How to Start a Business in High School

In March 2026, MyFitnessPal acquired Cal AI, a calorie tracking app built by two teenagers, after it hit 15 million downloads and $30 million in annual revenue in under two years. Zach Yadegari, the co-founder, was 19. He was also rejected by 15 of the 18 colleges he applied to, despite a 4.0 GPA and a company that a public corporation had just paid to own.

Most articles about student entrepreneurship would use Cal AI as a motivation story. This one uses it differently. Yadegari and his co-founder Henry Langmack made a specific set of decisions that most student builders don't make: about what to build, who to build it for, and how to structure the business underneath the product. Those decisions are why Cal AI got acquired. This article is about what those decisions were and how to apply them to whatever you're building.

Where are you right now?

This article is designed as a reference document. Find your stage and go to the sections that are useful now.

• "I have an idea but haven't started" — Sections 1, 2, 3
• "I've started but I'm not making money" — Sections 3, 4, 6
• "I'm making money but not sure the model is right" — Sections 3, 4, 5
• "I need legal and financial setup" — Section 5
• "I need to write about this for college applications" — Section 7

Is This a Business or a Project? A Diagnostic That Matters Before You Build

Most student ventures are projects. Projects are fine. They build skills, generate portfolio material, and teach execution. But if you want a business, you need to know the difference before you build, because the decisions are different from day one.

Three diagnostic questions to apply to your own idea before you write a line of code or pitch your first client:

• If you stopped working on it for two weeks, what happens? If revenue stops, you have a job. If the system keeps running, you have a business. Cal AI processed meal scans, charged subscriptions, and retained users while Yadegari attended school and slept. That's the test. A tutoring practice stops earning the moment the tutor stops showing up. The distinction is structural, and it determines how you build from the start.
• Does it get easier or harder as you add more users? A tutor with 10 students works roughly 10 times harder than a tutor with 1. A software product with 10,000 users does not require 10,000 times the work of a product with 1 user. If your model gets harder as it scales, you're building a job. If it stays flat, or gets easier through automation and systems, you're building something worth building.
• Is the value in you or in the system you built? A student whose clients leave when they leave has built personal brand equity, not a business. A student whose clients stay because the product or system is valuable has built something transferable. Transferable things get acquired. Personal brand equity doesn't.

Three categories, in order of leverage: time-for-money (job), productized service (some leverage), scalable product (business). Most students start in the first category. The goal is to understand which category you're in and make deliberate decisions about when and how to move.

How Do You Find a Problem Worth Solving? The Positioning Map

Generic solutions lose to incumbents every time. Incumbents have more money, more users, more data, and more brand recognition. A high school student cannot beat MyFitnessPal by building a better MyFitnessPal. Cal AI didn't try to.

The positioning map is a four-step exercise. Work through it for your own idea before committing to a build.

1. Step 1: Name the incumbent. For any idea you're considering, who already does this? If the answer is nobody, that's either a genuine gap or a sign that nobody wants it. Figure out which before proceeding.
2. Step 2: Map the tradeoffs the incumbent made. Every product makes tradeoffs. MyFitnessPal chose accuracy, comprehensiveness, and a database spanning 20 million foods. To make those choices, they gave up speed, simplicity, and frictionless entry. Write down what your incumbent chose — and what they gave up to make that choice.
3. Step 3: Find the underserved users. Who did the incumbent's tradeoffs leave behind? MyFitnessPal's choices left behind everyone who wanted to log a meal in five seconds and move on. That's a large, real, paying audience. Cal AI built for them.
4. Step 4: Write your positioning sentence. "[Product] is for [specific user] who wants [the tradeoff the incumbent didn't make]." Cal AI's version: "Cal AI is for people who want to track calories in under ten seconds without manual entry, unlike MyFitnessPal which optimizes for precision over speed."

Students who can write that sentence before they build have solved the hardest part of starting a business. Students who can't write it yet aren't ready to build.

The exercise works outside the app world too. Three examples:

• Tutoring. Incumbents like Wyzant and Tutor.com chose breadth: any subject, any level, any location. The underserved user: a student preparing specifically for IB Chemistry at a school whose teacher runs notoriously difficult Paper 2 questions. A tutor who builds materials around that teacher's style and past exam history makes the opposite tradeoff. Depth over breadth. That's defensible. Generic chemistry tutoring isn't.
• Social media management. Agency incumbents chose volume: many clients, standardized deliverables, monthly retainers requiring minimal local knowledge. The underserved user: a local restaurant owner who needs someone who knows the neighborhood, can show up for events, and communicates like a person rather than an account manager. A student with genuine local knowledge and a phone has an advantage no agency can replicate for that client.
• Event photography. Professional photographers chose polish: formal packages, high prices, equipment that signals seriousness. The underserved users are school clubs, student organizations, and youth sports teams that need usable photos quickly, at a price that fits a $200 budget, from someone who actually understands the context. A student photographer is not a worse version of a professional for this client. They're the right person for the job.

