free tool · benchmark of 705 YC 2026 homepages · 2026-09-28
YC Homepage Grader
Paste a homepage. We fetch it once, read the HTML the way a browser and a search engine do, and score it against the 705 Y Combinator 2026 company sites from our research. You get a number, a percentile, the evidence and the three fixes that move it most.
Takes about five seconds. Nothing is stored until you ask for the report.
What it measures
Six dimensions. Five are scored from rules you can read under each bar; the sixth, your stack, is shown for context. Clarity: is there a headline, is it 3–12 words, is there a description. Conversion: a recognisable call to action, a specific one, a link to pricing. Distinctiveness: a declared typeface, not Inter, no "AI" in the first line (60.7% of the batch has it). Craft: share image, favicon, canonical, structured data, viewport. Speed: HTML weight and script count against the YC distribution. Percentiles are real ranks: the same rules run over every site in the benchmark.
It reads served HTML only. No JavaScript is executed and no screenshots are taken, so it can't judge your visuals or your copy's tone; it can tell you whether the fundamentals a founder ships in a weekend are there, and how you compare with the batch.
Questions
What does the grader check?
Only what a browser receives: the served HTML and response headers of one URL. From that we read the headline, meta description, calls to action, whether pricing is linked, whether the first line leans on the word AI, the fonts declared, social image, favicon, canonical link, structured data, viewport tag, HTML weight and script count, plus the site builder and host. No JavaScript is executed and no screenshots are taken, so text injected by script can be under-counted and speed is a proxy, not a Lighthouse run.
Why compare against YC 2026 homepages?
Because it is the most current picture of what funded early-stage companies actually ship. For our research we fetched the public homepage of every company in YC's Winter, Spring, Summer and Fall 2026 batches: 705 reachable sites. 60.7% lead with AI, 18.0% link to pricing, 49.1% run on Next.js. Your percentile is your rank against those 705 sites on the same measurements.
Is my data stored?
A grade is cached for 24 hours, keyed by the URL, so repeat runs are instant and we don't hammer your server. If you ask for the report we store the email you give us, the URL and the score, nothing else, and we use it to send the report and, at most, an occasional note from the studio. Email atul@nonboring.gg to have it deleted.
How is it scored?
Five dimensions are scored 0–100 from fixed rules you can read in the evidence under each bar: Clarity (headline present and 3–12 words, description present and 60–170 characters), Conversion (a recognisable call to action, a specific one, a pricing link), Distinctiveness (a declared typeface, not Inter, no AI cliché in the first line), Craft (og:image, favicon, description, canonical, JSON-LD, viewport) and Speed (HTML weight and script count against the YC distribution). Overall is weighted 25/25/20/15/15. Stack is shown for context and not scored. Percentiles come from running the same rules over the benchmark.
Can I get help fixing it?
Yes. The Launch Sprint is a fixed-scope, fixed-price engagement: $12,000, two weeks, one homepage rebuilt with the message, the type and the conversion path sorted. Or email atul@nonboring.gg with your grade and we'll tell you honestly whether you need us.
Is it free?
Yes. No account, no card, and the report is yours to print. The only limit is 30 grades an hour per network so the tool stays up for everyone.
Building something people should remember?
non boring.gg is a product and brand design studio in Bangalore, India, working with startups and tech companies across the US, Europe and Australia. Average engagement around $20,000. atul@nonboring.gg