Key takeaways
- Accelerators select on founders and rate of progress: Techstars states its priority order as "team, team, team, market, progress, idea." YC's Paul Graham wrote that he cares more about the founders than the idea, because most funded startups change direction significantly.
- The evaluation inverts the VC framework: VCs underwrite market size and traction metrics. Accelerators underwrite the people who will handle the pivot they expect to happen, at a stage where traction is often legitimately zero.
- YC accepts roughly 1-2% of applicants per batch: With 150-200 companies selected from 20,000-30,000 applications, and an estimated 30-40% of accepted founders had at least one prior rejection.
- The 10-minute YC interview is a stress test, not a pitch: Partners ask the same handful of questions ("What do you do?" "What have you built?" "How do you know people want this?") and judge how founders think under pressure, not how well they rehearsed.
Founders preparing for accelerator applications study the wrong material. They read pitch deck guides and investor criteria posts, then walk into a YC interview ready to defend their market size. The interview does not ask about market size. It asks "what have you built?" and "how do you know people want this?" and moves on in under ten minutes.
Accelerators and VCs are scoring the same application on completely different rubrics. If you prepare for one using the other's criteria, you are sitting the wrong test. This post covers the one accelerators actually give.
Will your idea survive the market?
Preuve AI runs 10 agents against live market data and links every claim to a source. Free analysis in 60 seconds.
How do accelerators evaluate startups?
Accelerators evaluate startups primarily on the founders and their rate of progress, not on traction metrics or market evidence. This inverts what most founders expect, because most fundraising advice is written for VC rounds where market size and revenue growth lead the conversation.
Techstars publishes its priority order on its help center: "team, team, team, market, progress, idea, in that order." Their program directors use that literal hierarchy when screening thousands of applications per batch.
Paul Graham, YC co-founder, writing on YC's own site, put it plainly: "I care more about the founders than the idea, because most of the startups we fund will change their idea significantly." He added that if a group of founders seemed impressive enough, he would fund them with no idea at all.
The logic is structural. Accelerators invest at a stage where the idea will almost certainly change. A 12-week batch is built to force pivots. So they underwrite the founders who will handle that change rather than the business plan sitting in front of them.
Here is the criteria table that makes the difference concrete, the thing I could not find anywhere else online: each row scored by how an accelerator reads it versus how a VC reads the same signal.
| What is scored | How an accelerator reads it | How a VC reads it | What you can show at idea stage |
|---|---|---|---|
| Founding team | Primary filter. Speed of building, clarity of thought, domain insight, co-founder relationship | Important but secondary to market and traction at seed+ | One specific impressive thing each founder has built or achieved, described in one sentence |
| Traction | Growth rate matters, absolute numbers do not. Pre-launch is fine | Revenue, retention, unit economics. Absolute numbers matter | A working prototype, 5-10 committed beta users, week-over-week usage growth |
| Market size | "If this works, what does it become?" A narrow wedge into a large market is fine | Bottom-up TAM/SAM/SOM with sources, specific enough to model returns | A one-sentence expansion path, not a polished TAM slide |
| Idea quality | Ranked low. Evidence of insight matters more than the idea itself. Ideas change | Ranked higher. The business model must be defensible and underwritable | A non-obvious insight about the problem, described in a sentence a stranger understands |
| Competitive landscape | "Who are your competitors and why do you win?" Naming none is a red flag | Detailed competitive positioning, moats, switching costs, market share data | Name 3-5 alternatives, explain your specific wedge in one sentence each |
| Demand evidence | "How do you know people want this?" Conversations, waitlists, organic pull | Paying customers, retention curves, conversion metrics | Specific user conversations, letters of intent, a waitlist with conversion data |
The last column is the one that matters for founders at idea stage. You do not need a polished pitch deck or a revenue chart. You need specific answers to specific questions.
What does YC actually look for?
YC evaluates a small number of things intensely. The application form and the 10-minute interview both surface the same signals: whether the founders can build, whether people want the product, and whether the market could get big.
