We Scored 50 YC Startup Landing Pages. The Average Was 41.

There's something deeply ironic about a company that can build a real-time collaborative database engine but can't explain what it does on a webpage. And yet that was the pattern -- over and over -- when we ran 50 Y Combinator startup landing pages through roast.page's scoring engine.

These aren't random bootstrappers. These are companies that survived the most competitive accelerator on earth, raised real money, and hired engineers who could probably rebuild your entire product in a weekend. Their landing pages scored worse than the overall median.

I checked the data twice. I assumed the sample was skewed. It wasn't. We pulled 50 YC companies across recent batches -- dev tools, fintech, healthtech, AI, B2B SaaS, and a handful of consumer products. The distribution was almost identical to the broader dataset, just shifted slightly left. More pages clustered in the 30-45 range. Fewer in the 60+ range.

The question isn't whether YC startups are bad at building products. They're clearly not. The question is why good engineering teams consistently produce mediocre marketing pages. And the answer, once you see it in the data, is obvious.

Engineer-Led Pages Read Like Documentation

YC startups are, overwhelmingly, engineer-founded. That's a feature when you're building a product. It's a liability when you're selling one on a webpage.

The pattern was everywhere. Pages that described the architecture instead of the outcome. Headlines that named the technology instead of the problem it solves. Feature lists that read like internal Notion docs -- comprehensive, accurate, and completely unpersuasive.

One dev tools company -- 10K+ stars on GitHub, real community traction, genuinely loved by the developers who use it -- scored 29. Their entire landing page was, effectively, API documentation. Endpoint descriptions. Code snippets. A list of supported languages. If you already knew what the product did and why you needed it, the page was great. If you were a first-time visitor trying to understand the value proposition, you'd bounce in five seconds.

They had 10,000 people who'd already voted with their stars. They just forgot to explain why to anyone else.

Where YC Pages Failed: Dimension by Dimension

Copy & Messaging -- the weakest link

This was the worst dimension by a wide margin, and the single biggest drag on YC scores overall.

The syndrome has a name at this point: "AI-powered platform for [category]." I saw it on 19 of 50 pages. Nineteen. Nearly 40% of YC startups opened with some variation of "AI-powered platform for" followed by a category noun. Infrastructure. Compliance. Recruiting. Data ops. The words changed. The emptiness didn't.

Here's the thing about "AI-powered platform for compliance" -- it tells me exactly two things: you use AI (so does everyone), and you work in compliance (fine). It tells me nothing about what you actually do, what specific problem you solve, or why I should care. It's a category label, not a value proposition.

The second column isn't cleverer. It's just specific. It names the pain. It uses the language a real person would use when describing their frustration. That gap -- between how engineers describe their product and how users describe their problem -- is where most YC landing page copy goes to die. We covered this pattern in detail in our copy mistakes analysis.

Trust & Social Proof -- the ironic failure

This one genuinely baffled me. YC startups have trust signals that most companies would kill for. They have the YC badge. They have real investors. Many have real customers, real revenue, real traction numbers. And they just... don't show any of it.

My best guess: founders think it's tacky, or they assume their audience already knows. Both are wrong. The YC badge is a shortcut to credibility. It says "someone very selective vetted us." Leaving it off your page is like having a Stanford degree and not mentioning it in a job application because you assume the interviewer will figure it out.

Call-to-Action -- vague and generic

"Get Started." "Sign Up." "Request Access."

That was the CTA on 34 of 50 pages. No context. No hint of what happens next. No friction reduction. Just two words on a button and a hope that the visitor has already decided.

The problem with "Get Started" is that it answers none of the questions forming in the visitor's head: Is this free? Do I need a credit card? How long does setup take? Will I talk to a salesperson? "Get Started" says nothing except "click here," which is the job of every button ever made.

Compare that to the top performers: "Start free -- no credit card, 2-minute setup" or "See your first report in 5 minutes." These CTAs do work. They reduce anxiety, set expectations, and give the visitor a reason to click now rather than bookmarking the tab and forgetting about it.

