Deleting Pages on Purpose

Last year I deleted 140 pages off one of my own sites in a single afternoon. Over the next three months, organic traffic to that site went up by about a fifth. That sounds like a clean story about pruning. It isn’t, and it’s worth taking apart properly, because content pruning is one of the few SEO tactics where the advice out there is confident and the mechanism underneath it is a lot less certain than people let on.

What pruning actually means

You go through a site, find pages that aren’t earning their place, and remove them. Sometimes that’s deletion. Sometimes it’s a redirect into a better page. Sometimes it’s leaving the page up and telling search engines to stop indexing it. The tactic is old, and it comes back into fashion every couple of years.

The argument for it goes roughly like this: a search engine forms some view of a site overall, low quality pages drag that view down, and cutting them lifts what remains. There’s a version of this Google has said out loud, in the sense that they’ve talked about site-wide quality signals existing. There’s a stronger version people repeat, where every weak page is actively costing you rankings elsewhere, and that one is a lot further ahead of the evidence than its confidence suggests.

What I can tell you from running the sites is narrower and more useful. Pruning has produced real gains on large sites with a lot of near-duplicate or programmatic pages. On small sites, under a few hundred pages, I’ve run the same exercise several times and measured essentially nothing.

So the first question isn’t which pages to cut. It’s whether you have the kind of site where cutting helps at all.

Finding candidates

The method I use takes about an hour and starts in Search Console. Export the performance data on the page dimension over the full sixteen month window, which is the longest range it holds. That gives you every URL that got at least one impression. Then run a crawl of the site with whatever tool you have, which gives you every URL that exists.

Join those two lists and the interesting set falls out immediately. Pages that exist but appear nowhere in the impressions export haven’t been shown to a single person in sixteen months. That’s your starting pile.

Here’s where most people go wrong, so I’ll be blunt about it: zero impressions does not mean zero value. I’ve nearly deleted pages that were quietly holding a dozen referring domains. I’ve come close to cutting a page that got no search traffic at all and converted better than anything else on the site, because people reached it from a newsletter.

So before anything gets deleted, run the candidate list against two more things: your backlink data, to catch pages other sites point at, and your analytics across all channels rather than search alone, to catch pages that earn their keep some other way. What survives that filter is a list you can actually act on.

The other thing worth checking is internal links. A page nobody visits might still be a hub that passes links around your own site sensibly. Crawl tools will show you inbound internal links per URL, and anything with a lot of them needs a plan rather than a delete key.

Four options, and picking correctly matters more than the pruning

You have a list. For each page there are four options.

Deletion, with a 410 status code on the URL. That code says the page is gone deliberately. A 404 says it isn’t there, which leaves open the possibility of a mistake. In practice both get the URL dropped from the index, and 410 tends to get there faster. Use it when the content has no successor and nothing points at it.

Redirect it into a related page with a 301. This is right when a better page covers the same ground and someone landing on the old URL would be well served by the new one. The relevance test is the whole thing. Redirect a page about one narrow topic into a genuinely related page and the redirect is respected. Redirect it into something unrelated and search engines treat it as a soft 404, which is the same outcome as deletion with extra steps. The classic version of this mistake is redirecting every pruned page to the homepage. It feels tidy and it accomplishes nothing.

Keep the page and add a noindex tag. This suits pages that serve real users but have no business in search results: thank you pages, internal tools, filtered views that would compete with the pages you actually want ranking.

Merge. Take two or three thin pages covering overlapping ground, write one genuinely good page that covers all of it, redirect the old URLs into it, and update every internal link that pointed at them. This is the option that produces the results people attribute to pruning, and it’s the most work by a wide margin.

That last part is what I most want to land. When someone tells you they pruned a site and traffic jumped, ask what else they did that quarter. In my experience the answer is usually that they rewrote and consolidated the surviving pages at the same time. The rewrite is doing the heavy lifting. The deletion is the housekeeping that made the rewrite tractable.

The cleanup nobody does

Updating internal links is the step everyone skips. After a pruning pass, a site is typically full of internal links pointing at URLs that now redirect. It works, and it’s sloppy, and it makes the next audit harder to read. Run a crawl afterward, find internal links that resolve through a redirect, and repoint them at the destination. An afternoon of that leaves the site in a much better state than the pruning did.

