New · A dedicated AI SEO channel: see conversions from ChatGPT, Perplexity, Claude & Gemini →
← Blog
Part of: SEO Conversion Tracking: Measure Conversions and Value from Organic Search →

Content Pruning: What to Cut, Merge or Keep (With a Template)

Portrait of Samy ThuillierBy ··12 min read
Content pruning decision table sorting website pages into keep, update, merge, noindex and remove

Content pruning is the practice of auditing every page on your site and deciding, page by page, whether to keep it, update it, merge it into a stronger page, hide it from search, or remove it. Most pruned pages are improved or merged, not deleted. Done with data, it leaves you with fewer, better pages that search engines and AI assistants can understand, and visitors who land on something current.

This guide covers what to prune and what to protect, a five-step process with a copyable template, a decision table, a worked example with real arithmetic, the technical mistakes that quietly undo a prune, and how to measure whether it worked in traffic and in revenue.

What Content Pruning Is (and Is Not)

The gardening image is accurate: you cut what is dead so the healthy parts get the resources. On a website, the “resources” are crawl attention, internal links, backlinks and the reader’s trust. A page that is outdated, thin, off-topic or a near copy of another page competes for those resources without giving anything back.

Content pruning has five possible outcomes for any URL:

  • Keep: the page does its job. Leave it alone.
  • Update: the topic is right but the page is stale, thin or misses the search intent. Rewrite it on the same URL.
  • Merge: two or more pages answer the same question. Combine the best parts into one and 301 redirect the others to it.
  • Noindex: the page is useful to visitors but has no business in search results (tag archives, customer-only pages, campaign variants).
  • Remove: the page has no traffic, no links, no conversions and no reason to exist. Let it return a 404 or 410.

A content audit and content pruning are two halves of one job. The audit is the diagnosis: an inventory of every URL with its performance data. Pruning is the treatment: the changes you make based on that diagnosis. Pruning without an audit is guessing.

Why Pruning Can Help, and When It Does Nothing

Pruning tends to pay off through four mechanisms:

  • Less cannibalization. When three posts target the same query, they split clicks and links, and none of them ranks as well as one complete page would. Merging gives search engines one clear answer per topic.
  • Better site-wide quality. A library full of outdated or thin pages makes the whole site look less reliable to readers and less coherent to search systems that look at more than one page.
  • Simpler navigation. Fewer dead ends and fewer near-duplicate pages make it easier for visitors to reach the page that answers their question or asks for the sale.
  • Crawl efficiency on very large sites. When a site has far more URLs than search engines crawl regularly, removing junk URLs leaves more crawl activity for the pages that matter.

Pruning does nothing for a 60-page site whose real problem is that it has no links, a slow server or pages nobody searches for. Google is explicit about one tempting shortcut: its guide to people-first content says that removing a lot of older content mainly to make a site seem “fresh” won’t help rankings, and it lists changing dates on pages that have not substantially changed as a warning sign. Prune to improve what readers find, not to send a freshness signal.

Where the Top Guides Disagree

If you read several content pruning guides, you will see conflicting advice on three points. Here is what the primary sources say.

How big a site needs to worry about crawl budget

Several guides say crawl budget matters from about 10,000 pages. Google’s own crawl budget documentation is narrower: it is written for sites with more than a million unique pages whose content changes about weekly, or sites with more than 10,000 unique pages whose content changes daily. If your 2,000-page blog changes a few times a month, crawl budget is not the reason to prune. Quality, cannibalization and conversions are.

Whether noindex saves crawl budget

Some guides recommend noindex to fix index bloat and crawl waste. Noindex does keep a page out of search results, but the same Google crawl budget page says not to use it for crawl savings: Google still requests the page, then drops it when it sees the tag. Use noindex to keep useful pages out of results. Use removal or a redirect when you want crawlers to stop spending time on a URL.

