SEO Audit & Broken Link Checker

Crawl any website and get, in one run:

• a technical SEO audit of every page: title, meta description, H1, image alt text, canonical, robots/noindex, hreflang, lang, viewport, thin and duplicate content, mixed content, slow and heavy pages, each with a 0–100 score;
• every broken link, broken image and broken script/stylesheet, with the pages that link to it and the anchor text;
• every internal redirect and redirect chain, so you can update links to point at the final URL;
• a site summary: issues ranked by how many pages they affect, duplicate title/description/content groups, sitemap quality (broken, redirecting, noindex and orphan URLs), robots.txt status and HTTP status codes.

Plain HTTP and no browser, so it is fast and cheap. $2 per 1,000 audited pages. Link checks, link rows, summaries, dry runs and failed sites are free.

Why this one

	This Actor	Typical SEO audit actors	Typical broken-link checkers
Price	$0.002 per page, no start fee	$0.005–$0.04 per page	$0.001 per link checked, often plus a start fee
Per-page SEO audit (44 checks)	✅	✅	❌
Broken links with source page and anchor text	✅ free	partial	✅ (paid per link)
External links, images, scripts and CSS checked	✅ free	rarely	some
HEAD first, then GET fallback (no false "405 broken")	✅	❌	some
403/429/999 reported as blocked, not broken	✅	❌	❌
Canonical target validated (broken, redirect, noindex)	✅	❌	❌
Sitemap quality: broken, redirecting, noindex and orphan URLs	✅	rarely	❌
Duplicate titles, descriptions and content, grouped	✅	partial	❌
Respects robots.txt	✅ (option)	varies	varies
Free dry run with a cost estimate	✅	❌	❌
Stops safely before the run timeout and still reports	✅	❌	❌

Quick start

1. Enter one or more websites, e.g. https://crawlee.dev.
2. Set Max pages per site (default 100). Not sure how big the site is? Tick Dry run first: it is free and tells you how many pages the sitemap lists and what a full audit would cost.
3. Run. Open the Pages overview, Broken links & redirects and Site summary tabs in the output.

{
  "startUrls": [{ "url": "https://crawlee.dev" }],
  "maxPagesPerSite": 200,
  "checkExternalLinks": true,
  "linkRows": "brokenAndInternalRedirects"
}

How it crawls

• It starts at your URL, reads robots.txt and the sitemaps (including sitemap indexes and .xml.gz files), then follows links breadth-first: closest pages first.
• It stays on the same site. www. and the bare domain count as one site. Subdomains are optional.
• It audits HTML pages that answer 200. Redirecting, broken and non-HTML URLs are recorded as link results and are not charged.
• Tracking parameters (utm_*, gclid, fbclid…) are always stripped. Ignore query strings also merges ?sort= style variants.
• Only crawl / never crawl patterns take plain text or /regex/.
• After crawling, it checks every unique URL found on the audited pages once. Canonical targets go first, then internal links, internal images/scripts/CSS, external links and external resources. It tries HEAD and falls back to GET when a server mishandles HEAD. It retries 429/503, honouring Retry-After.
• If the run is about to time out, it stops crawling early and still writes every row and the summary.

Not covered: pages that render their content only with JavaScript. These are flagged JS_RENDERED so you know. Also not covered: Core Web Vitals, and pages behind a login.

Output

All rows go to one dataset. rowType tells them apart, and the tabs filter the columns.

page row (charged, one per audited page)

{
  "rowType": "page",
  "url": "https://crawlee.dev/blog/page/2",
  "statusCode": 200,
  "score": 86,
  "errors": 0, "warnings": 2, "notices": 2,
  "issueCodes": ["TITLE_TOO_LONG", "TITLE_DUPLICATE", "META_DESCRIPTION_DUPLICATE", "CANONICAL_MISSING"],
  "issues": [{ "code": "TITLE_TOO_LONG", "severity": "warning", "message": "Title is 108 characters (over 60); search engines will cut it.", "count": 108 }],
  "title": "Crawlee Blog - learn how to build better scrapers | Crawlee …",
  "titleLength": 108,
  "metaDescriptionLength": 52,
  "h1": "Crawlee Blog",
  "h1Count": 1,
  "canonical": null,
  "canonicalStatus": "missing",
  "indexable": true,
  "inlinks": 3,
  "inSitemap": true,
  "brokenLinkCount": 0,
  "brokenLinks": null,
  "linksToRedirects": 0,
  "imagesMissingAlt": 0,
  "wordCount": 1210,
  "responseTimeMs": 182
}

