SEO Audit

Loading tool…

About SEO Audit

Paste a public homepage and choose how many pages to crawl: 25, 50, 100, 200, 250, or 500. The crawl starts at that URL, reads robots.txt, prefers sitemap locations, then follows same-host links. Click Run audit. You get a score ring, five category bars, a filterable issue list, and a sortable page table with status, title, description, H1, words, depth, inbound and outbound counts, schema, and issue counts.

Identical issues collapse into one row so you are not reading the same warning on every URL. Category bars are summaries; open the issue list to see which pages they hit. Download PDF for a full report or CSV for the raw tables. This is not Lighthouse or PageSpeed. We read the HTML the server sends. JavaScript-only routes are missed.

When to use it

Run it after a shared template change: a new title pattern, a dropped H1, or a footer that now links to retired URLs. Those faults repeat on every crawled page, and the collapsed issue list shows the blast radius before you ship.

Also run it before a CMS or host move, when you need an HTML baseline of titles, H1s, and inbound counts.

How to use SEO Audit

  1. 1Enter the homepage. Use a public https URL. Localhost and private addresses are blocked.
  2. 2Choose pages to crawl. 25 is the default. 50, 100, 200, 250, or 500 take longer. The form shows an estimated time.
  3. 3Click Run audit. Progress shows crawled n of N, then scoring, pair comparison, and links.
  4. 4Filter, then download. Category and severity filters change the on-screen list. PDF and CSV are the full report.

How the score is built

A letter grade needs at least three crawled pages. Fewer than that shows crawl diagnostics (sitemap probes and homepage link counts) instead of a grade. Categories are Crawlability and Indexing (30), On-Page (30), Technical (20), Content (10), and Links (10). Each distinct issue type is penalized once, then scaled by how many crawled pages it affects.

Critical issues cap the overall score: one or two keep it at 89 or below (B at best), three or four at 79, five to nine at 69, ten or more at 59. Missing title or H1 on the homepage counts twice toward that ceiling. Category bars are not capped that way. After the crawl, pair comparison checks every crawled page against every other page for near-duplicate main text at 85% similarity. Link checks use the same unique-link cap as the page limit you picked (25, 50, 100, 200, 250, or 500).

How to read the results

The ring is the capped overall score and letter. A needs zero criticals. One critical locks it at 89 (B at best). Bars summarise the five categories. They do not list URLs.

Open Issues, then set Category and Severity. Critical first: those set the ceiling. Warning is a concrete HTML fault. The page table sorts by status, title, H1, words, depth, inbound, outbound, schema, and issue count.

Why this matters

A crawler that hits an empty shell or a missing homepage title cannot score the rest of the site honestly. That is why a missing title or H1 on the start URL counts twice toward the critical ceiling, and why a one-page crawl refuses a letter grade.

Collapsed rows exist because the same template fault is one defect, not fifty. Pair comparison at 85% similarity flags near-duplicate main text, so a thin site can look large in the page table while the Content bar still drops.

Limits and privacy

Three audits per hour on your network, including 25-page runs. Localhost and private addresses are blocked. Only crawl a site you own or have the right to scan. A short cache may reuse a recent run. No account and no crawl history.

Common mistakes

Auditing www when the live host is the apex (or the reverse). The crawl stays on the host you typed, so the page table looks tiny and sitemap probes miss the real file.

Reading no letter grade as a bug. example.com returns one 200 and crawl diagnostics instead of a ring, because a grade needs three pages.

Treating a high On-Page bar as a clean title set, or reading an A as Search Console coverage. Open Issues and sort the page table by title or issue count.

FAQs

Is this Lighthouse or PageSpeed?
No. There is no lab performance score, no Core Web Vitals, and no headless Chrome. The score is from HTML checks on crawled pages.
Why is there no letter grade?
Fewer than three pages were crawled. The report then shows sitemap and homepage diagnostics instead of inventing a grade.
How does the crawl pick pages?
Start URL first, then robots.txt, then sitemap locations, then internal links on the same host (www stripped). PDFs, images, CSS, and similar files are skipped.
Does JavaScript content count?
No. If the first HTML is a shell or a block page, you will see few pages. We do not guess that the site is JavaScript-rendered.
What is in the PDF vs the CSV?
PDF is a readable report. CSV has tables for pages, issues, issue URLs, categories, and a summary. Filters on screen do not shrink the downloads.
Does /seo/website-audit still work?
Yes. That path 301s here. Same tool.
Can I crawl more than 500 pages?
No. Presets stop at 500, the hard ceiling per run. Three audits per hour on your network, including 25-page jobs. Split a large origin by section, then merge the CSVs.
Why is the score different from Screaming Frog?
Frog is a desktop crawl you configure. This score is HTML-only, five weighted bars, and a critical ceiling. It does not run JavaScript or use Search Console, and it will not invent a grade below three crawled pages.
What do Critical, Warning, and Notice mean?
Critical sets the score ceiling and should be fixed first. Warning is a concrete HTML fault such as a short title. Notice is advisory, like a title that matches the H1. Pass rows are omitted from the severity filter.