Navigation and sitemaps need the same treatment. Pull deleted URLs out of your XML sitemap, because a sitemap listing pages that return 410 is telling search engines something you don’t mean. Check your category pages and any link lists you maintain by hand.

Tags, archives, and where WordPress breeds junk

Tag and archive pages deserve their own pass, because on WordPress that’s where junk URLs breed. A site with 400 posts and tags applied loosely can generate hundreds of near-empty archive pages, each one a thin list of two posts. Those rarely justify a slot in the index. The fix here is usually not deletion, since the pages do serve navigation. It’s a noindex on tag archives plus a decision about which taxonomy genuinely earns its place in search. Keep categories indexed if someone maintains them and they have real introductions written on them. Leave tags out. Date based archives almost never earn a slot either, and most SEO plugins have switches for all of this that take about thirty seconds to set.

E-commerce runs differently

An out of stock product page is not a pruning candidate the way a weak article is. If the product is coming back, keep the page up, say plainly that it’s unavailable, and put alternatives on it. If it’s gone for good and there’s a direct successor, redirect to the successor. If it’s gone with nothing comparable, redirect to the parent category. Deleting product URLs that have accumulated reviews, referring domains, and years of ranking history is one of the more expensive mistakes I’ve watched people make, and it nearly always happens because some bulk tool made it a single click.

Dated posts are the awkward middle case

A news item from 2019 about a tool that no longer exists is doing no harm and earning nothing. My default there is to leave it indexed, put a line at the top saying when it was written and what has changed since, and move on with my day. That costs ten minutes and it’s honest. Delete those in bulk only when there are hundreds of them, which on a site that used to run a news section there usually are.

Timing and measurement

Don’t run a pruning pass while a core update is rolling out. Rankings move for reasons unrelated to anything you did during those windows, and you’ll end up unable to tell what your own change accomplished. Wait for the update to finish, then act, then measure.

Measure over a proper window: eight to twelve weeks, minimum. Deindexing takes time, the effects are gradual, and two weeks of data after a structural change will tell you nothing except how noisy your traffic is.

How I actually run it now

One practical note on how to run the thing. I deleted 140 pages in one afternoon, and I wouldn’t do that again on a site that matters to me. Doing it in a single movement means that whatever happens over the following quarter, you can’t attribute any of it. Now I work in tranches: thirty or forty URLs, then a wait, then the next tranche. It takes longer and it produces something you can learn from, which is the whole reason for measuring at all. On a site you’ve just bought and have no attachment to, go ahead and do it in one pass. On your main property, be patient about it.

What pruning will not do

It won’t make a page rank that deserves to rank and currently doesn’t. If you have an article aimed at a competitive query sitting on page four, deleting its neighbors is not the intervention. That page needs to be better, or it needs links pointing at it, or the query needs to be one you can realistically win with the authority you have. Pruning is maintenance. It clears the ground so you can see what you’ve got. Treating it as a growth tactic is how people end up removing perfectly reasonable pages and then wondering why the number never moved.

Keep a record

I keep a simple sheet with the date, the URLs removed, what happened to each one, and why. A year later, when someone asks what happened to a page, or when I’m trying to work out whether a pruning pass actually helped, that sheet is the only thing that makes the question answerable. Reconstructing it afterward from a crawl archive is miserable.

The one scenario where pruning is the strongest call

An inherited site often has hundreds of pages published by a previous owner to hit a volume target. That’s the strongest case for pruning I know of, and it’s also the case where you need the backlink check most, because bought sites frequently have links pointing at pages that look worthless from the inside.

Where that leaves the tactic

Pruning is a genuinely useful piece of maintenance that has been oversold as a growth lever. It makes a site easier to work on. It removes pages that would embarrass you if a client read them. On a large site with a lot of duplication, it does appear to move the needle. On a 200 page site, the honest answer is that your time is better spent making eight pages excellent than deleting forty.

I can’t tell you it will work on your site, and anyone who does is guessing. What I can tell you is how to find the candidates, how to avoid deleting something valuable, and how to structure the change so that in three months you can actually tell what it did.

For more walkthroughs like this, straight from the sites we actually run, head back to the homepage.