentina has a baseline of English education in school, but because of cultural differences, Millennials and older generally avoid speaking English.
  • As of 2022, the average software engineer made $34k per year in Argentina, compared to typical salaries around 100k in the US.
  • While engineers in the US report wanting about a 25% raise for their next job, engineers in Latin America expect about a 50% raise.
  • Before 2020, many companies could hire engineers in Tier 3 US markets at a 50-70% discount. With remote work becoming mainstream, the location discount dropped to around 20% percent.

I believe the salaries in LatAm will continue to increase until they approach the same salaries as tier 3 US markets, likely around 2025-2028. There are only a handful of cities where this shift can take place en masse:

  • Sao Paulo, Brazil: the largest density of tech talent in LatAm, and the de facto hub for the Fortune 500 on the continent. The size of the country and Portuguese being its first language add a lot of friction: more often than not, you’ll need to hire someone who speaks English as their third language.
  • Buenos Aires, Argentina: second largest density of tech talent in LatAm, likely shifting to the dollar as a currency in 2024. Very high level of English spoken and a neutral accent that makes it easier to speak English.
  • Mexico City, Mexico: industrial re-shoring and US expats have already pushed up the currency and the prices. Flights make this a great option if you need to manage a team in person. Downside: drug cartels shooting at commercial airliners.
  • Bogota, Colombia: the talent hub for many offshore sales and marketing teams in tech, Bogota is often overlooked.
  • Medellin, Colombia: infamous for drugs, Medellin has invested a lot in becoming a tech hub. Downside: Narcos created a self-fulfilling prophecy and Americans are likely profiled more than in other cities.
  • Montevideo, Uruguay: the capital of a well-run, peaceful country, Montevideo is seeing an influx of talent from India of all places. Downside: costs are already spiking, and a small market.

After surveying the options, I settled on the engineering market in Buenos Aires. While Sao Paulo may have more raw potential, I wasn't ready to learn Portuguese as a prerequisite to validate my thesis.

Searching for a business model

“A startup is a temporary organization formed to search for a repeatable and scalable business model.” - Steve Blank

Armed with seed funding, I returned to Buenos Aires and dove head first into finding the business model.

My initial hypothesis was a staffing firm that sourced talent by teaching English. This combined a few of my beliefs:

  • Staffing firms generally focus too much on transactions and not enough on relationships
  • Teaching spoken English to an experienced engineer is faster than teaching effective engineering to a native English speaker
  • English schools and apps struggle because they lack differentiation. Teaching English to engineers has the dual benefit of a smaller scope (specific vocabulary) and a higher value add ($1M in their career)

I had received feedback from hiring managers that having supply would get their attention, so I started by building the supply side rather than the demand side of the staffing business.

With English as the bottleneck, I looked for solutions to rapidly teach English to engineers. Why do so few engineers speak English? I found these core reasons:

  • Many engineers learn to read, write, and listen to English, but they struggle with speaking. The first three can be done alone; the last one requires a partner
  • Learning to speak English is like dancing: you don’t need to get it perfect, you just need to get it correct enough to keep going. And you can’t learn it on Youtube
  • Certification programs like Cambridge focus on written English more than pronunciation
  • There are 200,000 words in English, but you probably only need 5,000 to be considered fluent for your job. The puzzle then becomes: which 5,000 words?
  • With so much information available online, the bottleneck to learn English is not education, but motivation

So I started solving these problems:

  • I partnered with an existing AI app that gives real-time feedback on your spoken English
  • I collected data sets of domain specific vocabulary to create custom courses around niche jobs, like MERN engineers who use Typescript
  • I took over an English-speaking meetup to create places where engineers could meet and practice speaking together
  • I set up a basic accountability system for students, where we would check on their progress and intervene when they plateaued

Within just a few months, we had students using the app and making progress in their spoken English.

Having fully explored the supply side, it was now time to turn my attention to the demand side and what opportunities I could uncover there.

Oops that’s not a business model

In focusing first on the supply side, I made a critical error: I looked for opportunities to add the most value, rather than opportunities to extract value.

