# SEO content pruning: Complete 2026 guide and best practices

> Discover SEO content pruning: definition, practical steps and best practices to optimize your website. Comprehensive guide with concrete examples.

- Source: https://julien-gourdon.fr/lexique/en/seo-content-pruning
- Author: Julien Gourdon (https://julien-gourdon.fr/auteur)
- Published 2026-01-17 · updated 2026-11-22

---

L’Essentiel :

SEO content pruning means updating, merging, redirecting or removing a site's weak pages after a page-by-page review. Google states that its ranking systems work at the page level, with site-wide signals also in play: a page with few clicks does not automatically drag the others down. The siteFocusScore and siteRadius attributes from the 2024 leak remain working hypotheses. This guide covers the audit, the equivalence test to pass before any 301, when a 404 or 410 is the right answer, and a real decision with its counter-signals.

Content pruning comes up in almost every SEO audit, often presented as a universal fix. In 2024, the publication of Google's internal Content Warehouse API documentation gave it a new argument: it lists attributes such as siteFocusScore and siteRadius, which describe how focused a site is on its topic. Their existence says nothing about their weight in ranking, nor about what happens when pages are removed. I therefore treat these signals as leads worth thinking about, and every pruning decision as a trade-off to document.

## What is content pruning?

Content pruning is an SEO practice that involves identifying, then updating, merging or removing obsolete, redundant or underperforming content on a website. The goal is to make the editorial offer easier to read, for users and search engines alike. Rather than accumulating pages, pruning aims to keep the ones that provide identifiable value.

The types of content usually reviewed during pruning are:

- Outdated content: information that is no longer up to date.
- Low-performing content: pages that generate little to no organic traffic.
- Low-quality content ("thin content"): pages with very little content or no real depth.
- Duplicate content: pages that replicate similar or identical content found elsewhere on the site.
- Off-topic content: articles that stray from the main theme of the site.

Falling into one of these categories opens a review. The decision comes afterwards, page by page.

## What do siteRadius and siteFocusScore actually reveal?

The documents leaked in May 2024 describe thousands of attributes, without saying which ones are used or how they are weighted. Two of them appear in almost every discussion of pruning:

- siteFocusScore: a score indicating how much a site concentrates on a single topic.
- siteRadius: how far a page's embeddings deviate from the site's average embedding, in other words the page's distance from the site's thematic core. The way [vector search compares embeddings](https://julien-gourdon.fr/article/en/vector-search) helps to understand this kind of calculation.

These definitions make a link between thematic dispersion and authority plausible. They do not prove that removing an off-topic page improves the site's score, nor that this score affects the ranking of other pages. Moving from the existence of an attribute to a deletion rule fills a documentation gap with a belief.

## Does a weak page hurt the whole domain?

Nothing in Google's public documentation supports that claim for any given page. [Google's guide to its ranking systems](https://developers.google.com/search/docs/appearance/ranking-systems-guide) states that these systems are designed to work on the page level, and that site-wide signals and classifiers also contribute to its understanding of pages. Google adds that good site-wide signals do not mean that all content from a site will always rank highly.

Site-wide signals do exist. The helpful content system was the most discussed example, before it was folded into Google's core ranking systems in March 2024. Knowing that an aggregate signal exists still does not allow you to conclude that a specific low-traffic page on your domain is pulling the others down. A page with little traffic may answer a rare query, attract links or serve readers in another language. Whether to remove it depends on what it gives readers. A supposed penalty is not enough to justify it.

## What can pruning realistically deliver?

The gains from content pruning show up when the starting problem has been identified. Their size varies widely with the size of the site, and they have to be measured on the pages involved:

- A clearer topical offer: merging overlapping pages stops several URLs from competing for the same intent and strengthens the surviving page.
- A better reader experience: outdated information, or information contradicted by a newer page on the same site, costs trust even when it is not penalized.
- Less maintenance: fewer pages to monitor leaves more time to update the ones that matter.
- Leaner crawling: the crawl budget argument mostly concerns very large sites. For a site with a few hundred pages, Googlebot is almost never the limiting factor.