How often to prune

Recommendations range from monthly, to every one to three months for big sites, to quarterly, twice a year or once a year depending on size. None of these come from a study; they are rules of thumb. A schedule tied to publishing volume is more useful, and it is covered in the frequency section below.

What to Prune and What to Protect

Pages that usually end up in the update, merge or remove columns:

  • Outdated pages: old pricing, discontinued products, expired events, statistics from years ago, advice that is now wrong.
  • Thin pages: a few paragraphs that do not answer the question, auto-generated pages, stub posts.
  • Overlapping pages: several posts that answer the same query with slightly different titles.
  • Off-topic pages: posts written to chase traffic that has nothing to do with what you sell or who you serve.
  • Low-value indexable URLs: tag and filter pages, internal search results, parameter variants.

Pages you should not remove, even with low traffic:

  • Pages that produced conversions in the last 12 months, from any channel.
  • Pages with backlinks from real sites (redirect them if they must go).
  • Pages published in the last six months or so, which have not had time to rank.
  • Legal, support, pricing and other pages people need even if few search for them.
  • Pages that AI assistants still send visitors to (more on this below).

How to Prune Content in 5 Steps

Step 1: Build the inventory

Pull every indexable URL from four places: a crawl of the site, a CMS export, your XML sitemap and the Pages tab of the Search Console performance report. Each source finds URLs the others miss, such as orphan pages with no internal links or old PDFs. Deduplicate so each URL appears once, and drop URLs that already redirect.

Step 2: Add the data columns

Use a 12-month window so seasonal pages are judged fairly. For each URL, add:

  • Search clicks and impressions from Search Console (Performance, Pages tab), plus average position.
  • Conversions and their value from all channels, by landing page. In GA4, the Landing page report under Engagement shows key events per landing page; add your value per conversion.
  • Referring domains from a backlink tool, and internal links from the Search Console Links report.
  • AI assistant referrals: sessions where the source is chatgpt.com, perplexity.ai, gemini.google.com, copilot.microsoft.com or claude.ai.
  • Last updated date and a yes/no for “contains outdated facts.”
  • Target query, so overlapping pages sort next to each other.

Here is a template you can copy into a spreadsheet:

URLTarget queryClicks 12mImpr. 12mConv. 12mConv. value 12mRef. domainsAI referrals 12mUpdatedIssueActionRedirect to
/blog/old-guidehow to x403,1000$0002021Overlaps /blog/x-guideMerge/blog/x-guide
/blog/x-guidehow to x2,90061,00014$1,6809222025NoneKeep
/blog/event-2022brand event31200$0002022ExpiredRemove

The rows above are illustrative placeholders. Replace them with your own export.

Step 3: Apply the decision rules

Agree on rules before you look at individual pages, and share them with the people who own those pages (product, sales, support). That turns a fight about favorite posts into a review of a rule. This table is a starting point; adjust thresholds to your site.

EvidenceActionHow
Steady clicks, on-topic, accurateKeepNo change. Add internal links from weaker pages.
On-topic, impressions but few clicks, stale or thinUpdateRewrite on the same URL. Change the date only if the content changed substantially.
Two or more pages answer the same queryMergeMove the best sections into the strongest URL, 301 the rest to it. A canonical tag is a softer, temporary option.
Any conversions, backlinks or AI referrals, but weak or off-topicMerge or redirect301 to the closest page that offers the same next step (same form, product or offer).
Useful to visitors, useless in searchNoindexKeep live, add noindex, do not block in robots.txt.
No clicks, no conversions, no links, no AI referrals, no business reasonRemoveReturn 404 or 410, remove internal links and sitemap entries. Repurpose good parts elsewhere if any.

Step 4: Make the changes correctly

For merges, rewrite the target page so it includes what was useful on the pages you are folding into it, then redirect. A redirect to a page that does not answer the old page’s question wastes the visit. For removals, update every internal link that pointed at the removed URL, take it out of the XML sitemap, and check that the server returns a real 404 or 410 rather than a 200 page saying “not found.”

