How we grade websites
Every grade our tool hands out can be traced to a measurement. This page explains exactly what we measure, how it's weighted, and what we can't measure, in plain English.
The short version
We fetch your live homepage the way a search engine does, run it through Google's own PageSpeed test, parse the page's code for structured data and SEO fundamentals, and score seven categories from 0 to 100. Your overall grade is a weighted average: A is 90 or above, B is 80 to 89, C is 70 to 79, D is 60 to 69, and F is below 60. Nothing is hand-adjusted; the same site graded twice gets the same treatment both times.
We report the checks you pass as well as the ones you fail. Most audit tools only list faults, which makes them read like a sales pitch and gives you no way to tell whether the tool actually looked. Your results show what's already working, with the measurement attached, before anything it thinks is broken.
The seven categories and their weights
| Category | Weight | What it means for you |
|---|---|---|
| Performance / Core Web Vitals | 22% | How fast your site loads and responds on a phone. |
| AEO / Structured Data | 18% | Whether AI assistants and Google's smart features can actually read what your business is. |
| On-Page SEO | 16% | The on-page basics search engines expect. |
| Content & Copy | 10% | Whether there's enough substance on the page for search engines and AI to work with. |
| Conversion & Lead Gen | 12% | Whether a visitor who likes what they see can easily become a lead. |
| Accessibility | 10% | Whether people using screen readers, or just a phone in sunlight, can use your site. |
| Technical Foundation | 12% | The health of the foundation your site is built on. |
Weights reflect what moves revenue for a typical local or small business: speed and AI-search visibility lead because they decide whether you get found at all.
How each category is measured
Performance / Core Web Vitals
We run Google's own PageSpeed test (the same one at pagespeed.web.dev) in mobile mode and use Google's 0-100 performance score directly. We also report the raw numbers behind it: how long your main content takes to appear, which element is holding that up, how long your server takes to answer, how long scripts freeze the page, and how much the layout jumps while loading. Where Google has enough traffic data for your site, we also show what your real visitors experienced over the last 28 days, which is a stronger signal than any lab test.
AEO / Structured Data
We parse your page's structured data (machine-readable labels hidden in the code) and check for the five types that matter most for a business: your business identity, your products or services, FAQs, reviews, and page structure. Each type present earns points. We don't guess: the markup is either there or it isn't. We also read your robots.txt to see whether AI crawlers such as GPTBot, ClaudeBot, and PerplexityBot are permitted to fetch your site at all, because markup an assistant is forbidden to read is worth nothing to it.
On-Page SEO
Half the score is Google PageSpeed's own SEO check. The other half is ours: a proper page title, a meta description, exactly one main headline, a canonical URL, descriptive image alt text, a robots.txt, and a sitemap. For the sitemap we use the location your robots.txt declares, whatever it's named, and only fall back to guessing the standard paths if it declares none. If a check can't run (say, our request for your sitemap fails), it's excluded from the math rather than counted against you. We also note when the address we asked for redirects somewhere else, since each variant should settle on one canonical home. One thing we do treat as serious: a robots.txt that tells every search engine to stay off the whole site caps this category, because no other on-page work matters while that line is live.
Content & Copy
We count the words a visitor can actually see (code and scripts excluded), look for FAQ-style content that answers real questions, and check heading structure. This category is heuristic, so we label it "estimated". Volume isn't the same as quality, and we say so.
Conversion & Lead Gen
We look for a real contact form (with enough fields to follow up on), a tap-to-call phone link, visible calls-to-action, and a clear path to a contact or booking page. Also labeled "estimated", because we measure the plumbing, not the persuasion.
Accessibility
Mostly Google's accessibility score, blended with our own check of image alt attributes. We follow the same standards Google does: images marked as decorative pass, and only images with no alt attribute at all count against you. We also report the specific failures Google found beyond alt text: form fields with no label a screen reader can announce, links and buttons with no accessible name, headings that skip a level, and a page that doesn't declare its language.
Technical Foundation
HTTPS, security headers, how many scripts and stylesheets the page carries, page weight, and whether the site runs on a heavy page-builder stack or legacy libraries like jQuery. Bloat caps how fast and stable a site can ever be, so it costs points. We also report measured dead weight: how many kilobytes of CSS and JavaScript Google found that the page downloads but never uses, leftover polyfills for browsers that no longer need them, and how long third-party tags and widgets hold up the page. We detect the platform from your site's asset paths and class names only, never from words in your copy, so a page that merely mentions a tool isn't reported as running it.
What we're honest about
- "Estimated" labels. Content and Conversion (and any category where Google's test was unavailable) are heuristic. We mark them "estimated" on your results and in the report rather than presenting them as lab measurements.
- Score wobble. Google's performance test naturally varies a few points between runs, and that's true at pagespeed.web.dev too. A site near a grade boundary can flip a letter on a re-run. The findings underneath don't change.
- "Not measured" means exactly that. If one of our external checks doesn't answer, we mark that category "not measured", leave it ungraded, and weight your overall score across the categories we did measure. We'd rather hand you a smaller honest number than a complete-looking one with a guess inside it.
- Client-rendered sites get told so. If your page is assembled in the browser, the HTML your server sends is nearly empty, which is what crawlers and AI assistants read first. We report that as its own finding rather than quietly marking you down for thin content, because the cause and the fix are different.
- Homepage only. The free grade audits your homepage, the page search engines and visitors judge first. It doesn't crawl your whole site.
- Client-rendered sites. We read your page's served HTML the way crawlers first do. If your site only shows content after running JavaScript, some categories will score low, which is itself a finding worth knowing about.
- We grade ourselves. panicciventures.com goes through this exact tool, and we publish the result. Run it yourself anytime.
The paid reports
The $24.99 standard report doesn't change your grades. It explains them: every finding with what we measured and why it matters, the raw metrics, and a Google Business Profile baseline (your rating, reviews, and visibility gaps on Google Maps).
The $49.99 Pro report adds four things no cheap tool bundles: a competitor comparison (you name one competitor and we run their site through the exact same audit, side-by-side), a 30-day action plan (your findings re-sorted by impact vs. effort into a week-by-week checklist), a live AI visibility check (we ask a leading AI assistant about your business at report time and print exactly what it knows, whether it would recommend you, and who it names instead), and a market position section.
Market position is the one part of the report that isn't measured from your page. It answers whether anyone actually finds you: how many keywords you rank for, roughly what that traffic would cost to buy through ads, how many domains link to you, and your authority score, next to your competitor's. It comes from a third-party search index rather than from your site, so we label it an informed estimate rather than a measurement, and if the lookup doesn't return usable data we leave the section out instead of filling it with zeroes.
Two rules govern what we’ll say about those numbers. First, every claim is tied to the measure that supports it: keyword count buys “covers more ground” and nothing stronger, because how many terms a site appears for is not the same as how many people arrive. Whether you or a competitor gets found more often is a statement about visits, so we only make it when we have traffic figures for both of you, and when the two measures disagree we say that instead of picking the flattering one. Second, because these are estimates rather than measurements, two sites within 10% of each other on a measure are reported as a dead heat rather than a lead. A competitor showing a bare zero never earns you a win either: that’s far more often a gap in the index than a genuinely invisible business, and it’s the direction you’d have no reason to question.
For context, a manual audit at this depth is typically a $500 to $2,000+ agency deliverable, and competitor analyses alone usually start around $500. And if you decide to work with us, your report purchase is credited in full toward your first monthso either report effectively costs nothing if we end up fixing it together.