Founders assume the companies a step ahead of them have this figured out. We checked.
On 25 July 2026 we crawled 800 company sites from Y Combinator's public directory, filtered to active B2B and SaaS companies, shuffled with a published seed (20260725). 675 gave us a page we could score. These companies have funding, hires and investors reading their site.
The pages are worse than you would guess.
Close to a third never ask
30.7% of the 675 had no call to action in the body of the page.
We exclude the navigation, the header and the footer from that count. Almost every SaaS site has a sticky header with "Get started" in it, and counting it would measure the header's existence rather than the page's argument. Strip it away and 207 of 675 funded startups publish a page that asks the reader for nothing.
83 sites showed a crawler nothing at all
Of the 800 sites we attempted, 83 served no meaningful HTML until a JavaScript bundle ran.
That is the single largest failure category in our ledger, larger than 404s, 403s, rate limits and network errors combined.
Some search crawlers execute JavaScript and some do not. AI retrieval systems make no guarantees. If your marketing page is a client-rendered application shell, you are betting your discoverability on someone else's rendering budget.
Test it in one minute: disable JavaScript in your browser and reload your landing page. If you see a blank screen or a spinner, that is what part of the web sees.
Full failure ledger, from 800 attempts:
| Reason | Count |
|---|---|
| No HTML without JavaScript | 83 |
| Network error | 13 |
| HTTP 403 | 8 |
| Body too small | 7 |
| HTTP 429 | 5 |
| HTTP 404 | 4 |
| Robots disallowed | 2 |
| 503, 526, 402 | 3 |
We report these as failures rather than scoring them zero. A scored empty page drags every median toward zero and quietly ruins the dataset.
They write too hard
Median Flesch-Kincaid reading grade across the 595 pages with enough prose to score: 10.4. The upper quartile reached 12.5.
Unbounce, working from around 41,000 landing pages with conversion data attached, found pages written at a 5th to 7th grade level converted at 11.1% against 5.3% for pages at a professional reading level.
Most of these startups write four or five grades above the level that performs best on a much larger sample.
They hide the price and skip the proof
15.4% stated a price. 42.2% showed any social proof. 11.6% put that proof before the first call to action.
Ziakis and colleagues surveyed 184 people on what drives landing page decisions and found trust ranked first among five dimensions, with reviews, guarantees and detailed descriptions all above 78%. Personalization ranked last. It is a convenience sample recruited over social media and it measures what people say rather than what they do, so discount it. The direction still contradicts what these pages ship.
The rest of the numbers
| Measurement | Median | p25 | p75 | n |
|---|---|---|---|---|
| Word count | 759 | 458 | 1,218 | 675 |
| Words before first in-body CTA | 34 | 21 | 101 | 464 |
| Distinct CTA labels | 1 | 0 | 2 | 675 |
| Total CTA elements | 2 | 0 | 3 | 675 |
| Form fields | 0 | 0 | 1 | 675 |
| Images missing alt text | 6 | 0 | 34.5 | 675 |
The words-before-CTA figure is a document-order position. It is not a claim about the fold, which needs a rendered viewport we did not use.
Two further findings: 8.1% shipped no H1 element, and 7.1% of the 620 pages with a measurable headline ran past ten words.
What a founder should do with this
The bar is low, which is good news if you are willing to clear it.
- Disable JavaScript and reload your page. Fix it if it disappears.
- Cover your header with your hand. Does anything left ask the visitor for something?
- Move your proof above your first ask. Reordering costs nothing.
- Publish a price, a range, or the shape of the model.
- Run your copy through a readability check and aim three grades lower than feels professional.
- Confirm an
<h1>exists in the markup, not a styled div.
Doing all six puts you ahead of most of a Y Combinator cohort.
Method
Frame: Y Combinator public company directory via the yc-oss mirror, active companies classified B2B or SaaS. Seeded shuffle, seed 20260725, 800 attempted, 675 scored, collected 25 July 2026. Robots.txt honoured, identifying user agent, bounded concurrency.
We did not measure conversion. Nothing here is a conversion claim.
Sources
- Ziakis, C., Papadopoulos, A., Antoniadis, I. and Saprikis, V. (2026). Conversion Rate Optimization in e-Business: Trust, Design, and Personalization Implications for Landing-Page Design. Springer Nature Switzerland. DOI: 10.1007/978-3-032-12968-0_37
- Unbounce, Conversion Benchmark Report. Approximately 41,000 landing pages, 464M pageviews, 57M conversions.
- LeadzLander, The State of SaaS Landing Pages (2026). Collected 2026-07-25, seed 20260725.