First Impression -- template syndrome

I could have swapped the hero sections of 20 of these pages and nobody would have noticed. Same gradient backgrounds. Same sans-serif headline. Same abstract pattern or mesh gradient. Same layout: headline, subheadline, two buttons, maybe a screenshot below the fold.

There's nothing wrong with using a template -- we covered this in our five-second test piece -- but when your page looks identical to every other startup in your batch, you've lost the one thing a landing page needs to do in its first three seconds: make the visitor feel like they're in the right place for their specific problem.

The B2C Gap: 15 Points

This was the biggest surprise in the dataset.

The reason isn't that B2C founders are better marketers. It's that B2C pages are forced to be clear. When your user is a regular person -- not a developer, not a VP of Engineering -- you can't hide behind jargon. You can't say "orchestration layer" and expect them to nod along. You have to say what the thing does in words a human would actually use.

B2B pages, especially dev tools and infrastructure products, have a built-in excuse to be vague: "our audience is technical, they'll get it." Maybe. But "getting it" isn't the same as being persuaded. Even a senior engineer comparing four tools in browser tabs will gravitate toward the one that most clearly articulates its value. Clarity wins regardless of audience sophistication.

What the Top 10% Did Differently

Eight YC pages in our sample scored 65 or above. Here's what they had in common -- and it wasn't design budget or headcount.

None of this is groundbreaking. That's the point. The top YC pages didn't do anything exotic. They just didn't skip the basics that most YC pages skipped. Our headline analysis breaks down exactly why leading with pain outperforms leading with product descriptions -- it's the same pattern here, just concentrated in the YC dataset.

Five Things YC Founders Should Fix Today

If you're a YC founder reading this and your score is somewhere in the 30s or 40s -- which, statistically, it probably is -- here's where to start. These are ordered by impact-per-hour, not importance.

1. Rewrite your headline to name the pain, not the product. Delete "AI-powered platform for." Replace it with the sentence your best customer used when they described why they signed up. If you don't know that sentence, ask them. Today. It's a Slack message, not a research project.

2. Add the YC badge and your traction numbers. Below the hero, one line: "Backed by Y Combinator. Used by [X] teams." If you have notable investors, add their logos. This takes 10 minutes and directly impacts your Trust score.

3. Put a real screenshot above the fold. Not an illustration. Not an abstract graphic. The actual product. If your product doesn't look good in a screenshot yet, that's a different problem -- but hiding it isn't the solution. People want to see what they're signing up for.

4. Rewrite your CTA with context. "Get Started" becomes "Start free -- takes 2 minutes." "Sign Up" becomes "Try it free -- no credit card needed." Tell the visitor what happens after they click. Remove the ambiguity that keeps their hand hovering over the mouse.

5. Add one specific testimonial. Not "Great tool!" from Anonymous. A real quote from a real user with a real name and a real outcome. "We cut our deploy time from 45 minutes to 3 -- Jamie Torres, Staff Engineer at Linear." One good testimonial outperforms ten generic ones.

You could do all five in an afternoon. Most of them don't require a designer. They definitely don't require an engineer. They require someone willing to think about the page from the visitor's perspective instead of the builder's.

The Uncomfortable Takeaway

YC startups build incredible products. That's not in dispute. But a landing page isn't a product. It's a sales conversation compressed into a single scroll. And the skills that make you great at building -- precision, technical depth, comprehensive feature coverage -- are often the opposite of what makes a landing page convert.

Good engineering is about completeness. Good landing page copy is about omission -- saying the one thing that matters and leaving the rest for the docs.

The founders who get this right treat their landing page like a product problem: who's the user, what do they need to see, what's the fastest path to the outcome? The ones who don't treat it like an afterthought -- something you throw together after launch and never touch again.

The average YC landing page scores 41. The median across all pages we've analyzed is 44. Being backed by the best accelerator in the world doesn't automatically make your marketing page good. If anything, it might make you less likely to prioritize it.

Run your page through roast.page and see where you actually stand. The scoring takes 30 seconds. The fixes, if you take them seriously, take an afternoon. And the gap between a 41 and a 65 is not talent or budget. It's the willingness to stop describing your product and start selling it.