URL Roasts In Depth

How a URL roast works

You paste a page URL. PageScorch fetches the page, extracts the visible text and copy, and sends it to the AI for analysis.

The AI reads the page content the same way a skilled copywriter would: headline, subheadline, body copy, CTA text, social proof, objection handling, and overall narrative structure. The result is a conversion score, problem list, prioritized fixes, and AI copy rewrites.

What the AI reads

PageScorch extracts visible text from the page — the content a reader actually sees — and strips out scripts, styles, and non-visible markup before analysis. This means the AI focuses on:

• Headline and subheadline copy
• Body copy and value propositions
• CTA button text
• Testimonials, review quotes, and social proof text
• FAQ content and objection-handling copy
• Pricing and plan descriptions

The AI does not read JavaScript-rendered content, images, or CSS. It analyzes the copy and messaging structure.

What URL roasts are best for

Use a URL roast when:

• You care about copy quality and messaging structure, not layout
• You want to analyze a live public page quickly without taking a screenshot
• You are auditing competitor pages to benchmark messaging
• You want to run bulk page analysis via CSV (Agency plan)

Bot-blocked pages

Some pages block automated fetching (Cloudflare, anti-bot services, paywalls). When this happens, PageScorch returns a BOT protection detected result rather than a score.

What to do: Use the manual paste fallback. On the failed roast result page, you will see a "Paste page source" option. Open the page in your browser, use View Source (Ctrl+U / Cmd+U), copy the HTML, and paste it into the field.

PageScorch sanitizes and processes pasted HTML through the same pipeline as a fetched URL. The result is a normal roast with a conversion score and full analysis.

See Manual HTML paste for bot-blocked URLs for step-by-step instructions.

URL validation and safety

PageScorch validates URLs before fetching:

• Only http:// and https:// schemes are accepted
• Private IP ranges (localhost, 192.168.x.x, 10.x.x.x, etc.) are blocked
• Non-standard ports are blocked
• Redirects are followed but validated at each hop

If a URL fails validation, you will see a specific error message explaining why.

Content limits

PageScorch reads up to a set amount of page content. Very long pages (e.g. long-form sales letters) may be analyzed based on the most conversion-relevant sections the parser extracts. The AI is optimized to find and prioritize headline, CTA, and above-the-fold copy.

Screenshot roast vs URL roast — which to use

| Situation | Best approach | |-----------|---------------| | Evaluating visual design and layout | Screenshot roast | | Evaluating copy and messaging | URL roast | | Page is behind a login or paywall | Screenshot roast | | Bulk auditing a list of public pages | URL roast (CSV import) | | Page is bot-blocked | Screenshot roast or pasted-HTML URL roast | | Analyzing a design mockup | Screenshot roast |

When in doubt, run both. Screenshot and URL roasts capture different dimensions of the same page and the scores complement each other.

After your roast

Once the URL roast completes, you can:

• Read the full problem list and copy-focused fix suggestions
• Generate an AI Fix Report for rewritten headline, subheadline, and CTA
• Re-roast the URL after updating the page to see if the score improved

See Generating AI fixes for the next step in the workflow.