What does this SEO audit check?
An SEO audit checks the tags and structure a search engine reads: the title, the meta description, the canonical link, the heading outline, image alt text, Open Graph tags, the viewport declaration and any JSON-LD. This tool does that two ways. Paste one page's source and the twelve checks run in your browser; enter a domain and a crawler reads every page the site publishes in its sitemap, scoring each one from 0 to 100.
Crawling needs a server, and only for one reason: a browser is not allowed to read another origin's HTML unless that origin opts in with CORS headers, and almost none do. So when you crawl, your browser asks this site's own endpoint for the pages a few at a time, which keeps every request short, makes the progress bar honest, and obeys the target's `robots.txt` — both its Disallow rules and any `Crawl-delay`. Pages you paste are never uploaded.
Both modes apply the same twelve checks, the same thresholds and the same arithmetic: 100 minus 15 for every failed check and 7 for every warning, floored at zero. That is a triage heuristic, not a ranking prediction — a page with no title lands in the fifties even when everything else is right. The crawler reads raw HTML with regular expressions where the pasted-source audit uses the browser's own DOM parser, so one page can land a point or two apart between the two.
How to audit a page or a whole site
- For a single page: open it, press Ctrl+U (Cmd+Option+U on macOS) to view the source, paste the whole document into the box and press Audit page. Nothing is uploaded.
- For a whole site: switch to Whole site, enter the domain and press Audit the whole site. Pages are discovered from robots.txt and the sitemap, then read 18 at a time.
- Watch the progress line while it works. You can press Stop at any moment and keep whatever has been read so far.
- Start with Fix these first: it counts each problem across every page crawled, so the template flaws that are worth one edit show up at the top.
- Work down Worst pages first, the lowest-scoring end of the crawl. Each row names the checks that failed, and each issue carries the score it cost.
- Download the CSV when you would rather sort and filter in a spreadsheet, then fix the template rather than the page when a check fails everywhere.
SEO audit limits and rules
| Item | Limit |
|---|---|
| Modes | One pasted page in the browser, or a crawl of one whole site |
| Pages found via | robots.txt sitemap directives, /sitemap.xml, /sitemap_index.xml, /wp-sitemap.xml, then sitemap indexes followed into up to 20 child files |
| Pages per crawl | 1,500 from the site's sitemap; the browser stops asking after 600 and renders the 100 lowest scores |
| Pace | 18 pages per request, 6 at a time, 12 seconds allowed per page |
| robots.txt | Obeyed: Disallow rules are skipped and reported, Crawl-delay is honoured |
| Cache | Each page and sitemap is reused for 10 minutes, so a repeated crawl costs the site nothing |
| Crawl budget | 1,500 pages per visitor per hour, so the server cannot be used as someone else's crawler |
| Checks | Title, description, canonical, robots, language, H1, images, links, Open Graph, viewport, structured data, word count |
| Title threshold | 15–60 characters; outside that is a warning |
| Description threshold | 50–160 characters |
| Scoring | 100 minus 15 per failed check and 7 per warning, floored at 0 |
| Not included | Core Web Vitals, backlinks, rankings, redirect chains, duplicate content, JavaScript-injected tags |
When to use it
- Finding the handful of pages on your site that are thin, untitled or missing a description, which is invisible one page at a time.
- Checking a canonical tag after a migration, where a single wrong template affects every URL.
- Auditing a site before quoting a fix, and exporting the CSV as the evidence.
- Spotting images that shipped without alt text across a whole content library.
- Confirming that no page accidentally carries a noindex directive, and that the sitemap lists what you think it lists.
Is it private?
Pasting a page's source keeps everything on your device: the parsing uses the browser's own DOMParser, and nothing is sent anywhere. Crawling is the deliberate exception, because a browser cannot fetch another origin's HTML. When you crawl, your browser sends the site's address to this site's own endpoint, which reads the pages you named, obeys the target's robots.txt and caches each page for ten minutes. The only thing recorded about you is how many pages your address has fetched in the last hour — the meter that stops one visitor turning the server into a crawler. Nothing is stored about the sites you audit, and no third-party service is involved.
Frequently asked questions
Because of the browser's same-origin rule. A page you paste is already in your hands, so the audit can be pure client-side. Reading pages you do not have means issuing requests to another domain from JavaScript, which the browser blocks unless that domain sends permissive CORS headers. Any tool that crawls your site either uses a server or an extension with extra permissions; this one uses its own PHP endpoint, and says so rather than hiding it.
From the site's own robots.txt `Sitemap:` lines and, failing that, from /sitemap.xml, /sitemap_index.xml and /wp-sitemap.xml. Sitemap indexes are followed into up to 20 child files. Nothing is guessed or discovered by following links, so if a page is not in the sitemap it will not appear — an incomplete sitemap is itself the finding.
A single audit takes up to 1,500 URLs from the sitemap and reads 18 of them per request. The browser stops asking after 600 pages and shows the 100 lowest scores, with the full list in the CSV. There is also a budget of 1,500 pages per visitor per hour, so a second large crawl in the same hour is refused rather than silently eaten.
It reads it first, picks the group that names this crawler or falls back to `*`, and skips any URL whose path a Disallow rule covers — reporting those URLs as skipped rather than quietly dropping them. A `Crawl-delay` is honoured too: the crawler drops to one request at a time and waits between them, up to two seconds per page.
The list at the bottom names the reason: an HTTP status other than 200, a response that is not HTML (a PDF or an image in the sitemap), a page robots.txt disallows, an address that did not answer, or a page that took longer than twelve seconds. A sitemap full of 404s is a real finding, not a tool failure.
They use the same twelve checks, the same thresholds and the same arithmetic, but they read the HTML differently — the crawler uses regular expressions, the pasted audit uses the DOM the browser builds. Odd markup can therefore move a page a point or two between them. Neither runs JavaScript, so tags injected at runtime are invisible to both; for those pages, copy the rendered DOM from the Elements panel and use the single-page mode.
Related tools
Written and maintained by Md. Nur Hossain Repon.