Every Link on Your Site, Checked Once

Links decay on their own

Nothing has to go wrong for a link to stop working.

A page you linked to gets renamed. A supplier redesigns their site. An association changes its address. A document you referenced is moved to a different folder.

None of that is anybody's mistake, and all of it happens continuously, so a site that was perfect at launch has a handful of dead links eighteen months later without anyone touching it.

Why nobody tells you

Because reporting a broken link requires a visitor to care enough to find your contact details and write to you about a problem that is not theirs.

Almost nobody does that. They press back and continue, or they leave.

The result is a fault that produces no signal at all, which is why link checking has to be a scheduled job rather than a response to complaints.

Two different problems

Broken links divide cleanly, and the two halves deserve different amounts of your attention.

Internal links point to your own pages. When one breaks it is your doing, it is entirely within your control, and it is nearly always a typo or a page that was renamed without the links being updated.

External links point at other people's sites. When one breaks somebody else made a decision, and your options are to find where the content moved, replace it with something else, or remove the link.

Fix all the internal ones. Triage the external ones.

How to check them

A crawler starts at your homepage, follows every link it finds, and records what each one returned.

Desktop crawling tools will do a small site in a few minutes and a large one in under an hour. Free versions typically cap the number of pages, which is more than enough for most business sites.

Your site's own tools may also flag some of this, though they generally report only what a search engine happened to encounter rather than everything that exists.

Run the crawl, export the results, and work from the export rather than from the screen.

Reading the report

The report will be longer than you expect and most of it is not urgent.

• Not found responses are the real breakages.
• Server errors mean the other end is having a problem, possibly temporary.
• Timeouts often mean a slow site rather than a dead one.
• Redirects are working links, though worth noting.
• Blocked responses sometimes mean the site is refusing crawlers, not that the link is dead.

Check a sample by hand before mass-editing, because a crawler reporting a failure is not the same as a person finding one.

Fix the internal ones first

They are the shortest work and the clearest benefit.

A broken internal link is a visitor you had, on a page you control, who was going somewhere you sent them, and did not arrive.

Most of them come from three causes: a page renamed without updating what pointed at it, a typo in a hand-written address, and a link to a page that was deleted at some point without anybody checking what referenced it.

All three are fixed by editing the link, and a site with fifteen of them can usually be cleared in an hour.

The external ones need judgement

For each dead external link, decide which of three things is true.

The content moved, in which case find the new address and update it.

The content is gone but the point still stands, in which case find another source that says the same thing.

Or the link was never important, in which case delete it and move on rather than spending twenty minutes finding a replacement for a passing reference.

A worked example

A trade association site crawled at just over four hundred pages and produced ninety-one failing links.

Thirty-one were internal. Most traced to a single renamed section from a redesign two years earlier, where the pages had been moved and about half the links updated.

Forty were external and pointed at member businesses, a third of which had simply changed their web addresses.

Twenty were duplicates: the same broken link in a footer that appeared on every page, counted once per page.

That last category is worth knowing about, because ninety-one sounds like a catastrophe and the actual work was one footer edit, one batch of internal corrections, and an afternoon of checking member addresses.

Total time was under four hours, and the footer fix alone accounted for a fifth of the report.

Redirect chains

The crawl will also show links that work but do so through a chain of redirects, each one a small delay.

A link pointing at an address that redirects to a second address that redirects to a third is functional and slightly wasteful.

These are not urgent and they are cheap to fix while you are already in there: point the link at the final address and remove the intermediate steps.

What the crawl will not find

A crawler follows links in your pages, which leaves several places it never reaches.

Links inside PDFs and downloadable documents, which are frequently the oldest and most neglected links you own.

Links in email templates and autoresponders, which sit in a separate system entirely and can point at pages you deleted years ago.

Links in printed material and on signage, including any short address or code that redirects somewhere, since those fail silently and nobody can press back.

And links buried in menus that are built by scripts rather than written into the page, which some crawlers see and some do not.

None of these are difficult to check. They simply have to be checked deliberately, because the automated pass will report a clean result while they are quietly broken.

The counter-case

You can overdo this.

Chasing every external link on a large site to a perfect replacement is an enormous amount of work for a small return, and it is work that will need doing again next year regardless.

A reasonable standard is that internal links should be perfect, links in your main navigation and footer should be perfect, and links buried in old articles should be correct where it is cheap to make them so.

Removing a dead link is an acceptable fix. Not every reference needs a successor.

Making it happen again

1. Crawl the whole site and export the results.
2. Spot-check a few by hand before editing anything.
3. Find repeated links in headers and footers first.
4. Fix every internal break.
5. Triage the external ones: update, replace, or remove.
6. Note the date and put the next one in the calendar.

Twice a year is enough for most sites, and the second run is always shorter than the first.

Links that work but say nothing useful are a separate problem, covered in links that all say read more.

Frequently asked questions

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.