How to prepare your Y Combinator or Techstars application

What accelerators actually screen for

Accelerators read thousands of applications, so reviewers move fast and look for a few signals. The first is the team. They want to know why you and your co-founders are the right people to build this, whether you work well together, and how quickly you learn. The second is the market. They want a problem that many people or businesses feel, and a sense that the market could grow large. The third is demand. They look for proof that someone wants what you are building, even in a rough early form. A first-time founder often assumes the idea itself is what gets judged. It is not. Reviewers assume most ideas will change. What they are really betting on is whether you understand your customer better than anyone else, and whether you can show early signs that the market agrees. If you can explain the problem in plain words, name exactly who has it, and point to real people who have used or paid for your product, you already stand apart from most of the pile.

You do not have to guess at the weighting, because both programs publish it.

Techstars states its order plainly in its own help center: team, team, team, market, progress, idea, in that order. The same page tells applicants to round out the team with the skills the company needs, to make progress on a prototype and reference it, and to show early traction or data points that back up the core assumptions.

Y Combinator publishes the stage mix instead. Its FAQ says that on average 40 percent of the companies it funds in each batch are just an idea, and that most have no revenue. The same page says the founding team should have the skills to build the product themselves rather than outsourcing it, and that nobody needs an introduction, because all applications are considered equally.

Two things follow, and they change how you spend the weeks before a deadline.

• Early is not the problem. Vague is. A company with no revenue and a precise account of who it spoke to and what it heard reads better than one with a little revenue and no account of where the revenue came from.
• The team question is about capability, not credentials. The question is whether this group can build and sell this thing without buying in the ability to do it. A long resume that does not answer that has not answered it.

The parts of the application that win or lose it

Most accelerator applications ask short, direct questions, and each one is a chance to earn or lose the reviewer's attention. The founder question matters most. Say clearly why you are building this and what you know that others do not. Skip the buzzwords and tell the truth. The what do you do question should be answerable in one plain sentence that a stranger would understand. If it takes a paragraph, keep cutting. The progress question is where many applications quietly fail. Vague claims like strong interest mean nothing. Specific facts, such as the number of users, paying customers, or signed pilots, carry real weight. The one-minute video, when a program asks for one, is not about production quality. Reviewers want to see how you and your co-founders speak, think, and handle a plain question. Answer each prompt directly and resist the urge to sound impressive. A short, honest answer with a concrete number beats a long, polished one with nothing behind it. Reviewers have seen every clever phrase, so clarity and evidence are what stand out.

The four questions that are harder than they look

Almost every program asks some version of these, and they are where good companies write bad applications.

1. What are you working on? The test is whether a stranger outside your industry can repeat it back. Y Combinator says this is typically the first question at interview, and that founders find it harder to answer clearly than they expect. Write the sentence, say it to someone who does not work with you, then cut it again.
2. What do you understand about this market that most people in it do not? The one question where being ahead of the reader is the point. A good answer is a specific mechanism or behavior, learned from someone real. If it would fit in a trade publication headline, it is not yours.
3. Why now? Either something changed, or this was possible five years ago and somebody would have done it. Name the change: a cost that fell, a rule that shifted, a habit that moved.
4. What is the biggest thing standing in your way? Programs read the evasions closely here. A candid answer with a plan attached reads as a founder who has looked at the problem. A dismissal reads as one who has not.

One mechanic worth knowing first. Y Combinator's FAQ says to submit one application per batch even with two ideas, and that editing stops at submission because review starts there. The draft you send is the draft that gets read, so write it where you can edit freely, give it to someone who will argue with it, and paste it in when it is done.

Why a real demand signal beats a polished deck

It is tempting to spend weeks perfecting a pitch deck before you apply. A clean deck helps, but it is not what convinces a reviewer. A deck is a set of claims. A demand signal is proof. Proof is a person who paid you, a company that signed a pilot, a waitlist that people actually joined, or a small group using your product every week. You do not need a finished product to gather this. You can build a simple landing page that explains the offer and ask people to sign up or pre-order. You can run a short pre-sale. You can interview twenty potential customers and record how many say they would switch today. Each of these gives you something specific to write in the application, and each one is far more persuasive than a slide claiming the market is large. The habit this builds also serves you long after the form is submitted. Founders who chase real demand early learn what their customers truly want, which means they waste less time building the wrong thing. So before you polish another slide, go get one honest piece of evidence that people want what you are making.

One more reason to keep the evidence exact rather than generous. Y Combinator's interview guide says that if you state numbers in the interview, it may ask you to verify them afterwards. A figure rounded up in a hurry in October is a figure you stand behind in November.

So keep one page beside the application: every number you cited, where it came from, and the date you pulled it. Active users, and what active means in your product. Revenue, and whether it recurs. Pilots, and whether money moved. Twenty minutes of work, and it is the page you will want ten minutes before the call.

A pre-application checklist you can finish this month

You can do most of this preparation on your own, without any special tool. Start with the problem. Write one plain sentence naming who has the problem and why it hurts. If you cannot, keep talking to customers until you can. Next, gather evidence of demand. Set a small, real target, such as ten paying customers, fifty waitlist signups from strangers, or three signed pilots, and go earn it. Then write your one-sentence description of the product and test it on someone outside your industry. If they understand it, keep it. Look up the specific program you want, since Y Combinator and Techstars ask different questions and value slightly different things, and read past accepted companies to learn their bar. Draft your answers early, then cut every sentence that does not add a fact. Record a rough practice video and watch it with a friend. Finally, line up the numbers you will cite so each claim is true and easy to check. If you finish this list, your application will rest on evidence rather than hope, which is the strongest position any first-time founder can be in.

