SEO Audit Crawler: SEO Checker, Website Audit & Broken Links
Pricing
from $7.00 / 1,000 page auditeds
SEO audit and site audit for your own website: crawl it from a start URL or sitemap and get a 0-100 SEO score per page and site from a published formula, plus broken links and images, redirects, title and meta description length, H1s, canonical, robots, hreflang, Open Graph and JSON-LD.
Versions follow MAJOR.MINOR.PATCH (src/version.py); Apify shows MAJOR.MINOR from .actor/actor.json.
Every run logs its version and records it in the RUN_STATS key-value record.
RUN_STATS keeps the reason as before.humble-echidna-<name>, and if that is
refused too it runs without memory for that run. Users' existing memory is untouched: the usual name is tried first.Retry-After answers: when a site or API asks it to wait before retrying, it now waits up
to 90 seconds (was 30), once per site, with at most 3 minutes of such long waits per run; a longer wait, or a
site that asks again after it was waited out, is reported with the reason instead of stalling the run.
A link check still has 60 seconds in all, so a site asking for longer than fits is reported as rate-limited at once.startUrls is no longer marked required, so AI agents (Apify's MCP server) see that a dataset alone (datasetId) is enough. Nothing else changes: a run with neither still stops with a clear message.Audit the websites in another actor's results, e.g. Google Maps leads with weak websites.
datasetId, a dataset picker with read access) and Field with the
website (datasetUrlField). With a dataset, each item's website becomes a start URL and startUrls is ignored;
with Crawl depth 0 that's one home page per business, and at higher depths every page found through a website
carries its source fields. The field is found automatically when left empty: the first of website, url,
domain, site, homepage with a website in the first 100 items, then the same names one level down
(contact.website); a dotted path names a nested one.N/A, an email) and Google
Maps links (the place's own url in a Maps scraper's output) are skipped, never fetched or charged, and counted in
the new RUN_STATS.dataset record (itemsRead, websites, withoutWebsite, notAWebsite, googleMapsLinks,
duplicates, the field used and whether it was found automatically). Tracking parameters (utm_*, gclid,
fbclid, ...) are dropped from the address. A dataset with no website at all, or one that can't be read, fails
the run with the reason.sourceTitle (the item's title, else name), sourcePlaceId (its
placeId) and sourceIndex (its position in the dataset, 0 = the first), to join results back to the leads;
null for URLs typed in. A new "Leads (from a dataset)" dataset view shows them next to the key columns.mms_common.dataset_input), as are its test stand-ins (mms_testing).http-error, request-failed), never below 0. slow-response isn't counted (response times vary run
to run). New field score; SUMMARY gets siteScore (the average over every page audited, halves rounded up),
scoreBands (90-100, 50-89, 0-49) and lowestScorePages; RUN_STATS gets siteScore; the status message
shows it. The README's issue table has the points for every code, checked by a test.imagesEmptyAlt (alt="" present but empty), imagesMissingSize (no width
and height attributes; new notice images-missing-size), imageLinksWithoutText (a link whose only content is
images without alt text; new warning image-links-without-text). With the new Check images input (on by
default), each distinct <img src> gets one HEAD request per run (GET only when HEAD answers an error), wherever
it's hosted, counted against Max link checks per run: brokenImages / brokenImageCount (new error
broken-images) and oversizedImages over 200 KB by Content-Length (new warning oversized-images);
imagesNotChecked counts the rest. SUMMARY lists broken and oversized images once each. RUN_STATS counts
imagesChecked apart from linksChecked. Not measured, and the README says so: files without a Content-Length,
pixel size, format and compression, srcset/<picture>/lazy-loading/CSS images.compareWithPreviousAudit / onlyChangedPages): a page whose score changed has score in
changedFields and previousScore; since the score is computed from the compared issue codes, a score change
always comes with an issue added or fixed, so it counts as a change (and in the only-changed mode the page is
returned). SUMMARY.comparison gets siteScore, previousSiteScore and siteScoreChange. Broken and oversized
images are carried like broken links when an image had no clear answer (not checked, or refused). An audit
remembered by 1.1 isn't flooded on upgrade: the image issues found on a page the first time it's compared again
are recorded, not reported as added (pages the crawl doesn't reach stay marked until it does). The memory record
gains checks, legacy and siteScore; its format is otherwise unchanged, and turning Check images off is a
separate audit (its own baseline), while the default keeps 1.1's comparison.compareWithPreviousAudit, off by default): each page gets changeType (new,
changed or unchanged), changedFields (status code, redirect target, title, meta description, H1s, canonical
URL, noindex, broken links), changedIssues (issue codes added and fixed, broken links new on the page, how many
were fixed) and previousStatusCode. The SUMMARY record gets a comparison: the previous audit's date, run and
dataset IDs; pages new, changed, unchanged and removed; issues added and fixed per code; new and fixed broken
links. The first run with it on is the baseline (nothing reported as new).onlyChangedPages, off by default; turns the comparison on): new and changed
pages are written and charged as usual ($10.00/1k); unchanged pages are still audited and counted in SUMMARY but
not written, and are charged the new unchanged-page event, $1.00/1k (a custom event, not a dataset row, so
never billed twice; charged in batches of 100; never for a page that refused, failed or answered 5xx again). The
baseline run charges every page as a row. In this mode the maximum cost per run limits the pages read: before a
page is read, the least it can cost is reserved (the unchanged price, or the page price on the baseline); a
changed page is written only if the rest of the page price fits, else it's reported again next run. Nothing is
reserved for unchanged pages while the event isn't priced on Apify.seo-audit-memory
key-value store: per page the status code and short fingerprints of the fields, issue codes and broken links, no
page content (about 200 bytes a page). Only pages the user got (written, or unchanged in the only-changed mode)
update it, so a page cut by the maximum cost per run is reported again next run.pagesRemoved is null,
incompleteReasons says why, and those pages stay remembered. A broken link without a clear answer this run (not
checked, refused, rate-limited) isn't reported fixed. slow-response isn't compared (timing noise).null (or empty) and SUMMARY.comparison is null.RUN_STATS gains compare, onlyChangedPages, baseline, pagesNew, pagesChanged, pagesUnchanged,
pagesRemoved, unchangedNotReturned, crawlComplete and incompleteReasons; pagesAudited now counts
unchanged pages that weren't returned (pushed is still the rows charged); unchangedPagesCharged and
unchangedPagesFree count the rechecked pages charged and not charged.common/mms_linkcheck, so Bulk URL & Broken Link Checker
uses it too. Same results; one difference in timing: a worker no longer waits for a busy host while another
host's link is ready.First release.
Sitemap: lines, else /sitemap.xml; the same discovery as Sitemap URL
Extractor), up to a crawl depth and a max pages per run.<html lang>, hreflang alternates, Open Graph and Twitter tags present, JSON-LD @type values (types only),
image count and alt coverage, internal/external link counts, word count, broken links, and a list of issues with a
severity (error, warning, notice).SUMMARY key-value record (not charged): status codes, issue totals per severity and per
issue, worst pages, duplicate titles and meta descriptions, every broken link with the pages it's on, slowest
pages, averages, structured data types, and the pages that were never requested.apify-default-dataset-item event. Max pages per run and the
maximum cost per run are honoured before a page is requested.