Page rows also include: finalUrl, redirectChain, depth, foundOn, notIndexableReason, metaDescription, h2Count, metaRobots, xRobotsTag, lang, hreflang, images, imagesMissingAltSample, internalLinks, externalLinks, structuredDataTypes, ogTitle, ogImage, htmlBytes, contentType and crawledAt.

link row (free, one per problem URL)

{
  "rowType": "link",
  "url": "https://crawlee.dev/blog/img/lukasp.webp",
  "state": "broken",
  "statusCode": 404,
  "internal": true,
  "kind": "img",
  "linkedFromCount": 1,
  "linkedFrom": [{ "page": "https://crawlee.dev/blog/…", "anchor": null, "kind": "img" }],
  "checkMethod": "GET"
}

state takes one of these values:

• broken: 4xx/5xx, a redirect loop or too many redirects.
• error: DNS, connection, SSL or timeout failure.
• blocked: 401/403/429/999, meaning the server refused the checker. LinkedIn, Stack Overflow and Cloudflare-protected sites often do this, so check these by hand; they are not counted as broken.
• redirect: the URL works but redirects; the row shows the full redirectChain.

Link rows decides which states get rows.

summary row (free, one per site)

The summary row holds:

• pagesAudited, averageScore, pagesWithErrors, nonIndexablePages;
• brokenInternalUrls, brokenExternalUrls, blockedUrls, redirectingInternalUrls, linksChecked, urlsDiscovered;
• issues: every issue code with the number of pages it affects, worst first;
• duplicateTitles, duplicateMetaDescriptions, duplicateContent: groups of pages;
• sitemaps: URLs listed, plus broken, redirecting, noindex and orphan URLs (in the sitemap but linked from no audited page, checked only when the whole site was crawled);
• robotsTxt, statusCodes, stoppedBy (maxPages, costLimit, timeout or null).

The same summaries are saved as JSON in the key-value store record RUN_SUMMARY.

Issue codes

Score = 100 − 10 per error − 4 per warning − 1 per notice (minimum 0).

