State of SaaS Landing Pages
We audited 15 of the most-used B2B SaaS landing pages with Croast. The median score is 82/100. The strongest pages share three patterns. The weakest share two.
Generated 7/1/2026 · re-runnable via npm run audit-batch
The median B2B SaaS landing page scores 82, firmly in the Solid band. The top quartile reaches 87, well into Strong. These are well-funded teams with professional copywriters.
Pages that score below 75 share two patterns: vague headlines (Figma, Slack, Asana rely on brand recognition rather than specific promises) and choice-paralysis from multiple primary CTAs (Buffer, Slack).
Pages that score 85+ share a counter-intuitive pattern: a single primary CTA, not three. Loom, Linear, Airtable, Intercom, Vercel each have one button. Visitors do not need options; they need clarity.
The Tools for X headline (Ahrefs, Buffer) reads as a feature list. The trio-of-verbs headline (Vercel, Notion) reads as a workflow. The former is Amazon; the latter is the homepage.
FAQ presence is universal across all 15 pages. What is not table stakes: specific customer counts (10,000+ teams) vs. vague customer categories (leading brands).
Pricing is mentioned on 15/15 pages. Where they differ: 11/15 hide the actual number behind a button, 4/15 show tiers. The show-tiers pages score on average 6 points higher.
Top scorers name a specific outcome in 5-12 words. Bottom scorers reach for poetic 2-3 word slogans that confuse with their brand tagline.
Verbs beat nouns. Move work forward is weaker than Develop. Preview. Ship. Both have rhythm, but only one names what you do.
Three winning patterns emerged: (1) trio-of-verbs (Vercel, Notion), (2) super-category plus qualifier (Loom, Intercom, Linear), (3) one-sentence promise with a number (Airtable, Stripe).
Bottom scorers rely on brand recognition. Top scorers do not need to, the headline sells a product, not a logo.
Each page was fetched by Croast's server and analyzed by our rule engine. The engine scores 0-100 based on nine categories: headline, CTA, social proof, trust, specificity, page depth, SEO, FAQ, and pricing.
Headlines were additionally classified by an LLM into one of five strategies: outcome, feature, audience, category, or vague.
The corpus-level insights were generated by feeding all 15 (page, score, headline) triples to the LLM and asking it to find patterns. To re-run with fresh data, set LLM_API_KEY and LLM_BASE_URL and run npm run audit-batch.
Free · 60 seconds · no signup