The application's most important question, according to Paul Graham's published essay, is: "Please tell us in one or two sentences about something impressive that each founder has built or achieved." It is deliberately open-ended: "It could be that you did really well in school, or that you wrote a highly-regarded piece of software, or that you paid your own way through college after leaving home at 16."
The achievement's size matters more than its category. Generic claims like "Jordan is an exceptionally dedicated person" carry zero weight. One specific accomplishment beats a paragraph of adjectives about your character.
Garry Tan, YC's president and CEO, described the pattern he looks for in a 2024 TechCrunch interview: founders who come in with "some new discovery that they had discovered interacting with the technology itself." He used Brian Armstrong (Coinbase) as an example, someone who read the Bitcoin whitepaper while working as an anti-fraud engineer at Airbnb.
The interview is 10 minutes and feels adversarial on purpose. Partners interrupt, challenge assumptions, and compress months of due diligence into a stress test. They are watching how you think in real time, not how well you memorized a script. Common question territory, drawn from published accounts:
- "What do you do?" Answer in one clear sentence with no jargon. Fumbling this loses the room in the first 30 seconds.
- "Why you? Why now?" Your insight and the timing shift that makes this the moment.
- "What are your numbers?" Users, revenue, growth rate, retention. Know them cold.
- "Who are your competitors and why do you win?" Naming zero competitors is a red flag. Dismissing all of them is worse.
- "What have you learned from users?" Specific, surprising insights from real conversations, not survey data.
Partners are testing whether you answer quickly and honestly. Saying "I don't know" when you genuinely do not know lands better than bluffing, and they can tell the difference in about three seconds. What gets founders through is holding their ground with evidence when challenged.
Skip weeks of manual research
Get complete market research, sourced proof, competitor map, and pricing data for your idea instantly.
How is Techstars selection different from YC?
YC and Techstars select for different founder traits because they run fundamentally different programs. YC is batch-led: one centralized program, 150-200 companies per batch, weekly group office hours with partners. Techstars is mentor-led: dozens of programs worldwide, each run by a Managing Director who curates a batch of roughly 10-12 companies, with intensive one-on-one mentor matching during the first month.
That structural difference shapes what each program looks for. YC's published criteria prioritize velocity and first-principles thinking. Techstars, per its help center, lists "team, team, team" first and then looks for founders who will benefit from its mentor network, people who are coachable and have rounded out their team with the right skills.
| Dimension | YC (batch-led) | Techstars (mentor-led) |
|---|---|---|
| Batch size | 150-200 companies | 10-12 companies per program |
| Selection filter | Velocity, clarity of thought, first-principles insight | Team completeness, coachability, mentor fit |
| Investment | $500K standard deal (7% equity) | $120K ($20K + $100K convertible note, 6%) |
| Mentorship model | Weekly group office hours with YC partners | "Mentor Madness" month, then 2-3 dedicated lead mentors |
| Application process | Written application, then 10-minute partner interview | Multiple conversations with Managing Directors and mentors |
If you are applying to YC, optimize for speed of explanation and evidence of building. If you are applying to Techstars, optimize for team completeness and demonstrate that you will absorb mentorship. One application will not satisfy both filters. Write two.
Do accelerators care about traction?
Yes, but they define it differently than VCs do. A VC evaluating traction wants revenue, retention curves, and unit economics. Accelerators care about growth rate on any meaningful metric, and "meaningful" includes pre-revenue signals like weekly active users or waitlist conversions.
A startup doing $500 a week in revenue and growing 20% week-over-week is more interesting to a YC partner than one flat at $50,000 a month, according to published analysis of YC criteria. The first shows a market pulling. The second shows a plateau.
Both YC and Techstars accept pre-launch companies. If you have not launched, the substitutes are a working prototype plus a handful of committed users, letters of intent, or a waitlist with conversion data. The question is not "how big are your numbers" but "are your numbers moving, and do you know why."
If you are preparing for pre-seed investors, you need sourced market sizing, competitive analysis, and a go-to-market plan. If you are applying to an accelerator, you need to show that you have talked to potential users, that something is moving, and that you understand why it is moving.
I built Preuve to answer the question accelerators actually ask, "how do you know people want this?", with sourced evidence instead of a guess. A free scan pulls from 60+ live data sources in about 60 seconds.