Code	Severity	Meaning
TITLE_MISSING	error	The page has no <title>.
TITLE_EMPTY	error	The <title> is empty.
TITLE_MULTIPLE	warning	The page has N <title> tags.
TITLE_TOO_LONG	warning	Title is N characters (over 60/160); search engines will cut it.
TITLE_TOO_SHORT	notice	Title is only N characters (under 30/70).
TITLE_DUPLICATE	warning	Same title as N other page(s).
META_DESCRIPTION_MISSING	warning	No meta description.
META_DESCRIPTION_EMPTY	warning	The meta description is empty.
META_DESCRIPTION_MULTIPLE	warning	The page has N meta descriptions.
META_DESCRIPTION_TOO_LONG	notice	Meta description is N characters (over 60/160).
META_DESCRIPTION_TOO_SHORT	notice	Meta description is only N characters (under 30/70).
META_DESCRIPTION_DUPLICATE	notice	Same meta description as N other page(s).
H1_MISSING	warning	No <h1> heading.
H1_EMPTY	warning	The <h1> is empty.
H1_MULTIPLE	notice	N <h1> headings.
IMG_ALT_MISSING	warning	N image(s) without an alt attribute.
CANONICAL_MISSING	notice	No canonical link.
CANONICAL_MULTIPLE	error	N canonical links with different targets; search engines may ignore all of them.
CANONICAL_RELATIVE	notice	Canonical uses a relative URL (…).
CANONICAL_INVALID	error	Canonical is not a valid http(s) URL (…).
CANONICAL_OTHER_DOMAIN	warning	Canonical points to another domain (…).
CANONICAL_TO_OTHER_PAGE	notice	Canonical points to a different page (…); this page will not be indexed itself.
CANONICAL_INSECURE	warning	Canonical uses http:// on an https:// page.
CANONICAL_TARGET_BROKEN	error	Canonical target returns ….
CANONICAL_TARGET_REDIRECTS	warning	Canonical target redirects to ….
CANONICAL_TARGET_NOINDEX	error	Canonical target is noindex.
CANONICAL_ON_NOINDEX	notice	Page is noindex but also has a canonical to another page (mixed signals).
NOINDEX	notice	Page is noindex (…).
NOFOLLOW	notice	Page is nofollow (…).
BROKEN_LINKS	error	N broken link(s) on this page.
BROKEN_IMAGES	warning	N broken image(s) on this page.
BROKEN_RESOURCES	warning	N broken script, stylesheet or frame URL(s) on this page.
LINKS_TO_REDIRECTS	notice	N internal link(s) go through a redirect.
PAGE_REDIRECTED	notice	The URL redirects (…).
REDIRECT_CHAIN	warning	The URL goes through N redirects.
LANG_MISSING	notice	No lang attribute on <html>.
VIEWPORT_MISSING	warning	No mobile viewport meta tag.
THIN_CONTENT	notice	Only N words of text.
CONTENT_DUPLICATE	warning	Exactly the same text as N other page(s).
MIXED_CONTENT	warning	N resource(s) loaded over http:// on an https:// page.
SLOW_RESPONSE	notice	Server took N ms to respond.
LARGE_PAGE	notice	HTML is N KB.
JS_RENDERED	notice	Very little text in the HTML; the page is probably rendered by JavaScript, which this audit does not run.
HREFLANG_INVALID	warning	Invalid hreflang value(s): ….

Thresholds: title 30–60 characters, meta description 70–160, thin content under 200 words, slow over 3 s, large HTML over 3 MB. Pages that are noindex or canonicalised elsewhere are left out of duplicate checks.

Pricing

$0.002 per audited page saved ($2 per 1,000). Nothing else is charged:

• link checks and link rows: free, however many;
• summary rows and the RUN_SUMMARY record: free;
• dry runs: free;
• sites that fail to open: free;
• broken, redirecting and non-HTML URLs: free (they are not audited pages);
• with Save only pages with issues, pages without issues: free.

Set a maximum cost per run in Apify and the Actor stops before going over it. The summary then shows stoppedBy: "costLimit".

Examples: a 50-page brochure site costs $0.10. A 1,000-page blog costs $2. A 10,000-page shop audited on 2,000 sample pages costs $4.

Tips

• Big sites: run a dry run, then audit a sample with Max pages per site and an Only crawl pattern per section (e.g. /blog/).
• Shops and faceted navigation: tick Ignore query strings. Exclude ?filter=, /tag/ and similar.
• Fragile servers: lower Max parallel requests to 2–3.
• Only want broken links? Set Link rows to "Broken links only" and Save only pages with issues. Use the Broken links & redirects tab.
• Scheduled monitoring: schedule a weekly run and compare summary rows, or connect the dataset to Google Sheets, Slack or Zapier via Apify integrations.

FAQ

Does it render JavaScript? No. It reads the HTML that servers send, which is what most search-engine first passes see. Pages whose text appears only after JavaScript are flagged JS_RENDERED.

Why is a link "blocked" rather than broken? Some sites (LinkedIn returns 999, Stack Overflow 403, many Cloudflare sites 403/429) refuse automated requests even though the link works in a browser. Calling these broken would be wrong, so they are reported separately.

Does it obey robots.txt? Yes, by default, for internal pages. External links are checked with a single request each.

Is it legal? It requests public pages, like a search-engine crawler, at a polite concurrency. Audit sites you own or are allowed to audit.

Support

Open an issue in the Actor's Issues tab with the run ID. Feature requests are welcome.