How to Scan a Whole Website for Broken Links

Start from one URL, follow internal links, use the sitemap, pause, export. A practical whole-site workflow in Chrome without a desktop crawler.

A whole-site scan is not “click every page.” It is a crawl: start URL, queue of internal links, optional sitemap, a report of statuses plus where the href was found.

Before you start

Use a URL the public (or your staging user) can actually open. Login walls will look like failures or miss everything behind them.
Decide a page limit so a calendar with infinite dates does not become your weekend.
Know that JavaScript nav only exists if the crawler renders. Browser crawls do. Dumb HTTP crawlers often do not.

In Chrome

Install Advanced Broken Links Search (currently free).
Open any page on the domain.
Start full-site crawl / the scanner window.
Let it read links and sitemap. Pause if you need the machine.
Filter to 4xx, export CSV, open a source page with highlight.

That is the same extension as a one-page scan. There is no second SKU.

Speed and false positives

Three speed presets — Slow / Normal / Fast — control how many pages and links get checked in parallel. On a real 1,100+ page site, a full crawl took 6–9 minutes depending on the preset. If the server briefly rate-limits the crawl (429 or 503), a link is not marked broken right away — it gets a couple of retries with a short backoff first, and only stays in the report as an error if the server keeps rejecting it.

What it will miss

URLs only in ads or email, never linked, never in the sitemap.
Other people’s sites (outbound is tested, not crawled).
Files your robots.txt forbids if you respect robots.

After the report

Group by destination URL. One dead product linked from 40 posts is one redirect, not 40 tasks. Flatten redirect chains. Recheck 5xx. Then ship.

If the site is 80,000 URLs of enterprise CMS, you may still want a desktop crawler. If the site is everything you can browse this afternoon, you do not. We said the same on the Screaming Frog comparison.