How to find broken links by hand, without a paid tool

A broken link is one of the few website problems that is trivial to fix and almost impossible to notice. Nothing turns red, no alert arrives, the page looks exactly as it did yesterday. There are four ways to find them without paying for anything: opening them by hand, a browser extension, Search Console, and a crawler on its free tier. Each one works, each one misses something specific, and this article says what.

The four free methods, compared

None of them is complete on its own. The combination that covers the most ground for nothing is simple: a free crawler for the overview, Search Console for what Google actually sees, and your own hands for the twenty pages that matter most.

1. By hand: the procedure, actually written down

"Open the pages and click the links" is not a method, it is an intention. Here is the one that survives three hours without you losing the thread.

1. List the pages in order of importance. Homepage, menu pages, your ten most-read articles, contact and pricing. The rest comes later, if there is time.
2. Open each page with the mouse, not the keyboard. Middle-click every link: it opens in a background tab without losing the page you are on. That detail is what makes the method bearable.
3. Deal with tabs in batches of ten. A glance is enough: either the page displays, or it shows an error.
4. Write it in a file, never in your head. Two columns: the page the link is on, and the broken link. The first column is the only one that lets you fix anything.
5. Do not forget the footer and the menu. They are on every page, which is exactly why you stop seeing them. Check them once, right at the start.
6. Count outbound links as well as internal ones. A link to a site that shut down breaks in exactly the same way, and it will appear in no automated report on your side: it is the most common blind spot.

This method is excellent up to about thirty pages. It has a merit the tools do not: you reread your own site, and along the way you find things no crawler ever reports, like a 2023 price or a link that points at the right page but describes it badly.

2. The browser extension

An extension that highlights broken links on the page you are viewing costs a minute to install and does real work: it turns step 3 above into a glance.

What it misses is structural: it only knows the page in front of you. It visits nothing, it keeps nothing, and it will never tell you that a link broke last week on a page you had no plans to open. It is an accelerator for manual checking, not a method in its own right.

A second, less known limit: many sites answer an extension differently from a normal visitor. A link flagged red while working perfectly in another browser is the most frequent case, and it produces pointless fixes.

In between sits the on-demand check of a single page: paste an address and see every link on it tested avoids the install, and covers exactly the need when you want to check one specific page rather than a whole site.

3. Search Console

Free, fed by Google's actual crawling, and underused. The indexing report lists URLs returning an error, with the date they were detected, and URL inspection tells you which pages point at them.

Two blind spots, both important:

• It only covers what Google chose to visit. A recent page, or a page no internal link reaches, may never appear at all.
• It ignores your outbound links entirely. It watches your pages, not your references. That is half the problem, and it is the half nobody looks at.

A third point is worth knowing: the delay. An error Google detected can take days to appear in the report, and weeks to leave it once fixed. It is not a verification tool, it is a record of what happened.

4. A crawler on its free tier

Several desktop crawlers will explore a site for free up to a few hundred URLs. For a fifty-page site that is plenty, and it is by far the best ratio of time spent to ground covered.

What the free tier withholds is almost always the same three things: the URL cap, the absence of scheduling, and above all the absence of history. You get a photograph, not a comparison. And a site does not need to be audited once, it needs to be watched, because links die continuously and without warning.

Where they come from, and why they come back

You rename an article and forget the three pages that pointed at it. You delete a category. A partner shuts down. A newspaper reorganises its archive. Each of these is reasonable on its own, and each leaves a link somewhere that now leads nowhere.

Two years of ordinary maintenance is enough for a fifty-page site to accumulate dozens. That is why a single check, however thorough, settles nothing durably: the stock rebuilds itself, and it rebuilds faster after a redesign or a migration, which are the two moments when everything breaks at once.

Reading the results without panicking

A report, even a free one, is longer than people expect, and most of it is not urgent. Sort it before you start:

• A page in your menu returning 404: fix today, it is on every page of the site.
• A link inside a popular article: fix this week.
• A link on a page nobody visits, to a page nobody visits: put it on the list, no urgency.
• A page reported as blocked rather than broken: often a site refusing automated clients. Open it yourself before doing anything.

Above all, do not fix everything. Not every broken link should be repaired, and a 404 is sometimes the honest answer: here is how to triage them. Incidentally, a pile of 404s is not a crawling problem either, whatever you may have read: crawl budget concerns almost nobody.

The point where doing it by hand stops being reasonable

It is not where people expect. It is not a question of page count, it is a question of repetition.

A full manual check costs roughly two minutes per page, so an hour for thirty pages and three to four hours for a hundred. Done once, that is acceptable. The problem is that it has to be redone: monthly for a site that changes, quarterly at the very least. Nobody blocks out three hours a quarter, so the check slips once, then twice, and becomes annual without anyone deciding it.

Three signs tell you the line has been crossed:

• you can no longer say when the last complete check happened;
• you find broken links by accident, while reading your own site;
• somebody else depends on the answer, a client for instance, and you cannot answer without spending an afternoon on it.

As long as none of the three is true, free and manual is enough, and nothing on sale will serve you better.

What free will not give you

This needs saying plainly, because it is the one piece of information missing from the usual comparisons. None of the four methods above runs by itself, none of them keeps last week's state to compare with today's, and none of them warns you. They all answer the question "where do I stand right now?", never "what changed since last time?".

It is exactly the same limit as an unwatched redesign, where links break all at once and nobody sees it. If that question concerns you, automation starts to earn its keep. If it does not, keep your money: a regular manual check beats a subscription whose reports you never read.