When I studied the anonymized data from 6,000+ ideas, the ideas that score highest tend to be the ones where the founder can point to specific demand signals and name who the first buyers are, regardless of projected market size. That is exactly the signal an accelerator partner is listening for.
What happens if you get rejected?
Rejection from a program that accepts 1-2% of applicants is the default outcome. YC receives approximately 20,000-30,000 applications per batch and accepts 150-200, according to Garry Tan's 2024 public remarks.
An estimated 30-40% of accepted YC founders had at least one prior rejection, according to YCInsight's compiled data. YC partners explicitly encourage reapplication and track the delta between submissions. The median reapplicant applies twice. The maximum publicly documented is 4-5 attempts with no penalty for multiple applications.
Named examples: Brian Chesky (Airbnb) was rejected before being accepted. Drew Houston (Dropbox) was told to find a co-founder, teamed up with Arash Ferdowsi, and reapplied. Even YC alumni starting second companies report rejections on follow-up applications, per the same source.
What changes between rejection and acceptance clusters around five things, per YCInsight:
- Launched and got first paying users
- Added a technical co-founder
- Pivoted to a more specific problem
- Showed week-over-week growth on any meaningful metric
- Reframed the application around what they had actually built, not what they planned to build
Every item on that list is about demonstrating progress. That maps directly to the evaluation criteria in the table above: accelerators select on rate of progress, so the way to turn a rejection around is to make visible progress and reapply.
A rejection does not close the VC path either. Most investors outside YC do not ask whether you applied, and the ones who do care about your traction, not your YC status. I covered what investors actually check in diligence in a separate post. A rejected accelerator application and a weak investor pitch are different failures. Figure out which one you actually had, and fix that.
FAQ
How do accelerators evaluate startups differently from VCs?
Accelerators evaluate startups primarily on the founding team and rate of progress, while VCs evaluate primarily on market size, traction metrics, and unit economics. Accelerators invest at a stage where the idea will likely change, so they underwrite the founders who will navigate that change. VCs invest at a stage where the business model should already work, so they underwrite the evidence that it does. Techstars publishes its priority order as "team, team, team, market, progress, idea." A typical VC due diligence process leads with market size, revenue growth, and retention metrics.
What does Y Combinator actually look for in applications?
YC looks for founders who can explain what they are building in one clear sentence, have built something impressive before (not necessarily a startup), show domain insight into the problem, and think clearly under pressure. Paul Graham wrote on YC's site that he cares more about the founders than the idea, because most YC-funded companies change their idea. The application's most important question asks founders to describe "something impressive that each founder has built or achieved," and the best answers are specific, not generic.
Do you need traction to get into an accelerator?
No. Accelerators accept pre-launch companies. YC and Techstars both accept startups with zero revenue and no launched product. What matters is evidence of progress: a working prototype, a handful of committed users or letters of intent, or week-over-week growth on any meaningful metric. Growth rate matters more than absolute numbers. A startup doing $500/week growing 20% week-over-week is more interesting to YC than one flat at $50,000/month.
Should you reapply to YC after being rejected?
Yes. An estimated 30-40% of accepted YC founders had at least one prior rejection. YC partners explicitly encourage reapplication and track the delta between submissions. Airbnb and Dropbox are among the companies whose founders were rejected before being accepted. The strongest reapplications show concrete progress since the last attempt: launched product, first paying users, added a co-founder, or pivoted to a more specific problem.
What is the YC acceptance rate?
YC accepts roughly 1-2% of applicants. Recent batches accepted 150-200 companies from approximately 20,000-30,000 applications. YC now runs four batches per year (Winter, Spring, Summer, Fall), each roughly half the size of the older two-batch format. The Summer 2025 batch had an acceptance rate reported at approximately 0.6%.
Vincent
5 years in B2B growth, building Preuve AI in public. 82% of ideas it scores aren't ready, the point is finding out in 8 minutes, not 3 months.
Follow on X →
Building is expensive. Validation is free.
Run your idea through 10 AI agents before you write a line of code. Every claim source-linked.
