Why prune: pages that serve no purpose have a cost
A website piles up pages over the years: sold-out products, duplicate variants, town pages with no content of their own, 200-word articles written for one keyword, pages generated by a plugin. Each one takes Google crawl time and, above all, says something about the overall quality of the site.
Google says so in its guide to ranking systems: its systems work at page level, but also use site-wide signals. Thousands of empty or near-identical pages therefore weigh on the good ones. And its spam policies specifically target producing many pages with the main purpose of manipulating rankings.
Pruning is not a punishment for the site, it is a sort: fewer pages, each one useful, better linked.
The sorting rule: keep, merge, remove
This is the rule we apply on our own sites, page by page, from the last 90 days of Search Console:
- Keep: a page that got at least ten or so impressions in 90 days, or that carries unique content (a data point, a calculation, a product on sale). Improve it rather than touching it blindly.
- Merge with a 301: several pages that answer the same intent (the same question asked differently, a colour variant, the same town on two pages). First move the useful content to the page that stays, then redirect the others to it.
- Remove with a 410: a page with no impressions at all in 90 days and no content of its own. It returns the 410 code (“gone”) and leaves the sitemap and the internal links.
- Never: mass-redirect to the home page. A redirect must lead to a page that answers the same question; otherwise the visitor arriving from an old link lands on a page that is no use to them, and the redirect no longer means anything to Google either.
301, 410, 404: what Google really does with them
A 301 (permanent) redirect says the page has moved. Google follows the redirect and uses it as a strong signal that the target should be treated instead: the new URL is the one shown in the results. Google prefers server-side redirects, which it interprets better than JavaScript redirects.
Avoid chains: by default Google follows up to 10 redirect hops, but every hop slows down crawling and the visitor. An old URL redirects straight to the final page.
410 and 404: Google says it clearly, all 4xx errors (except 429) are treated the same way. A page that returns a 404 or 410 is not indexed, and an already indexed page that starts returning one of these errors is removed from the index. We use the 410 because it states the intent (permanent removal) to other search engines and to our own tools, but the 404 does the same job for Google.
Do not block a page you are removing in robots.txt: Google could no longer see that it has gone. In an emergency (a page that should never have been public), the Search Console removals tool hides it for about six months, long enough to remove it for good.
- Apache (.htaccess): Redirect 301 /old-page /new-page
- Apache, removal: Redirect gone /removed-page
- Nginx: location = /old-page { return 301 /new-page; }
- Nginx, removal: location = /removed-page { return 410; }
Our case: more than 8,000 pages removed, and what changed
On a van comparison site, we removed 8,194 pages with a 410 on 23 September 2026: every page that had received no impressions in 90 days, about 87% of the site’s URLs. Then we merged 378 pages with a 301 on 28 September 2026 (the town × vehicle type variants grouped into their town page). 810 indexable pages remain.
What happened, over comparable periods: 104 impressions a day over the 28 days before, 98 over the 25 days after. The average position went from 69.4 to 64.5, and the site appears on 665 queries against 663 before.
Our reading, without dressing it up: removing most of the URLs cost nothing. The removed pages were not doing any work. But pruning alone did not bring back the traffic lost over the summer: it clears the ground, it does not replace better pages.
Over the same period, 3 other sites of ours, which had merged their variants with 301s (no 410) just before a Google update, came back clearly. The cause is not proven, and other pruned sites stayed at the bottom. We take caution from it, not a recipe.
- impressions per day
- 104
- impressions per day
- before, then 98 after
- average position
- 69.4
- average position
- before, then 64.5 after
- pages with at least one impression
- 239
- pages with at least one impression
- before, then 189 after
Pruning, step by step
- Export: every URL of the site (sitemap or crawler), with impressions, clicks and position over the last 90 days from Search Console.
- Sort: keep, merge, remove, using the rule above. For each merge, note the destination page.
- Merge first: move the useful content to the page that stays, then set up the 301s.
- Remove: a server-side 410 code, removal from the sitemap, removal of every internal link to these pages.
- Publish fresh sitemaps, split by page type, and resubmit them in Search Console and Bing Webmaster Tools.
- Measure after 4 weeks: impressions per day, position, pages with impressions. Compare equal periods.
Online shops: sold-out products and variants
- Sold-out product that will come back: keep the page, show “back soon” and offer alternatives.
- Discontinued product with a direct replacement: 301 to the replacement.
- Discontinued product with no replacement but with visits: 301 to the closest category, provided it answers the same search.
- Discontinued product with no visits and no links: 410.
- Colour or size variants on separate pages: a single product page with options, the others merged with a 301.
Frequently asked questions
Does deleting pages lower traffic?
Not if you remove pages with no impressions. On our own pruned site, daily impressions stayed at the same level. The risk comes from pages that have visits or links: those are merged with a 301.
Should I use a 410 or a 404?
For Google, both are treated the same way: the page leaves the index. The 410 better expresses the intent of permanent removal; the 404 works too.
How long should a 301 redirect be kept?
As long as possible, at least a year, and indefinitely if the old URL still gets external links or visits.
What is a zombie page?
An indexed or indexable page that gets neither impressions nor visits and brings nothing of its own: a vanished product, a duplicate, a generated page with no content. It is merged or removed.
Sources
- Google Search Central, HTTP status codes, network and DNS errors (2026)
- Google Search Central, Redirects and Google Search (2026)
- Google Search Central, Remove information from your site (2025)
- Google Search Central, A guide to Google Search ranking systems (2025)
- Google Search Central, Spam policies: scaled content abuse (2026)