Your advantage is almost never being better than the incumbent on their own terms. It's being the right choice for a specific user the incumbent built the wrong product for.

What Business Model Are You Actually Building? The Leverage Spectrum

Cal AI's team of seven employees handled 15 million downloads. That ratio is the only number a student needs to internalize about leverage. According to the TechCrunch report on the MyFitnessPal acquisition (March 2026), the team remained at seven people even as the product scaled to $30 million in annual revenue. Seven people. $30 million. That is what a high-leverage business model looks like.

Four models on a leverage spectrum, with honest assessments of each:

• No leverage: time-for-money. Tutoring, lawn care, photography billed by the hour, design projects sold as custom work. Revenue scales linearly with time. Your ceiling is your schedule. A student running this model at 8 hours a week at $40/hour earns $320/week before expenses. That's real money and real experience, but there's no path to growth that doesn't require more of your time. This is a fine starting point and a dead end as a long-term model.
• Some leverage: productized services. Same deliverable, defined scope, fixed price, documented process. A student who offers "one college essay edit, one round of revisions, delivered in 48 hours, $75 flat" has productized their service. They're not reinventing the engagement every time. The process gets faster with repetition. Prices rise with demand without requiring more hours. This is the most underutilized model for students because it requires saying no to custom work, which feels like leaving money on the table. It isn't. It's how margin gets built.
• Real leverage: digital products. Build once, sell repeatedly. Marginal cost of one additional user approaches zero. Cal AI. A set of IB study guides sold on Gumroad. A Notion template for college application tracking. A Chrome extension that solves one specific problem. The ceiling here is distribution, not time. The challenge is that you have to build something people actually want, which is harder than it sounds and is why Section 2 comes before this one.
• Maximum leverage: platforms. Connect two sides of a transaction and take a cut. Highest ceiling, hardest to build, requires solving the chicken-and-egg problem of attracting both sides simultaneously. Not the right starting point for most students. Worth knowing exists.

How to move between categories. The path most students overlook: the tutor who builds curriculum. You tutor IB Chemistry for two years using materials you've developed from scratch. At some point you have 60 hours of custom notes, practice papers, and mark schemes. Those materials are valuable to other students who can't afford a tutor and to other tutors who need a head start. You package them and list them on Gumroad for $30. Now you have a digital product running alongside your service. You've moved from no leverage to some leverage without changing your core skill. The next move, the one that takes real work, is building the product to the point where it sells without the tutoring attached. Where students find it through search or word of mouth and buy it without knowing you exist. That transition, from productized service to scalable digital product, is the path Cal AI's founders walked before most people had heard of them.

Do the Unit Economics Work? Three Questions Before You Commit

These three questions tell you whether your idea is structurally sound before you spend a week building it. Answer all three honestly. If the math doesn't work, the section ends with what to do about it.

• Question 1: What does it cost to deliver this once? Include time at an honest hourly rate, materials, tools, platform fees, and client communication overhead. Client communication is the number most students ignore and the one that most frequently kills service business margins. A student who spends two hours on a design project and another hour on emails, revisions, and follow-up has a 3:1 time-to-delivery ratio. That hour of communication costs the same as an hour of work. Price it accordingly.
• Question 2: What's the marginal cost of one more unit? For a service business, it's your time, which is finite and non-renewable. For a digital product, it's compute and storage costs, which is near zero at small scale. Cal AI's marginal cost per additional user was a fraction of a cent in cloud compute. That's why 15 million downloads is achievable with a team of seven. If your marginal cost is your time, you're capped. Know that before you start, not after.
• Question 3: What's the ceiling on what you can charge? This is anchored to the value to the buyer, not the cost to you. A student who saves a local business owner four hours per week has created four hours of that owner's time. If the owner bills at $150/hour, the student created $600 of value per month. Charging $80/month for the automation is not aggressive pricing. It's leaving $520 on the table. Value-based pricing requires knowing the buyer's alternative: what would they pay someone else to do this, and what is their time actually worth?

When the numbers don't work, pull one of three levers:

Raise prices first. Most students underprice because they're anchoring to their own discomfort, not the buyer's math. Test a price 30% higher than your instinct before concluding the ceiling is where you think it is. The number of students who have discovered their service was worth more than they charged is far larger than the number who priced themselves out of a market.

Reduce delivery time second. If unit economics are tight because delivery takes too long, document the process step by step and find what can be templated, automated, or removed. Productizing a service forces this discipline. The student who can deliver in two hours what used to take four has doubled their effective margin without raising prices.

