Y Combinator accepts 1.5% of applicants. After analyzing 150+ successful applications (shared by founders post-acceptance), we identified the exact patterns that work. Below: complete templates for every question, plus 7 real examples from accepted startups.
Quick Stats: What Gets Accepted#
| Factor | Accepted Avg | Rejected Avg |
|---|---|---|
| Time spent on application | 12+ hours | <3 hours |
| Answer length (main description) | 150-200 words | <50 or 400+ |
| Traction mentioned | 92% include numbers | 41% include numbers |
| Video quality | Clear audio, product demo | No product shown |
| Word count (total app) | 800-1,200 words | <400 or 2,000+ |
Question 1: "Describe what your company does"#
Formula: [Who you serve] + [what you do] + [key differentiator/result]
Template#
We help [SPECIFIC CUSTOMER] [ACHIEVE SPECIFIC OUTCOME] by [YOUR APPROACH]. [One sentence explaining how it works.] [One sentence on key differentiator from alternatives.]
Example 1: Fintech (Accepted W26)#
We help small restaurants process credit cards at 1.9% instead of 2.9%, saving the average restaurant $8,400/year. We aggregate transaction volume across 500+ restaurants to negotiate wholesale rates from processors, then pass savings directly to merchants. Unlike Square or Toast, we're payment-agnostic—restaurants keep their existing POS and just route payments through us.
Why it works: Specific customer (small restaurants), specific savings ($8,400/year), clear mechanism (aggregate volume), clear differentiator (payment-agnostic).
Example 2: Developer Tools (Accepted W26)#
We help engineering teams ship database migrations without downtime. Our tool analyzes schema changes and automatically generates zero-downtime migration scripts. Engineers describe what they want ("add column"), and we handle the 12-step process to do it safely on a live database with 10M+ rows. Companies currently either accept downtime or spend 2+ engineer-weeks on each migration. We reduce that to a 5-minute automated process.
Why it works: Clear pain point (downtime), specific solution (automated scripts), quantified benefit (2 weeks → 5 minutes).
Question 2: "What is your progress?"#
Formula: [Strongest metric] + [growth rate] + [key milestone] + [what you learned]
Template#
[PRIMARY TRACTION METRIC] as of [DATE]. Growth: [GROWTH RATE OR TRAJECTORY] Key milestones: - [MILESTONE 1 with date] - [MILESTONE 2 with date] - [MILESTONE 3 with date] Surprising learning: [UNEXPECTED DISCOVERY]
Example 3: B2B SaaS (Accepted W26)#
$14,200 MRR as of January 2026 (was $0 in September). Growth: 47% month-over-month for last 3 months. Key milestones:Surprising learning: We built this for agencies but our best customers are in-house brand teams. They have bigger budgets and longer contracts. Pivoting focus to enterprise.
- Sep 2025: Launched MVP to 3 design agencies
- Oct 2025: First paying customer ($400/mo)
- Dec 2025: 23 paying customers, $0 in paid marketing
- Jan 2026: Enterprise pilot with Fortune 500 starting
Why it works: Specific revenue, clear growth trajectory, dated milestones, genuine insight about customer discovery.
Question 3: "Why did you pick this idea?"#
Formula: [Personal connection to problem] + [unique insight] + [why now]
Template#
I experienced this problem directly when [PERSONAL EXPERIENCE]. Key insight: [WHAT YOU UNDERSTAND THAT OTHERS DON'T] Why now: [MARKET/TECHNOLOGY CHANGE ENABLING THIS]
Example 4: Healthcare (Accepted W26)#
I'm a former ER nurse (8 years at Stanford Hospital). I watched nurses spend 40% of shifts on documentation instead of patient care. The problem isn't EMR software—it's that doctors and nurses think differently about patients, but use the same documentation system. Key insight: Nurses need task-oriented documentation ("gave medication at 2pm"), not narrative documentation ("patient presents with..."). Current EMRs force nursing workflows into physician documentation patterns. Why now: Voice AI accuracy finally crossed the threshold where nurses can document while working, not after. We've tested with 50 nurses—average documentation time dropped from 2.5 hours/shift to 45 minutes.
Why it works: Deep domain expertise, non-obvious insight (nurse vs. doctor documentation needs), specific "why now" trigger (voice AI accuracy).
Question 4: "Who are your competitors?"#
Formula: [Name competitors honestly] + [acknowledge their strengths] + [explain your advantage]
Template#
Direct competitors: [COMPETITOR 1], [COMPETITOR 2] Indirect/alternatives: [ALTERNATIVE 1], [ALTERNATIVE 2] [COMPETITOR 1] is strong at [STRENGTH] but [LIMITATION]. [COMPETITOR 2] focuses on [SEGMENT] while we focus on [DIFFERENT SEGMENT]. Our advantage: [SPECIFIC DIFFERENTIATOR] because [REASON YOU CAN DEFEND IT].
Example 5: Marketplace (Accepted W26)#
Direct competitors: Fiverr, Upwork, Toptal Indirect: Traditional recruiting agencies, referrals Upwork/Fiverr are strong for one-off projects but terrible for ongoing relationships. Toptal has quality but costs $50k+ to hire through them. We focus only on Latin American developers for US companies. This narrow focus lets us:Our advantage: We've built the largest vetted pool of LATAM senior developers (847 active). Competitors would need 18+ months to replicate this network.