The interview is a different test

The form and the interview measure different things, and preparing for one as though it were the other is a common way a strong application stalls.

The form is read in a few minutes by someone who has never met you, so it rewards precision and cutting. The interview is a conversation, and the programs are open about how short it is. Y Combinator publishes its format: ten minute conversations over video, two or three partners on the call, all founders present, and interviewers who have already read your application and have it open in front of them.

Read that last detail twice. Nothing in those ten minutes repeats the form. There is no presentation and no slot for one, and a rehearsed opening spends a minute you will want back.

The guidance Y Combinator gives is unusually direct, and worth reading in full rather than through anyone's summary of it. The short version: do not rehearse the interview, because heavy practice makes the conversation stiff. Practicing the application video is a separate job, since that one is a recording you get to redo. Make real progress between applying and interviewing instead. Know your users and your numbers, and have the numbers written down rather than in your head. Be ready to show a working version of what you built. Expect every founder to answer at least one question.

The register is what experienced founders most often get wrong. This is not an investor meeting where composure is the product. Both programs describe choosing people they will work with closely for months, so a candid answer to an awkward question does more for you than a smooth one, and "I do not know, and here is how I would find out" is an answer rather than a failure. If you have raised before, the other adjustment is length: ten minutes will not hold the three minute narrative you built for a seed round. Answer what was asked, stop, and let them ask the next one.

What a rejection contains, and how reapplying works

Most applicants are not accepted. The acceptance rate is not worth staring at, because you cannot move it. What you can do is understand what a no contains.

The two stages return different things. Y Combinator's FAQ says it does not give feedback on written applications, because of the volume, but that teams who reach an interview get detailed feedback afterwards either way. Its interview guide adds that it is common for teams to take that feedback, apply again the following batch, and be accepted. The FAQ puts the pattern plainly: in a typical batch, about half the companies applied more than once before they got in, and progress since the last application is described as a strong signal.

So reapplying is a normal route rather than a consolation prize. What it asks of you is easy to write and hard to do: something has to be different, because a rewritten answer is not different while a new fact is, such as twenty more customer conversations, a pilot that started, or a channel you tested and can report on.

A practical way to hold it: save a copy of your application the day you submit, because you cannot edit it afterwards and memory will smooth it. When the decision arrives, set that copy beside your current answers and mark every line where the underlying fact has changed. If nothing has changed after a full cycle, the problem was never the writing.

Whether an accelerator is the right choice for you

This is the question most application advice skips, and it deserves an hour before you spend a week on a form.

An accelerator is a trade, and both sides are published. Y Combinator's standard deal is $500,000: $125,000 converting into a fixed 7 percent, and $375,000 on an uncapped most favored nation SAFE, a simple agreement for future equity that converts later on the best terms you give another investor. Techstars publishes its own: $220,000 for most of its programs, of which $200,000 is an uncapped most favored nation SAFE and $20,000 buys 5 percent in common stock. Techstars calls the program three months of mentorship-driven support, and Y Combinator's batch runs in person in San Francisco.

The money is rarely the point. Y Combinator says at least half the startups it funds do not need it, and offers a test worth borrowing whichever program you are weighing: ask whether joining would improve your company's outcome by at least 7 percent.

Situations where the honest answer is often no, at least this cycle:

• Your constraint is one unknown, not a network. If what is blocking you is whether anyone will pay, three months of introductions will not settle it and a two week demand test might. Validating the business idea is the cheaper move first.
• You cannot commit full time, or cannot be where the program is. Both expect founders present and working on the company, and a plan to sort that out later comes apart at the point it matters.
• The timing belongs to the next cycle. A round closing now, a contract landing in four months, or a cofounder conversation you have not finished are all reasons the batch after this one may serve you better.
• You want it mainly for the signal. It is a real signal, and a thin reason on its own, because the work that earns a place is the work that makes the signal less necessary.

None of that argues against applying. Both programs tell early companies to apply rather than wait, and an application costs an evening. It argues for knowing what you are buying before a deadline makes the decision feel already made.

The work that cannot be done in the month before a deadline

The checklist above covers the weeks before a deadline. There is a second list underneath it, and it decides more applications, because none of it can be started in a month.

• Talked to enough people that you have stopped being surprised. Not ten conversations to fill an answer box. Enough that you can predict what the next person will say, and are occasionally wrong in a way that teaches you something. Market research for startups covers how to run them so the output is usable afterwards.
• Settled the cofounder question. Equity split, roles, vesting, and what happens if one of you leaves. A team still negotiating its own terms shows up in an interview inside two minutes, and the team is read first.
• Put something in front of a stranger. A working version, a landing page, or a service you deliver by hand. A fake door test takes a weekend and turns a paragraph of intention into a fact you can cite.
• Decided what you would do without the program. Every good application answers this without being asked, and founders who have decided sound different from founders who have not.
• Written your numbers down once, with sources. The pre-seed fundraising checklist is the same evidence an accelerator reviewer, an angel and a grant committee all want, so assembling it once serves all three.

Do those and the form takes an evening. Skip them and it takes three weeks and reads like three weeks of writing.

That groundwork is the part LaunchValid builds: research on your own market, a demand test you can run before a deadline, and a deck and memo written from the same evidence so the numbers agree. It changes what you have to show. It cannot change a decision that belongs entirely to the program, and nothing sold to you can.