352 roasts and counting
Roast. Fix. Re-test.
Prove it.
Paste a URL. We measure the page — PageSpeed, accessibility, every link, the actual DOM — then AI turns those measurements into a scored, prioritized fix list you can act on today.
One free roast on signup · no credit card
"Mock It has a genuinely strong foundation — 530k mockups, 5,000+ templates, transparent pricing, and a 7-day free trial — and adding real attributed testimonials with star ratings would be the single fastest unlock for your conversion rate."
Read the full roast
The deliverable
What a roast looks like
"Your landing page is like a newspaper that forgot to write headlines — all the infrastructure, none of the persuasion."
#1 priority fix
Rewrite the hero to lead with the outcome, not the feature list. Your headline names the product; it never names the problem it kills.
Every roast includes
-
Scored 8-dimension breakdownWeighted for your market, benchmarked against every page we've measured -
Prioritized fixesRanked by impact, specific to your page — not a generic checklist
-
Drop-in copy rewritesHeadlines and CTAs rewritten, ready to paste -
Paste-ready AI fix promptHand the whole fix list to Claude, Cursor, or your engineer in one paste -
Markdown export & share linkClient-ready, or public in the Roastery if you want the badge
Why this isn't a GPT wrapper
Measured first. Explained second.
A chatbot guesses what your page probably looks like. We fetch it, render it in a real browser, and measure it — then the AI explains what the measurements mean and what to do about them.
What we measure
verified-
Google PageSpeed — LCP, CLS, the real Core Web Vitals
-
WCAG accessibility scan
-
Every link checked with a real HTTP request
-
Console errors from a real browser render
-
Technical SEO — meta, schema, structure
-
Full DOM parse — headlines, CTAs, social proof, pricing
What the AI does with it
interpretation-
Scores the 8 dimensions, weighted for your market
-
Writes the roast — in the tone you pick, from brutal to boardroom
-
Drafts the copy rewrites
-
Builds your paste-ready fix prompt
The model explains measurements. It doesn't invent them. That's the difference between an audit and an opinion.
The part a chatbot can't do
The loop that moves your score
01
Roast
One run measures everything and scores it. Your baseline is on the record.
02
Fix
Paste the ready-made prompt into Claude or Cursor, or hand the fix list to your engineer. The fixes are specific enough to ship.
03
Re-test
Same measurements, run again. The score delta is your proof — side by side, category by category.
Score history and before/after comparison are built in — that's what the subscription is for.
The scorecard
Eight dimensions, weighted by impact
Your value proposition and trust signals carry 40% of the score, because they carry the conversion. Weights shift when we detect your market — SaaS, e-commerce, and more.
20%
20%
15%
13%
12%
10%
5%
5%
Reading the score
What the number means
Measured across every site we have audited. The median is 58, and fewer than one site in a hundred has ever cleared 80 — so this is not a school percentage, and a 60 is not a D.
Fewer than one site in a hundred we've audited has reached this range.
Strong nearly everywhere we look. What separates this from the top is usually precise and short.
Most areas are in reasonable shape, so individual changes start to move the number noticeably.
Much of what we measure is holding up. What's left tends to be a handful of things rather than everything.
There's real substance here, with the gaps spread widely enough that the order you tackle them in matters.
Several fundamentals need attention at once, and fixing any single one of them won't move things far on its own.
The gaps here are structural rather than cosmetic — until those close, smaller improvements won't show up in the number.
Built for
People who ship pages
Founders & builders
You built the page with AI in an afternoon. Close the loop the same way: roast it, paste the fix prompt back into your tools, re-test before launch.
Marketers & CRO
Defensible numbers for the argument you're already having: a measured baseline, a prioritized backlog, and a delta you can put in a slide.
Agencies & freelancers
Client-ready audits in a minute instead of an afternoon. Volume plans up to 1,000 tests a month, exports included.
From the test kitchen
What early testers said
Real quotes, unedited beyond anonymizing names.
"Genuinely helpful! The feedback I got on my SaaS homepage was better than what Lovable's /redesign skill gave me on the same page."
"The analysis included many useful details that tools such as PageSpeed Insights do not provide, which adds real value."
"I got really good feedback about my website and it was very actionable. I actually plan to make the changes suggested and come back and get feedback again. It was very easy to figure out. The signup was straightforward. I was able to get my first result ASAP."
"The results feel actionable. I love the ability to export as markdown (so I can feed back to Claude)."
"I love the 'Priority fixes' section with clear steps and suggested fixes… great results for developers with a lot of clients."
The honest comparison
"Couldn't I just ask Claude?"
| RoastMyPage | Asking a chatbot | Agency audit | |
|---|---|---|---|
| Real measurements |  |
Guesses from training data | |
| Same test twice → provable delta | |
 |
At consulting rates |
| Time to results | ~1 minute | Minutes, unverifiable | 2–4 weeks |
| Cost to start | Free | Free-ish | $3,000+ |
| Fixes ranked by impact | |
Sometimes | Buried in a PDF |
And against the tools you already know: crawlers like Screaming Frog and Ahrefs Site Audit inspect your site's plumbing, and graders like HubSpot's check the marketing basics. None of them scores whether the page persuades — value proposition, trust, CTA clarity — weighted by impact and measured the same way twice, so a re-test proves your fix moved the number. That's the part we built.
Pricing
One roast finds the problems.
The plan is the loop.
Credits never expire. Enough to fix and re-test a redesign — and any purchase unlocks the public/private toggle on your roasts, permanently.
Questions
Fair questions, straight answers
Is this just ChatGPT with a coat of paint? +
No — and not because we say so, but because of what runs before any AI does: Google PageSpeed, a WCAG scan, HTTP verification of every link, console-error capture from a real browser render, and a full DOM parse. The AI's job is to explain those measurements and rank the fixes. Ask a chatbot the same question and it never loads your page.
How long does it take? +
About a minute. The slowest step is Google's PageSpeed run — we show you what we're measuring while it works.
What do I get free? +
One full roast — every measurement, the complete scored report, priority fixes, and the AI fix prompt. Sign in with Google or GitHub; no credit card.
Can I keep my roast private? +
Free roasts are public in the Roastery — that's what keeps it useful to browse. Once you've paid, even for a single month, you keep permanent control: list or unlist any of your roasts at any time. The share link still works for anyone you send it to.
Too harsh? Too soft? +
Pick the tone: roast, professional, coach, or executive. Same measurements, same rigor — the delivery is up to you.
See what's costing you signups
Measured audit, prioritized fixes, about a minute.