Content Pruning

Here's a hard truth: a 1,000-page site with 200 excellent articles and 800 thin ones isn't treated like a site with 200 excellent articles. It risks being treated like a low-quality site with occasional bright spots.

Google says it uses both page-level and site-wide signals in ranking. The helpful content system, which became part of the core ranking systems in March 2024, made that site-wide dimension famous: sites with large amounts of unhelpful content saw sitewide declines, not just page-by-page losses. Thin content drags down the whole.

Pruning isn't deletion for its own sake. It's quality control. Its sibling is the content refresh: most underperforming pages should be fixed, not removed. This guide covers the removal side.

The content audit process#

Step 1: Export all indexed URLs#

Combine a crawl (Screaming Frog or similar) with Search Console's Pages report (Indexing > Pages) and your sitemaps. You want every URL Google is actually crawling and indexing, not just what your CMS lists as "published". The two lists are often different.

Step 2: Pull 12 months of Search Console data per URL#

Export Performance > Pages with clicks, impressions, CTR and position, and join it to the crawl:

Step 3: Sort into three buckets#

The thresholds below are a starting rule of thumb, not Google numbers. Scale them to your site's traffic.

Working: more than about 10 clicks a month, or meaningful impressions with positions inside the top 30. These are earning their keep. Leave them alone unless there's a specific reason to touch them.

Underperforming: some impressions, few clicks, positions 20 to 50. These need a judgement call: can they be meaningfully improved? Are they duplicating a better page?

Dead weight: zero clicks, near-zero impressions, or positions beyond 50 for a full year. These are the first queue for pruning action.

The three decisions#

For every underperforming or dead-weight URL, you have three options:

Improve#

The page has a real keyword opportunity but hasn't delivered because the content is thin, outdated, or poorly structured.

Indicators: Position 15 to 40 with meaningful impressions, clear intent, an obvious gap between your page and what's ranking above you.

Action: Rewrite substantially: add depth and first-hand detail, update the data, restructure for clarity. Don't add 300 words of filler and call it done. The full process, including how to date the update honestly, is in Content Refresh & Decay.

Consolidate#

Multiple pages cover the same topic without any of them being the authoritative resource. They're splitting relevance signals and competing against each other.

Indicators: Two or more URLs ranking for the same query cluster, high content similarity, none individually ranking on page one.

Action: Pick the strongest URL as the destination. 301 redirect the others. Merge the best content from each into the winner. Update all internal links to point to the survivor. See Duplicate Content for the canonical side.

Delete or noindex#

The page has no viable keyword opportunity and no user value. No amount of improvement will make it worth ranking, or worth reading.

Indicators: Targeting an irrelevant or zero-opportunity query, auto-generated tag/category pages with minimal content, outdated event or campaign pages with no evergreen relevance.

Options:

• 301 to the most relevant live page: best for users when a genuinely related page exists
• noindex: if the URL must stay accessible (for users, ads or internal tools) but shouldn't be indexed
• Delete and return 404 or 410: for content with no natural redirect destination. Both drop the URL from the index; 410 just says "gone on purpose" more explicitly

The usual suspects#

Thin tag and category pages#

CMS auto-generated pages for tags or categories with only 2 or 3 posts beneath them. Noindex or consolidate.

Date-archive pages#

/2019/07/ archive pages almost never satisfy real search intent. Noindex or redirect to the corresponding category.

Duplicate location/service pages#

"We serve the [city] area" × 50 cities, all with the same copy. This is what Google's doorway abuse spam policy describes, and it can lead to a manual action. Either write genuinely unique content per location (local case studies, local team, local details) or collapse them into a single well-structured service area page. More in Local SEO.

Scaled AI or programmatic pages#

Hundreds of pages generated from a template or an LLM with little added value are exactly what the scaled content abuse policy targets, "no matter how it's created". John Mueller said in September 2026 that low-value programmatic SEO "often leads to a site that's either spam, borderline spam, or low quality", and that Google's systems may have "lost faith" in a site because of the old pages (as reported by Search Engine Roundtable). Prune these aggressively. See Programmatic SEO and AI-Assisted Content Without Penalties.

Stale dated content#

"Best SEO Tools for 2018" still indexed in 2026 disappoints every visitor, and an obsolete date in a title is one of the reasons Google lists for rewriting title links. Either refresh it properly (new testing, new picks, then a new year) or redirect it to your live equivalent. Don't just change the year in the title: Google's helpful content guidance specifically warns against changing dates to look fresh when the content hasn't substantially changed.

"Me too" blog posts#

Posts written to cover a topic because a competitor covered it, not because your site adds anything different. Google calls this commodity content. These are the most uncomfortable to acknowledge and among the most important to fix or remove.

What to watch after pruning#

In the 30 to 90 days after a significant pruning operation:

• Indexed pages count (Indexing > Pages): should fall to match your intended footprint
• Clicks for surviving pages: should improve as quality signals concentrate
• Crawl stats (for larger sites): Googlebot should spend proportionally more time on the pages you kept

Add a Search Console annotation on the date you pruned so the change is visible on the chart later.

Recovery from site-wide quality problems is slow but real. Since December 2025 Google's core updates documentation says it also makes "smaller core updates" that aren't announced, which "are another way that your content can see a rise in position (if you've made improvements)". So you don't strictly have to wait for a named core update, though larger shifts often still arrive with one. Nothing is guaranteed: pruning removes a drag, it doesn't add value by itself.

Next: Content Refresh & Decay, keeping the pages that survive the prune from sliding back.