Link checking

What link checking is

Link checking is a scheduled sweep of every link on your website — the ones pointing at your own pages and the ones pointing at other people's — followed by the repair of the ones that no longer work. It runs on a set day, monthly or quarterly, and it ends with the site fixed. You are not handed a spreadsheet of errors to deal with later.

It is not link strategy and it is not link wording. Deciding which of your pages ought to point at which others is internal linking, a planning job done once and then followed. Replacing every “click here” with words that say where the link goes is descriptive link text, an editing job done page by page. This is the maintenance sweep, and it asks one question of each link: does it still land somewhere, and if not, where should it land now.

When you need it

Any site more than about three years old has dead links in it. They rot quietly. A supplier rebuilds their website and every deep link into it breaks at once. A partner closes. A product is discontinued and its page deleted, while the six pages linking to that product are never touched. Somebody renames a page in the CMS on a Tuesday afternoon.

It is worth doing on a schedule if your pages cite outside sources: a trade association, a provincial regulation, a manufacturer's specification sheet. Those move most often, and they are the links a reader is most likely to follow, because they are the ones you offered as proof. It is worth doing immediately after a site rebuild, and after any round of deleting or merging pages.

You do not need to buy this separately if you already have the whole pipeline run for you or a content maintenance plan, because a link check is already inside both. Buy it on its own when you publish for yourself and simply want somebody to deal with the rot twice a year.

What gets checked

• Internal links. Every link from one of your pages to another, including the ones buried in old blog posts. Anything landing on an error, and anything landing on a page that has been replaced by a better one.
• Outbound links. Links to other people's sites, checked for errors and also for the quieter failure: the link still works but now lands on a parked domain, a login wall or a page about something else entirely.
• The furniture. Navigation menus, footers, sidebars, calls to action and buttons. A broken link in the footer is on every page you have, and it is the one nobody thinks to test.
• The places a checker cannot see. Links inside downloadable PDFs, email templates and scheduled newsletters. These are checked by hand, because no automated tool opens them.
• Redirect chains. Links that work, but only after two or three hops. They are tidied to point straight at the destination.

What you get

The fixes, made. Internal links repointed at the right current page. Outbound links either updated to the source's new address, swapped for a better source, or unlinked with the reference left in the text where there is no live equivalent. The work is done in your CMS, not described to you.

A short list of decisions. Some broken links are not mine to settle — a link to a service you no longer offer, or to a partner you have parted company with. Those come back as a handful of questions with a recommendation against each, answered in a five-minute reply.

A one-page record. What was checked, how many links were found broken, what was fixed and what needed your decision. One page per sweep, filed so that the pattern over a year is visible.

A worked example

An illustration, not a client. A Halifax accountancy practice has about ninety pages, half of them blog posts written over eight years explaining tax changes as they happened. A great many of those posts link out to federal and provincial pages that have since been reorganised, and the practice's own service pages link to a fee schedule that was replaced two years ago.

The first sweep finds sixty-odd broken links. Most are outbound to government material, and are repointed at the current equivalent page. Eleven are internal links to the old fee schedule, all repointed at the current one. Four posts link to a seminar series the practice stopped running, and those are the decisions sent back to the owner, who asks for the links removed and one sentence added saying the seminars ended. Three posts turn out to be about rules that no longer exist at all, which is a different problem, and goes on the list for a content cleanup. The next quarterly sweep finds four broken links, which takes under an hour.

How it runs

1. A half-hour call. How big the site is, what it runs on, how often you want the sweep, and whether PDFs and email templates are in scope. No charge.
2. A fixed price in writing. Per sweep, or per year for a schedule, agreed before anything starts.
3. Access. An editor login to the CMS, and copies of any PDFs or email templates you want included.
4. The sweep. Two to four working days for a first pass on a site of a few hundred pages, because the first one is always the big one. Later sweeps are usually a day.
5. The fixes and the record. Repairs made, decisions sent to you, the one-page record delivered, and the next sweep put in the calendar.

What it costs

Quoted in CAD after the half-hour call, because the size varies enormously: a forty-page site and a twelve-hundred-page site are not the same job. The first sweep costs more than the ones that follow it, and a standing quarterly arrangement is quoted as a single annual figure. The pricing page shows the four ways of working, and link checking is included in the monthly Content Engine rather than billed on top. Whatever the number is, it is fixed in writing before I start.

What happens next

Two things usually come out of the first sweep. Pages that turned out to be broken in more than their links go for an update of the stale dates, prices and names. Pages that should not exist at all go into deleting, merging and redirecting the old material. Most owners then put the sweep on a schedule as part of a plan that gives every page a review date, so it never builds up again. None of that is assumed. A link check is a complete piece of work on its own, and if you want one sweep and nothing else, that is what you get.