## How do you prune without losing value?

Pruning starts with an inventory and ends with a measurement. In between, every page gets a reasoned decision:

1. Audit your content: combine Google Search Console, GA4 and a backlink report over at least twelve months (clicks, impressions, positions, conversions, inbound links). Impressions alone do not prove real demand.
2. Look for counter-signals: an editorial backlink, a rare but qualified long-tail query, seasonality, a version in another language or an assisted conversion often justify keeping the page.
3. Decide: update if the topic is still relevant, merge if several pages answer the same intent, redirect only if an equivalent page exists, remove in all other cases.
4. Measure the affected group: compare, over windows of equal length, the clicks and impressions of the modified pages and their targets. Site-wide traffic is too noisy to isolate the effect of a batch.

## Which pages deserve a 301 redirect?

A 301 is justified when the destination meets the same need as the source. Defaulting to the homepage or a parent category is not enough. Google's documentation on site moves with URL changes advises against redirecting many old URLs to one irrelevant destination, such as the homepage, and warns that such redirects may be treated as soft 404 errors.

Before every redirect, I run the target through an equivalence test:

- Does it answer the same search intent as the source?
- Will a reader arriving through an old link find the information they came for, or its updated version?
- Has the useful content of the source been carried over before the redirect?
- Do internal links to the source now point directly to it?

A single negative answer is enough to postpone the redirect. Without a relevant replacement, a 404 or 410 response is appropriate: Google drops the URL from its index over time, and readers understand that the content no longer exists. Removal does not erase every copy, since AI search engines can keep a cached version, as shown when [ChatGPT kept indexing content that moderation had deleted](https://julien-gourdon.fr/article/en/black-hat-geo-when-chatgpt-indexes-deleted-content). For redirected pages, a redirect plan then keeps track of the mapping between old and new URLs.

## A worked example: why this guide was kept

A previous version of this guide presented siteRadius and siteFocusScore as proof that an underperforming page harms the entire domain, and listed the homepage among possible redirect targets. Both points are corrected above. On traffic alone, the page could have been removed: its French version received 33 Google clicks in twelve months, this English version 6.

The table below sets out what was weighed in my content audit of 13 October 2026 for the French guide, for this English version and for a group of news pages on the same site. Read it column by column: counter-signals often carry more weight than low traffic.

| Page | Signals for removal | Counter-signals | Decision |
| --- | --- | --- | --- |
| French guide | 33 clicks in twelve months, an overstated claim about the effect of a weak page | Two followed backlinks whose anchors quote the very claim being corrected | URL kept, text corrected at the source |
| This English version | 6 clicks in twelve months | It serves English-speaking readers, and redirecting it to the French page would break the hreflang pair | URL kept, correction mirrored |
| News pages with no equivalent (Apple and Gemini, ads in ChatGPT, antitrust trial) | Low traffic, dated topic | No page on the site meets the same need: a 301 to a neighbouring page would be a soft 404 | No redirect: update, or a 410 response if removal is decided |

Deleting the French guide would have cut two inbound links that point precisely to the faulty claim. Readers who follow them now land on a version that repairs the error instead of making it disappear. Measurement covers this group of two pages: clicks and impressions of the French and English versions over 28 days before and after the correction, without diluting them in site-wide traffic.

## Which mistakes make pruning lose value?

Pruning loses value when a page disappears along with what it brought: links, an answer, an audience. A few safeguards limit that risk:

- To avoid:
  - Deleting a page without checking its backlinks and internal links.
  - Mass-redirecting to the homepage or to a generic category.
  - Justifying a deletion with a supposed site-wide penalty, without any evidence specific to the page.
  - Pruning only once, while content keeps ageing.
- Best practices:
  - Work in small batches, so that observed effects can be attributed.
  - Prefer updating or merging over removal.
  - Update the XML sitemap and internal links as soon as redirects are live.
  - Record every decision and its reasons, so that it can be reviewed later.

A leaner site can become easier to understand, for readers and search engines alike. Pruning nonetheless remains an editorial decision, made page by page. Primary evidence describing exactly how a weak page affects the rest of a domain would change the wording of this guide, without retroactively validating deletions made without review. What protects a site over time is the record of decisions: why each page was kept, merged, redirected or removed, and what the measurement showed afterwards.