Change the model third. If you're in the time-for-money category and the ceiling is genuinely too low, working harder within the model won't fix it. Adding a fifth tutoring student to a practice that's already at capacity doesn't solve the structural problem. Moving up the leverage spectrum does.

When It's Not Working: Diagnosis Before Pivot

Every student business hits a wall. The most common mistake is pivoting too early. Many change the product or the idea before understanding why it isn't working. The second most common mistake is pushing too long on something that's structurally broken. The diagnostic below is designed to tell you which situation you're in.

1. The problem is distribution, not the product. 
   
   1. Sign: people who try the product like it, but not enough people are trying it. Retention is decent but acquisition is broken. 
   2. Response: don't change the product. Change how people find it. Pick one distribution channel, such as a subreddit, a teacher network, a Discord community, a specific TikTok niche and run it seriously for four weeks before evaluating. One channel mastered produces more than five channels managed poorly.
2. The problem is the product, not the positioning. 
   
   1. Sign: people try it and don't come back. Acquisition works but retention is broken. 
   2. Response: go back to direct conversation. Find three people who tried the product and left, and ask specifically what didn't work. Don't guess. Don't survey. Call or message them and ask. The answer is almost always in those conversations and it's almost never what you expected. Students who skip this step spend weeks building the wrong fix.
3. The problem is the model, not the product or the distribution. 
   
   1. Sign: people use it and like it but won't pay, or they pay once and don't renew. 
   2. Response: the pricing, packaging, or monetization structure is wrong, not the product. Test a different price point, a different structure (monthly versus annual versus one-time purchase), or a different buyer entirely. Selling a study tool to students is harder than selling it to parents. Selling an automation to individual business owners is harder than selling it to a franchise that can deploy it across 20 locations.

When to actually pivot. You've run all three diagnostics, made specific changes based on real user conversations, and the numbers haven't moved in 60 days. At that point the issue is the original positioning, or the wrong tradeoff for the wrong user. Go back to Section 2 and run the positioning map again with what you've learned. A pivot informed by six weeks of real data is a decision. A pivot driven by frustration after two weeks is a guess.

Cal AI's founders built multiple products before the calorie tracker found its footing. The acquisition wasn't the result of a single brilliant idea. It was the result of accumulated judgment about what users actually wanted; judgment that came from earlier attempts that didn't work.

Legal and Financial Setup: The Honest Minimum

Most articles on student entrepreneurship tell students to incorporate early. That advice is wrong for most students. It exists because it sounds serious, not because it's useful at the idea stage. Here is what you actually need to know.

The default: sole proprietor until a specific trigger applies. No registration required in the US, UK, Canada, Singapore, or UAE for a sole proprietor operating under their own name. No filing fees. No overhead. You are the business. Your income is your income. You can open a business bank account, accept payments, and sign contracts as an individual. This is the right structure for the large majority of students who are testing an idea.

The four triggers that change the default:

1. A business partner. Without an entity, there's no legal structure governing what happens if the partnership breaks down. Who owns what, who decides what, and what happens if one person wants to leave; none of that is answered without an operating agreement inside an LLC. This trigger applies the moment you bring on a co-founder with meaningful equity in the outcome. It's the most important trigger and the most commonly ignored one.
2. A client that writes checks to businesses rather than people. School districts, companies, and institutions require a vendor entity for their accounts payable systems. When a client tells you they can only pay a business entity, form an LLC in your home state. It takes one afternoon and costs between $50 and $200 depending on the state. That's the right response to that specific trigger. Not a C corp. Not Delaware. An LLC in your home state.
3. Real liability exposure. A health app, a food product, a physical service with injury risk, or anything else where something going wrong could result in a claim against you personally. An LLC creates a liability shield between the business and your personal assets. For most student software products, this risk is theoretical. For anything touching health, safety, food, or physical activity, it's real enough to warrant the filing.
4. Raising outside investment. If you're taking money from investors, you need an entity. Delaware C corp is the correct structure in that case because it's what investors expect and what lawyers know how to work with. This applies to a very small number of students. If it applies to you, you already know it.

On taxes. Any student earning more than $400 in a calendar year in the US owes self-employment tax on that income. Track every transaction from day one: date, description, amount received, amount spent. A Google Sheet is sufficient. Set aside 25 to 30 percent of every payment before you spend any of it. Do not wait until April to think about this — the money is easier to set aside when it arrives than to find six months later. Involve a parent early, not for permission, but because payment platforms and bank accounts require adults for users under 18, and the conversation about tax infrastructure is easier before the revenue exists than after.

Delaware C Corp. The right structure for three situations: raising venture capital, issuing equity to employees with vesting schedules, and planning a future institutional fundraise. Wrong for everything else. Cal AI was certainly incorporated by the time they raised any money or structured the acquisition. At $0 in revenue, they weren't. Don't incorporate based on aspiration. Incorporate based on a specific trigger.

