Website Link, Redirect & SEO Auditor

Check a website release or migration for broken URLs, redirects, missing HTML titles and canonical links. Start with up to 10 public pages or XML sitemaps. Export one row per URL actually checked, with HTTP status, redirect chain, final URL, referring pages discovered before the check, response time, and evidence for each issue.

Quick start

{
  "startUrls": [{"url": "https://example.com/"}],
  "maxPages": 10,
  "maxChecks": 25,
  "maxDepth": 2,
  "checkExternalLinks": false
}

For the public Example Domain page, a check returns title: "Example Domain" and a missing_canonical issue. Its link to iana.org is skipped with the default checkExternalLinks: false.

{"url":"https://example.com/","title":"Example Domain","canonical_url":null,"issues":[{"code":"missing_canonical","evidence":"No valid link rel=canonical"}]}

The snippet shows the source-backed fields; full rows also include HTTP status, final URL, redirect chain, referrers, robots meta, and measured response time.

What it checks

The crawler checks each normalized URL once. It follows at most five HTTP redirects, then records the final status and URL. It parses HTML links, title, canonical link and meta robots; XML sitemap loc entries can seed more checks. Same-origin HTML pages and sitemaps can enqueue links through maxDepth. External links are checked only if enabled and are never crawled for further links. A 4xx status other than access/challenge codes is reported as broken_http; 5xx as server_error. HTTP 401, 403, 407 and 429 are classified as access_or_challenge. DNS, TLS, timeout, blocked private address and oversized response failures are unreachable with evidence, rather than a fabricated 404.

This is deterministic technical QA on anonymous public HTTP(S) responses. It does not run a browser, assess visual layout or Core Web Vitals, verify search indexing, or promise SEO rank. Login, JavaScript-only links, robots directives beyond the exposed meta_robots field, and nonstandard ports are outside scope. It does not guess a URL's status when a fetch fails.

Limits and partial results

startUrls accepts 1–10 URL strings or { "url": ... } objects. maxPages defaults to 10 and permits 1–1000 same-origin checks. maxChecks defaults to 25 and permits 1–1500 total checked URLs. maxDepth defaults to 2 and permits 0–5. Responses are capped at 1 MB, each request has a 12-second timeout, the Actor has a 100-second internal deadline, and the transport allows at most 3000 requests including redirect hops. Redirects, page caps, and spending limits can produce partial coverage. OUTPUT in the default key-value store reports fetched, saved, skipped, error and duplicate counts, request count, and reached limits. The dataset contains only checked URL rows; no summary or placeholder rows.

referring_pages captures referring URLs known when the row is saved. Later-discovered duplicate links can increase the duplicates summary without changing an already-saved dataset row. Sitemaps themselves receive a URL-check row when fetched; their entries are separately checked subject to the same limits. Every checked URL may make multiple HTTP requests if it redirects, so the requests count can exceed the URL-check count.

Billing

The billing unit is one saved checked URL, event url-check. The current price is shown by Apify when you run the Actor. A discovered but unfetched link is not billed. The Actor stops checking URLs when the run's charge limit is exhausted.

Output fields

Each dataset row has url, referring_pages[], status_code (null if no HTTP response), redirect_chain[], final_url, title, canonical_url, meta_robots, issues[] with code and evidence, and response_time_ms (null if no response). title, canonical, and robots are null for non-HTML responses. Empty issues means none of the implemented checks found an issue, not a full SEO pass.

Access and safety

The Actor verifies HTTPS certificates and rejects private/local destinations, credentials, unusual ports, unsafe redirects, response bodies over 1 MB, and URLs over 2048 characters. It does not use AI or paid external APIs to produce checks.