The finding that broke my assumptions first: a bootstrapped founder selling a $29/month invoicing tool scored 78. A Series B startup with $14M in funding and a 6-person marketing team scored 31. I checked the data twice because I assumed I'd mixed up the rows.
I hadn't.
Four months. 1,000 landing pages. SaaS, e-commerce, agencies, creator tools, startup launches -- roughly 60% B2B, 40% B2C. Solo founders to public companies. We didn't cherry-pick. We took them as they came through roast.page's scoring engine and let the numbers do the talking.
The Headline Number
Top 10%: 72+. Top 1%: 85+. Bottom 10%: below 24.
And that bootstrapped-vs-funded comparison wasn't an outlier. Across the full dataset, we found near-zero correlation between company size and score. The r-squared was 0.04. Essentially noise. What did correlate: specificity of copy (r=0.61), presence of quantified social proof (r=0.54), and having a single clear CTA above the fold (r=0.49).
Budget doesn't make a good landing page. Clear thinking does. I know that sounds like a platitude, but the data backs it up in a way that's hard to argue with.
Dimension by Dimension
Our system weights 8 dimensions. I'm going to spend the most time on the ones where the data was most interesting, and breeze through the ones where the story was straightforward.
First Impression & Hero -- 20% weight
Most polarized dimension in the dataset. Pages either nailed it or completely fumbled -- very little middle ground. The distribution was almost bimodal: a cluster around 25-35 and another around 65-80, with barely anything in between.
My theory on why: hero sections are a design commitment. You either thought about it carefully and executed intentionally, or you threw your tagline over a stock photo and moved on. There's no "accidentally mediocre" hero -- you either did the work or you didn't.
The top performers shared exactly three traits: a headline you could understand in under 3 seconds, a clear visual hierarchy with one dominant element, and a single obvious CTA above the fold. No tricks. No cleverness. Just clarity.
The worst offenders? Rotating carousels. Yes, people still build these in 2026. I'd love to find the person who keeps putting carousels into landing page templates and have a conversation. Also: hero sections that required a scroll to find the headline (I counted 23 of these -- nearly all in the bottom quartile), and abstract illustrations that looked nice but communicated absolutely nothing about what the product actually does.
We wrote a whole piece on why those first seconds matter so much, and our hero section playbook breaks down the patterns that actually work.
Copy & Messaging -- 20% weight
Lowest-scoring dimension overall, and it's not even close.
This one is personal for me because I think it's the most fixable problem in the entire dataset. You don't need a redesign. You don't need new tooling. You need someone to rewrite your headline and your first three paragraphs.
These phrases communicate literally nothing. They're placeholder text that never got replaced. I'm convinced half the time they came from a ChatGPT prompt that said "write me landing page copy" -- which is ironic because ChatGPT is also perfectly capable of writing specific, concrete copy if you give it actual information about your product.
Here's what separated the top-scoring pages:
"Trusted by thousands"
"AI-powered analytics engine with real-time dashboards"
"Used by 4,200 accounting teams"
"See what's actually happening in your business -- without waiting for someone to build a report"
Specificity costs nothing. It's the single highest-leverage copywriting change you can make. More on this in our copy mistakes breakdown.
Second biggest issue: feature-focused copy where benefits should be. "AI-powered analytics engine with real-time dashboards" vs. "See what's actually happening in your business -- without waiting for someone to build a report." Same product. Completely different emotional response. The first one sounds like a spec sheet. The second one sounds like someone who understands your Tuesday afternoon frustration.
Call-to-Action -- 15% weight
Highest-scoring non-design dimension, partly because the bar is low -- having a visible button with text on it gets you about halfway. But what separated the top tier was more nuanced than I expected.
| CTA Pattern | Avg Score | % of Dataset |
|---|---|---|
| "Start your free 14-day trial" | 71 | 8% |
| "Get Started" / "Sign Up Free" | 52 | 34% |
| "Learn More" | 38 | 22% |
| "Contact Sales" (only option) | 29 | 11% |
| Multiple competing CTAs above fold | 34 | 19% |
The "Contact Sales" finding was stark. When it's the only option on a page targeting individual users or small teams, it's not a call to action. It's a wall. You're asking someone who just arrived on your page for the first time to commit to a phone call with a stranger. We dig into the psychology behind this in the CTA optimization piece.