Helping engineers learn English is by far the most effective way to add value. Some students literally need just a small nudge to get back on track learning English. Others just need another friend who is trying to learn the same thing. Others just want a course that feels relevant to their interests.

These small interventions will help some engineers earn $1 Million more in their career.

In focusing on this, however, I failed to find a clear pathway to make money. Engineers don’t make for great buyers: consumers generally don’t invest a lot into their education, engineers hesitate even more when it comes to buying technology, and Argentineans have the luxury of abundant high quality public education which makes paying for it particularly suspect.

I intended to find the business model on the demand side, finding employers for the engineers and getting a standard 10-20% fee. Unfortunately, while I was developing the project, the market went from “Quiet Quitting” to “Quiet layoffs”.

Most of Big Tech stopped "redundant hires" and started looking more carefully at their current staff. Today, even as the US economy declares a tight labor market, and parts of the market are picking up, much of the economy is still in a "vibecession".

Even in a hot market, I would have struggled to find clients, as it meant literally building two businesses at once. But I didn't need a great business: I just needed place about two engineers per quarter to break even.

As I made these changes and refocused my efforts, I started to see the repurcussions of broader economic factors.

The 1% of China Fallacy

A common error for startups is to say “If we just get 1% of the Chinese market, we’ll reach $x Billions.” The error here is in thinking the floor of your business is a number that rounds to 1%. In practice, the floor is always a number that rounds to zero percent.

I avoided the mistake of thinking I would be guaranteed 1 percent of the market, but I failed to notice the second part, that by default your market rounds to zero.

In thinking "I just need to place two engineers per quarter" I failed to notice that placing one is nearly as difficult as placing ten.


To be fair, I made some of these choices expecting more funding which never came.

When it became clear that I would need to reach cash flow without additional funding, I simplified the model to something I better understood: we would focus on local software agencies who need to either find English-speaking talent or get more US clients.

At first glance, this looks like a major pivot, but it actually simplified things:

  • We went from serving two audiences to mostly serving one
  • We went from having a B2B2C model to just a B2B model
  • We went from supporting two geographical markets to just one
  • I went from building a business I don’t know well (staffing) to one I understand very well (agency marketing)

Although the margins would be lower, this looked like it would solve the need for cash flow: my initial outreach consistently averaged over 10% reply rates, and I was getting regular meetings with business owners in the city. I signed one agreement to source talent for an agency, and had a half dozen offers for commission-based referral agreements.

Then the election season started.

Milei, Inflation, and the future of Argentina

Just when I thought I had figured out a path forward, a significant political event shook the foundation of my plans.

The elections in Argentina are different from the US: everyone is expected to vote, and every party finalizes one candidate after the primary results.

On August 13, Argentina held the primary elections, and the clear leader to emerge was Javier Milei.

Milei is a big unknown: the press wants to portray him as an authoritarian who wants to legalize firearms and let you sell your kidney to pay for them. In my informal polling, few expected Milei to win the primary.

The very next day, the government reacted by increasing the interest rate from 97% to 118%, and the official currency devalued by 18 percent. Unofficially, the value of the peso experienced such large fluctuations that no one could agree on which website to use as reference: it was anywhere from 700 to 790 pesos to the dollar, after being around 600 only weeks before.

For context: when I arrived last October, I was getting close to 300 pesos to the dollar. The peso had lost half of its value since October, but then seemingly overnight it lost thirty percent of its value.

Between the shock election and the shock devaluation, the market froze. Prospects started canceling meetings and ghosting, which was not standard before the election. Although these agencies needed US clients more than ever, they were stuck in the paradox of location arbitrage: with so much of their business expensed in Pesos, they were unwilling or unable to spend anything close to US market rates, even with the higher upside.

Even if they wanted to spend to get US clients, they now had to manage the stress of their entire team handling the currency devaluation and the ambient fears around what comes next.

Shutting Down

Against this backdrop, I had to decide whether to continue or not. When I stopped taking a salary to keep operations going, I set a milestone for when I would shut everything down. I reached that milestone, ignored it, moved the goal posts, reached those, etc. until one day I had already passed clear indicators to shut down.