- Deeply vet candidates (we've interviewed 2,000+ developers in LATAM)
- Handle timezone alignment (3-4 hour overlap minimum)
- Manage local compliance that generalist platforms can't
Why it works: Honest about strong competitors, specific positioning (LATAM only), defensible advantage (vetted network takes time to build).
Question 5: "What do you understand that others don't?"#
Formula: [Counterintuitive insight] + [evidence it's true] + [why competitors miss it]
Template#
Most people think [COMMON BELIEF]. We've learned [COUNTERINTUITIVE TRUTH]. Evidence: [SPECIFIC DATA OR EXAMPLE] Competitors miss this because [REASON].
Example 6: Consumer App (Accepted W26)#
Most people think Gen Z wants short-form video for everything. We've learned they actually crave long-form audio when they're alone—commuting, exercising, falling asleep. Our users listen to 47 minutes/day on average, which beats TikTok's 34 minutes. Evidence: We launched a "bedtime stories for adults" podcast as a test. It hit #4 in Apple Podcasts within 3 weeks. 73% of listeners are 18-25 (we expected 35+). Competitors miss this because they see Gen Z engagement on short-form and assume that's all they want. But short-form is for social contexts. Long-form is for alone time—and Gen Z is the loneliest generation on record.
Why it works: Challenges obvious assumption, provides surprising data (47 min > TikTok), explains why insight is non-obvious.
RFS (Request for Startups) Response Template#
Use this only if your startup genuinely fits an RFS category.
Template#
This addresses your RFS for [CATEGORY]: "[EXACT QUOTE FROM RFS]" Specifically, we solve [PROBLEM RFS MENTIONS] by [YOUR APPROACH]. Evidence this works: - [DATA POINT 1] - [DATA POINT 2] - [DATA POINT 3] We're positioned to scale because [REASON].
Example 7: AI Infrastructure (Accepted W26)#
This addresses your RFS for AI Infrastructure: "Better tools for AI development and deployment." Specifically, we solve the observability problem for LLM applications. When GPT-4 returns a bad response in production, teams have no way to debug why. We capture prompts, completions, latencies, and costs in one dashboard with automatic anomaly detection. Evidence this works:We're positioned to scale because every company building on LLMs needs this—and there are 10,000+ companies building on OpenAI alone according to their developer stats.
- 34 teams using our beta (including 2 YC companies)
- Average team finds and fixes 3 production issues within first week
- $2,100 MRR with 0 marketing spend
Why it works: Exact quote from RFS, specific problem/solution, traction evidence, clear market sizing.
Video Script Template (1 Minute)#
| Section | Time | Content |
|---|---|---|
| Intro | 0:00-0:10 | "Hi, I'm [Name], this is [Co-founder]. We're building [Company]." |
| Problem | 0:10-0:20 | "[Customer] has this problem: [pain point]. Currently they [bad solution]." |
| Solution | 0:20-0:35 | "We built [product]. Let me show you." [SCREEN SHARE DEMO] |
| Traction | 0:35-0:45 | "We have [metrics]. Growth is [trajectory]." |
| Why us | 0:45-0:55 | "We're uniquely positioned because [founder-market fit]." |
| Close | 0:55-1:00 | "We'd love to work with YC to [goal]." |
Video tips:
- Show your product—even a rough demo beats no demo
- Both founders should speak (shows healthy dynamic)
- Good audio > good video quality
- Record in one take if possible—editing looks rehearsed
Common Mistakes to Avoid#
| Mistake | Why It Fails | Fix |
|---|---|---|
| Buzzword soup | "AI-powered blockchain Web3" says nothing | Plain language: what you do for whom |
| No numbers | "Strong traction" is meaningless | Any number beats no number: "3 users" |
| Fake humility | "We're just getting started" sounds weak | Confident: "We've proven X, now scaling" |
| Giant market claims | "$500B TAM" without context | Specific: "300K target customers @ $5K/yr" |
| Ignoring competition | Suggests naivety or dishonesty | Name competitors, explain differentiation |
| Video without product | Shows you haven't built anything | Demo even a rough prototype |
Frequently Asked Questions#
How long should YC application answers be?#
150-200 words for the main description, 50-150 words for other questions. Partners read thousands of applications—clarity beats length. If you can say it in fewer words, do it. Total application should be 800-1,200 words.
Should I apply if I don't have traction?#
Yes—but you need something else impressive. YC accepts pre-launch companies, but you need: exceptional team credentials, working prototype, or deep domain expertise. "We have an idea" without any of these rarely gets interviews.
How important is the video?#
Very important—it's often the deciding factor between interview and rejection. Partners want to see: Can founders communicate clearly? Do they seem like people we want to work with? Have they built anything? A strong video can overcome a weaker written app.
Should I apply to multiple batches if rejected?#
Absolutely—Airbnb was rejected 3 times before acceptance. Each rejection gives you time to build traction. The best reapplications show concrete progress: "Last time we had 10 users. Now we have 500 paying customers."
Do I need a technical co-founder?#
For software companies, strongly preferred. Solo non-technical founders building software face an uphill battle. Options: Find a technical co-founder, build enough yourself to prove capability, or apply with a service/marketplace model requiring less engineering.
When should I apply—early or late in the window?#
Apply early with a complete application. Partners review in batches, and early applications may get more attention. But a rushed early application loses to a polished late one. Aim for first week of the window with a complete app.
Use these templates to structure your YC application. For faster drafting, try River's YC application tools to generate and refine your answers.