One thing that surprised me: "No credit card required" near the button correlated with measurably higher CTA scores. Not because it's clever copy -- it's the opposite of clever. It works because it removes the one specific fear people have at the moment of clicking. Simple friction reduction, every time.
Trust & Social Proof -- 15% weight
Second-lowest dimension. Most pages either had no social proof at all (28% of the dataset -- almost three in ten) or used it so generically it might as well not be there.
The "generically" part is key. I saw hundreds of pages with testimonials that read: "Great product, would recommend." Who said that? Where do they work? What specifically did the product do for them? A testimonial without a name, photo, and company is just text you wrote yourself. Everyone knows this. It builds negative trust.
What actually worked: logos with context ("Trusted by teams at Stripe, Notion, and Linear" -- not just a logo strip with no explanation), testimonials with specific outcomes ("We increased signups by 40% in 6 weeks"), and -- this was interesting -- case study links. Pages that linked to a detailed case study from their testimonial section scored 18% higher on Trust than pages that just had the quote.
More in our trust gap piece.
Visual Design & Layout -- 10% weight
Highest median. Weakest correlation with overall score.
I'll say that again because it's important: visual design had the weakest correlation with overall landing page quality in our dataset.
This doesn't mean design is irrelevant. It means the floor has risen so much -- thanks to Framer, Webflow, good Next.js starters, Tailwind UI -- that most pages look "fine." The differentiator isn't aesthetics anymore. It's copy, trust, and structure. A gorgeous page with vague copy, no social proof, and three competing CTAs still tanks.
The most common design mistake wasn't bad aesthetics. It was inconsistency. Different card styles on the same page. Mismatched button treatments. Three icon sets fighting each other. One SaaS page I looked at used Heroicons in the features section, hand-drawn illustrations in the benefits section, and stock photography in the testimonials section. It looked like three different companies designed three different sections.
The Other Three Dimensions (Quick Hits)
| Dimension | Weight | Median | Key Finding |
|---|---|---|---|
| Page Structure & Flow | 8% | 43 | Pages with a logical narrative arc (problem → solution → proof → action) scored 31% higher than those with random section ordering |
| Technical & SEO | 7% | 51 | Median page loaded 4.2 MB of assets. Top-scoring pages averaged 1.8 MB. The correlation between page weight and load time was nearly linear. |
| Differentiation | 5% | 36 | Lowest median of any dimension. 71% of pages never explained why them over the alternatives. Not once. |
That differentiation number is the one I keep coming back to. 71% of pages never answer the most basic question a visitor has: "Why you and not the other twelve options I have open in tabs right now?"
The pages that did answer it used one of three approaches: naming competitors directly and explaining the difference (rare, bold, effective -- Basecamp is the best example), identifying a specific underserved use case ("built specifically for 3-person accounting firms"), or leading with a contrarian take on the problem ("Most analytics tools give you more data. You don't need more data. You need fewer, better insights.").
Seven Findings That Challenged What I Thought I Knew
1. Hero videos underperformed static images
Before you screenshot this and tweet "video is dead" -- it's not. The problem isn't the format. It's the execution. Most hero videos were slow-loading, auto-playing brand reels that took 3-4 seconds to even start, and by then the visitor had already formed their impression from the blank loading state or the first blurry frame. The handful of pages with instant-loading, short product demos actually outperformed static screenshots. But they were maybe 5% of all video heroes.
2. Longer pages scored significantly higher
Pages with 1,500+ words scored 18% higher on average than short pages. This surprised me because the conventional wisdom in startup-land is "keep it short, attention spans are shrinking."
But here's the catch: only when the content was well-structured. Headers, visual breaks, progressive disclosure. A 2,000-word wall of text scored worse than a focused 400-word page. Length isn't the variable. Comprehensiveness + structure is.
The best analogy I can think of: a well-organized 20-page report is more useful than a confusing 2-page summary. Length is a feature when it's organized.
3. Custom illustrations didn't help (and sometimes hurt)
This one stung because I personally love custom illustration work. But the data was clear.