Still I kept going.

I had to return to the US for some family health issues, and almost as soon as I landed, I had an eye infection that prevented me from grinding at the computer everyday. Suddenly I had endless time away from the screen to reflect on my situation.

Faced with the realization that I had already compromised my commitments, to myself and others, I concluded it was time to let go.

Lessons Learned

“Look for the thing that this market is especially good at.” - Sean Murphy

When I shared the project with Sean in January, he had this great advice. In Argentina, I believe the answer is accounting.

Imagine growing up with prices so uncertain that one of your earliest childhood memories is your mother yelling across the grocery store to get foods before the prices go up.

Imagine having 17 currency exchange rates, depending on what you are buying or selling.

In the US, accounting is a profession; in Argentina, accounting is a daily reality. With so many exchange rates, and so much volatility and inflation, keeping track of expenses and prices is so innate to the culture that porteños don't even see it.

As a profession, Argentine accountants are also more equipped: where accounting in the US has become a lot of rule following, accounting principles in Argentina have far more room for judgment, attracting more intelligent, creative people than you would expect.

Other insights:

  • Hiring devs offshore is difficult because you can't validate their skill level. Accountants have a baseline exam that makes it easier to test
  • While devs are often hired alone or sporadically, accountants are often hired in cohorts based on a predictable regulated calendar. This makes the business far more robust
  • While junior devs are oversupplied due to bootcamps, accounting in the US actually had a shortage last year of people taking the exams. This shortage is so great that some states are considering a change in the education requirements

If I could start over, I would have focused on solving for the CPA market. I’m not sure I would have enough personal interest to scale this business, but it would at least have a shot at reaching positive cash flow.

“If you don’t make money, you have a hobby not a business.” -IRS

I’m still interested in helping agency owners and talented engineers learn English and get work from the US - but it doesn’t have to be a business.

I’ve actually mentored people into the global job market for a long time, and I’ll continue to do this. For business owners, I’ve created a lot of helpful resources to help them approach marketing their business better, and I hope to find work that will enable me to continue doing this.

However, a hobby doesn't necessarily have to be a business.

I have clear areas of experience where my skills can earn high hourly rates while working remotely, and I’m ultimately in a better position to help others if I earn US hourly rates, and can help others without the need to earn as much.

What’s next

I’m honestly not sure what comes next. Three possibilities stand out:

Case Study Development

In 2020 I created a novel process to predictably generate more case studies and referrals. I wanted to build the SaaS for this, but clients really wanted me to offer the service: it turns out I'm exceptionally good at interviewing executives and uncovering their stories.

I actually took a job to better understand the buying process, and then lost sight of the goal as I focused on what the job needed.

Recently I've returned to this, and found that I have more to say about this space than anyone else: for example, there are actually distinct case study formats that engage different personality types.

I'm planning to produce 100 case studies (14 so far) and welcome any clients who want to explore this.

MarTech Product

One thing I sacrificed in the past year is writing code. I completed the technical portion of Hack Reactor last year, and since then, I snuck in app development whenever I could. Often I would have to make a conscious choice to not write code, realizing that it was more fun but it didn't have a clear pathway to revenue.

In just the last six months, I build a Vercel app with GPT3, refactored it to support GPT4, and dusted off a 2013 Swift project to reproduce it with React Native. I had a lot of fun doing these and would love to get paid to create more.

In particular I would be very interested to work on a MarTech app: I have enough experience in MarTech to question the feature requests and push for better product, but I'm new enough to engineering culture to accept a company's internal processes and best practices.

MarTech Agency

I’ve recently uncovered some novel patterns to exploit in scaling a MarTech agency, and I’d be interested in a COO or exec role to help an existing agency grow their business. In particular I'm very interested to help an agency build a best in class offshore team for both growth and delivery.

The ideal work will get me into flow, with skills where I'm clearly useful, and I have some flexibility in location. If you have ideas or suggestions, please feel free to reach out.

In reflecting on the lessons from this journey, the biggest lesson is the reminder that going it alone isn't always the best route. As the next chapter of my career beckons, I'm most excited to start building with others again.