What happens to the pages you no longer want

Every website that has grown carries content nobody needs any more. News from 2017. A campaign that ended years ago. A section someone once started and never finished.

During a rebuild comes the moment when it is decided what comes along. And then almost always the same thing happens: the pages are deleted, and with that the matter counts as settled.

It is not. Deleting is not an answer, only the removal of the answer. The address lives on — in search results, in bookmarks, in links from other sites, in an email from three years ago. Whoever opens it now gets something. The question is what.

There are four answers, not one

Redirect. There is a successor that answers the same question. The visitor lands there.

Report as gone. There is no successor, and that is a decision. The content is gone and will not come back.

Keep, but take out of search. The content stays reachable for visitors but should no longer appear in search results.

Not found. The leftover category. It is the weakest answer and should occur least often.

Most websites use only the fourth.

Redirecting — and the mistake in it

If there is a successor that fits in content, a permanent redirect is the right choice. The visitor lands at what they were looking for, and the value the old address built up over the years passes to the new one.

The mistake lies in the words in content.

It is tempting to send everything old to the home page. No error arises, the numbers look clean, and the work is done in five minutes.

For the visitor, though, it is a dead end with better manners. They clicked a link to a particular topic and are now standing in front of your welcome. They do not know what happened, and they do not find what they were looking for.

If there is no fitting successor, a redirect is the wrong answer. Then comes the second one.

Reporting as gone — the answer hardly anyone uses

There is a response code of its own for “this page existed, it was removed on purpose, it will not come back”. It is not the same as “not found”.

The difference is a statement about the future. “Not found” means: there is nothing here right now, maybe it was a typo, maybe it will come back. “Gone” means: that was a decision, stop looking.

For search engines that is a clear signal. An address reported as gone disappears from the index more quickly than one that is simply not found.

And for you it has a practical side effect: in the analysis after the rebuild you can tell apart what you meant to throw away and what is missing by accident. On a site full of “not found” you do not see that difference.

The important part here is the second one, which tends to be forgotten: the response code is for machines. The person who clicked the link still needs a readable page that says what happened and where they can go instead.

Whole sections at once

When a complete section goes — a news archive, an old blog, a discontinued area — you do not enter every single address.

Naming the directory is enough. One rule instead of two hundred lines, and everything beneath it is taken care of.

This is also the point where the decision is easier than many think: an archive of news items that has had hardly any visitors and no incoming links for years makes the site bigger and nobody wiser. Keeping it costs maintenance and brings nothing. Removing it is the simpler decision, once you make it consciously instead of postponing it.

Keep, but not in search

The case that also exists and is often overlooked.

Some content should stay reachable for visitors but does not belong in search results. A section with posts from users. An archive that is needed internally. Something someone opens through a direct link but is not meant to find through a search.

That is not a deletion. The content stays, it is just no longer indexed.

The difference matters, because the two cases are often confused. What a visitor still needs gets deleted. And what nobody needs any more stays in, because deleting feels too final.

The page nobody wants to land on

Even with careful work, a remainder is left. Typos in addresses, wrong links from other sites, old addresses nobody thought of.

These visitors get to see your error page. And on most websites that is the worst it could be: an empty surface with a number on it.

Whoever lands there is gone. They jump back to the search results, and the next hit gets their visit.

Yet the effort for a usable error page is small. It needs:

The normal navigation. So the visitor sees where they are and can move on.

A search. The most direct way to what they actually wanted.

A sentence in normal language. “This page no longer exists” is understood by everyone. A three-digit number is understood by nobody outside the trade.

A visible way to the home page.

And the same goes for the page that appears for deliberately deleted content. It is a page of its own with a statement of its own — not the same as the error page, because here you do know what happened and can say so.

A sentence on honesty

In the source part of this article comes from, the author says about the technical setup of this message that they had to experiment for a long time before it worked, and that several of the obvious approaches are only partial solutions.

That matches my experience, and I write it down because nobody else does. This thing is not hard to understand, but it is fiddly to set up — in different ways depending on system and server. Anyone reading a guide that does it in three lines should measure afterwards whether what it says really happens.

How to check what your site answers

You do not need a tool if it is only about one address.

Open your browser's developer tools with F12, switch to “Network”, and open the address. The first row of the list shows the response code. That way you see whether your site actually reports what it is supposed to report.

For a whole website it only works with a tool that fetches all addresses and records the answers. That is the same crawl you should do after a rebuild anyway — and the list of old addresses you hold it against is a topic of its own with an article of its own.

The one check that is always worth it: open some address that does not exist on your site. Add a few random characters to your domain. What you see then is what everyone coming through a broken link sees.

The decision behind it

The actual point of this article is not technical.

It is that “gone” is not a decision. For every piece of content that goes, there is someone who still opens it — and the question is what that person should get instead.

Four possible answers, one conscious choice per address. That is all. For a normal website it takes an afternoon, and it is the difference between a rebuild nobody notices and one after which visitors run into nothing for months.

Sources

• Udemy, Der perfekte Webseiten-Relaunch, phase 3 — that an error page matters in a relaunch and is usually neglected; that such requests arise through faulty internal links, typos, links from other sites and addresses that used to be in the index; that an empty error page sends the visitor back to the search results, because they learn neither where they have landed nor why; that a usable error page keeps the navigation, a search and a way to the home page; that a search engine does not take such an address out of the index immediately and visitors therefore keep arriving there; that deliberately deleted content needs both the fitting response code for the crawlers and a separate, readable page for the person; that a single entry is enough for a whole directory instead of one line per address; and that there is the further case of keeping content available for visitors while taking it out of the index, for example with user-generated content.
• Udemy, Der perfekte Webseiten-Relaunch, phase 2 — that the decision is made per address between staying reachable, redirecting and deleting, and that sections with hardly any visitors, hardly any impressions and no incoming links bloat the site unnecessarily.
• Taken over explicitly from the source — the author's statement that the practical setup of the deletion message took considerable effort and that several obvious approaches are only partial solutions.
• Own practice — the classification of the four answers as a conscious choice per address; the note that a redirect to the home page is a dead end with better manners; the distinction between deliberately removed and accidentally missing content in the analysis after the rebuild; and the check through the Network tab and through opening a made-up address.