People want to see the actual product. Abstract visuals -- even beautiful, custom-made ones -- don't build the specific kind of confidence that converts visitors into users. Screenshots say "this is real." Illustrations say "we hired a good artist." One of those moves the needle on conversion.
I want to be clear: I'm not saying ditch all illustrations. I'm saying product screenshots should be the primary visual, and illustrations should support, not replace them.
4. Showing pricing boosted trust by 22%
Price transparency signals confidence. Hiding the price signals that you think it's a problem. Visitors read that subtext instantly. Even enterprise products benefit from showing a starting price -- it anchors expectations and filters out truly mismatched prospects, which is good for everyone.
The counterargument is always "but our pricing is complex and contextual." Sure. But "Plans starting at $49/mo for small teams" still beats a black box. You can be transparent about the starting point without committing to a full pricing table.
5. "Simple" and "minimalist" are different things
The best pages weren't the most minimalist -- they were the most focused. Big difference.
A focused page has plenty of content, all serving one conversion goal. A minimalist page that's missing key information -- no social proof, no how-it-works section, no pricing hint, no product screenshot -- performed worse than a longer, more complete page. Every time.
I think the "make it minimal" instinct comes from a good place (reduce clutter, respect attention) but it's been cargo-culted into "remove information," which is the wrong takeaway. The goal is focus, not emptiness.
6. Pages with FAQ sections scored 15% higher on Trust
Didn't see this one coming. Pages with a well-written FAQ section near the bottom scored materially higher on Trust & Social Proof, even controlling for other trust signals. My hypothesis: FAQs demonstrate that you've thought about objections. They preemptively answer the "yeah, but..." that's forming in the visitor's head. And they signal maturity -- you've been asked these questions before, which implies you have customers.
7. The "above the fold" obsession is mostly justified -- but not for the reasons people think
87% of top-10% pages had their primary CTA above the fold. But I don't think it's because people literally won't scroll. They will. The real reason above-the-fold CTAs matter: they anchor the page's purpose immediately. Even if the visitor scrolls down to read more, that early CTA sets the frame. "This page wants me to start a free trial." Everything below the fold is then read through that lens.
Pages that put their CTA only at the bottom scored lower because visitors scrolled without a framework for interpreting the content. Is this a product page? A content page? An about page? The early CTA answers that question in one second.
What the Top 10% Actually Look Like
After four months of staring at this data, I tried to distill the top-10% pattern into the smallest number of traits. I wanted five. The data insisted on six.
2. Specific, quantified proof. Not "thousands of customers" but "4,200 teams" or "$2.3M saved." Specificity was the strongest single predictor of a high overall score (r=0.61).
3. One clear next step. Never two competing CTAs. Never "Learn More" when they mean "Sign Up."
4. Visible product. Screenshots, GIFs, or embedded demos. They showed the product early and often.
5. Narrative structure. The page told a story: problem → solution → how it works → proof → action. Each section earned the next scroll.
6. Differentiation. They answered "why us?" explicitly. Not as an afterthought. Often in the hero or immediately below it.
Here's what bothers me about this list: none of it is surprising. These aren't secrets. There's no weird trick. It's just... doing the basics really well. Which apparently only 10% of landing pages manage to do.
I keep thinking about that bootstrapped founder with the 78. He didn't have a design team, didn't use Framer, didn't hire a copywriter. His page was clean, specific, showed the product, had three real testimonials from real customers, and had one big green button that said "Start free trial -- takes 2 minutes." That's it. And it beat 92% of every page we analyzed, including pages built by teams with 10x his resources.
The gap between 44 and 72 is not talent or budget. It's the willingness to be specific instead of vague, to show instead of describe, and to answer the questions visitors actually have instead of the ones you wish they had.
Where Do You Stand?
Paste your URL into roast.page and see your score across all 8 dimensions. The median is 44. The top 10% is 72+.
If you want to go deeper on specific dimensions: our hero section playbook for First Impression, the copy mistakes piece for Messaging, the CTA psychology breakdown for Call-to-Action, and the 5-second test for understanding how visitors actually experience your page in those first moments.
The data is clear on one thing: most pages leave conversions on the table not because the product is bad, but because the page doesn't communicate clearly. And that's the most fixable problem in all of marketing.