About roast.page
A decade of CRO and SEO consulting for SMBs, distilled into an AI scoring engine. The same caliber of audit a senior strategist would deliver — in about a minute, for free.
Why this exists
For more than a decade we ran CRO and SEO engagements for small and mid-market businesses — SaaS, ecommerce, fintech, agencies, professional services. The same handful of problems showed up on almost every page: a vague headline, a CTA that hides, trust signals that arrive too late, copy that talks about the company instead of the visitor.
The fixes were knowable. The bottleneck was always the same: a real audit costs thousands of dollars and takes weeks. Most founders never get one. They ship pages, hope the conversion rate is good enough, and never know what they could have changed.
roast.page closes that gap. We codified what consistently moved conversion across years of consulting work into an 8-dimension scoring framework, then trained an AI on those patterns. The result is a structured critique that mirrors what a senior strategist would write up after spending a day with your page — delivered in about a minute, for free.
The methodology
Every analysis runs the same four-step pipeline. Same engine for free users and Pro subscribers — there is no “light” version.
Capture
Every analyzed page is screenshotted at desktop and mobile viewports through ScreenshotOne, scraped for metadata, headings, copy, CTAs, trust elements, and benchmarked through Google PageSpeed Insights for Core Web Vitals.
Score
An AI vision model evaluates the page against eight weighted dimensions — first impression, copy, CTA, trust, design, structure, technical SEO, and differentiation. Each dimension scores 1–10. Weights aggregate to a 1–100 overall score.
Verify
Every score is recalculated server-side against the published weights so the AI cannot inflate the overall number. Industry, audience, and conversion goal are auto-detected so benchmarks are like-for-like.
Recommend
The model ranks fixes by expected conversion impact. Every issue surfaces with a specific recommendation that names the exact element on the page — never generic advice.
The 8 dimensions we score
Weights are based on what moves conversion most across thousands of real pages — not what feels important.
What we stand for
Specificity over generality
Generic advice ("improve your CTA") is worthless. Every recommendation names the element and the change.
Data over opinion
Benchmarks come from real analyzed pages — not recycled blog stats from 2018.
Industry-aware judgement
An enterprise SaaS hero is judged differently from a fitness coach landing page. Same framework, different lens.
Honest scoring
Most pages score 35–65. Scoring above 80 is rare and earned. We don't grade on a curve to flatter the visitor.
Get in touch
Questions about the methodology, requests for an industry not yet covered, or scores you want to push back on — we read every message.
Where the data comes from
Every benchmark cited on this site comes from real pages analyzed through roast.page. Conversion-rate references draw from public third-party studies, each cited inline. Where industry numbers are extrapolated rather than observed, we say so.
If we made a methodology mistake or a benchmark looks wrong, tell us — we publish corrections rather than quietly editing.
See how your page scores
Free, no signup. The full eight-dimension analysis in about a minute.