What Your Business Actually Signals on a College Application

Yadegari was rejected by 15 of 18 colleges despite a 4.0 GPA and a company doing $30 million in annual revenue that was subsequently acquired by a publicly traded company. According to TechCrunch's reporting (March 2026), he had not initially planned to attend college at all and reconsidered only after a summer at a hacker house surrounded by successful founders made him value the optionality of a degree.

Revenue does not guarantee admission. Understanding why matters; this is not to discourage students from building, but to ensure they document and frame their work in a way that makes it legible to admissions readers.

What admissions officers evaluate, in order of weight:

1. Specificity of problem identification. Can you articulate the exact problem you noticed, why existing solutions didn't address it, and what made you the person to address it? This quality separates a business from a project in an admissions reader's evaluation. "I noticed that every calorie tracking app assumed users were cooking at home or eating at restaurants, but students eating school cafeteria food had no reliable option" is specific. "I built a health app because I'm interested in fitness technology" is not. The first sentence demonstrates that you understand markets. The second demonstrates that you had an idea.
2. Evidence of real users and real iteration. Numbers matter, but the narrative around numbers matters more. A student with 200 weekly active users who can explain what changed after the first 50 users, what the retention data told them, and what they would build next demonstrates a feedback loop. A student with 10,000 downloads and no insight into user behavior demonstrates that they found a distribution channel. Those are different things.
3. Demonstrated learning from failure or iteration. Every business breaks something before it works. The student who describes specifically what broke, what they tried, and what they learned from fixing it is demonstrating intellectual honesty and adaptability: two qualities that college essays are explicitly designed to surface. The student who presents only the success story leaves the most compelling material unused.

The three containers for a college application:

1. The activities section holds one line. Use it for the most impressive quantifiable outcome: "Founded [product name], [X] weekly active users, [$X] revenue."
2. The additional information section is where the business model, the problem, and the traction narrative belong. This is the place for the positioning story — why this problem, why this solution, what you built, what you learned. Two to three paragraphs. This section is underused by most students and overloaded by the ones who use it.
3. The essay is where one specific moment becomes the story. Not "I built a business." Interesting topics can include a specific conversation with a user that changed how you understood the product, a specific failure and what it cost you to recover from it, or a specific decision that turned out to be right for a reason you didn't anticipate. The business is the context, but the moment of judgment is the essay.

What to document from day one. Keep a running document, updated monthly, with three things: the original problem statement in the words a real user used to describe it to you; every significant product decision and what prompted it; and the metrics that told you whether each decision was working. That document is the college application, the competition submission, and the interview answer. Students who build it as they go have options. Students who try to reconstruct it six months before applications are due face a much harder task.

Nova Scholar's Entrepreneurship Studio pairs students with experienced mentors who help build this documentation structure from the start of a project connecting the business-building process directly to the college application narrative.

Frequently Asked Questions About Starting a Tech Business in High School

Do I need a business plan before I start? No. A business plan is useful for raising money and for organizing your thinking once a business is established. At the idea stage, the positioning map in Section 2 and the unit economics questions in Section 4 replace it. Answer those questions first. They take two hours, not two weeks, and they surface the problems a business plan would bury under formatting.

Can I run a business during the college application year without hurting my application? Yes, with one condition: the business needs to be running on a system rather than on your constant attention. A business that requires 20 hours per week during application season damages both the business and the application. A business running on three to five hours of maintenance, because you built the system before senior year, is an asset on an application, not a liability. Build the system before junior year ends.

What's the difference between a startup and a small business for a high school student? A startup is designed to scale quickly to a large market with minimal marginal cost per additional user. Cal AI is a startup. A small business is designed to serve a defined market profitably and sustainably at a manageable scale. A tutoring practice or a photography service is a small business. Neither is better. They require different decisions from the beginning and signal different qualities to different audiences. Know which one you're building before you start, because the decisions diverge early.

Should I skip college and focus on my business instead? Yadegari considered it and ultimately chose college after a summer surrounded by successful founders convinced him that the optionality of a degree outweighed the short-term opportunity cost. The honest version: if your business is generating $30 million in revenue and has been acquired by a public company, the calculation is genuinely complex. If it's generating $5,000, the calculation is not. Know which situation you're actually in before making a decision that's difficult to reverse.

How do I handle taxes as a minor running a business in the US? Track every transaction from day one. Set aside 25 to 30 percent of every payment before spending any of it. File as a sole proprietor using Schedule C on your personal tax return. Involve a parent early because payment infrastructure (Venmo, Stripe, PayPal, business bank accounts) requires adult authorization for users under 18. If you're earning consistently above $10,000 annually, a one-hour conversation with a CPA will save more than it costs. Do that before April of the tax year, not after.

‍