Step 5: Ship in batches and measure

Release one batch at a time, grouped by type: all merges in one topic, then all expired event pages, then a subfolder. Wait a few weeks, check the numbers in the measurement section below, then move to the next batch. If something drops, a small batch is easy to trace and to roll back.

Check Revenue and Lead Value Before You Delete Anything

Most pruning guides pick pages by traffic. Traffic is the wrong filter on its own. A comparison post with 30 visits a month can produce more demo requests than a glossary page with 3,000. Before any page goes into the remove column, put a dollar figure on it.

Value at risk per URL = sum over 12 months of (conversions that started on that page x value per conversion), all channels

If you do not have values yet, set them first: for a lead, value is roughly close rate times average deal value; for a purchase, it is the order value. Our guide on how to calculate conversion value walks through it. Then apply one rule: any URL with value at risk above zero is never left to 404. It gets kept, updated, or redirected to a page that offers the same next step, so the visitor who would have converted still can.

Count conversions from all channels, not only organic. Pages that look dead in search often earn their keep from email, social or direct visits. For the organic part specifically, see organic conversion attribution for how to tie a conversion to the search landing page when the keyword is hidden.

Content Pruning Worked Example: A 420-Post Blog

This example is illustrative. The numbers are invented to show the arithmetic, not taken from a real site. A B2B company has 420 blog posts. The team exports 12 months of data and applies the rules above.

A traffic-only rule (fewer than 10 search clicks in 12 months, no referring domains) flags 176 posts for removal. Before anything is deleted, the team adds the value column. Two conversion types count: newsletter signups valued at $3 each and demo requests valued at $250 each (a 10% close rate on a $2,500 average first-year deal).

Group found in the 176 flagged postsPostsConversions 12mValue 12m
Posts with newsletter signups only1322 signups22 x $3 = $66
Posts with at least one demo request69 demos9 x $250 = $2,250
Posts with zero conversions1570$0
Total17631$2,316

Deleting all 176 would have thrown away $2,316 a year in lead value, almost all of it from six posts. Those 19 posts move to the merge column and redirect to the closest page with the same signup or demo form. The final plan:

ActionPostsNotes
Keep96Steady clicks, on-topic
Update71Impressions but low clicks, stale facts
Merge + 3018465 overlapping posts plus the 19 with conversions, folded into 27 target pages
Noindex12Customer-only release notes
Remove (404/410)157No clicks, no links, no conversions, no AI referrals
Total42096 + 71 + 84 + 12 + 157 = 420
Illustrative result of the plan
420 posts become 263 live posts (420 - 157 removed), of which 251 are indexed (263 - 12 noindexed)
$2,316 a year of lead value is kept by redirects instead of lost to 404s
27 target pages now carry the content and signals of 84 merged posts

Check AI Assistant Referrals Before Removing a URL

ChatGPT, Perplexity, Gemini, Copilot and Claude link to web pages in their answers, and an old URL can keep receiving those clicks long after it stopped ranking in Google. A page with zero search clicks can still have a steady trickle of assistant referrals. If you delete it, every one of those visitors hits a 404.

In GA4, open Traffic acquisition, switch the dimension to Session source, and filter for the assistant domains, then add Landing page as a secondary dimension. Any URL with assistant referrals goes into the merge or keep column, not removal. One caveat: some assistant clicks arrive without a referrer and show up as Direct, so the count is a floor. Our guide to AI conversion tracking covers how this traffic is identified.

Failure Modes That Quietly Undo a Prune

These are the mistakes that show up weeks later as “pruning did not work.” Each has a quick check.

FailureWhat happensHow to check and fix
Noindex page also blocked in robots.txtGoogle never fetches the page, never sees the noindex, and the URL can stay in results.Remove the robots.txt block for noindex pages. Google’s noindex documentation requires the page to be crawlable.
Removed page returns 200 with a “not found” messageGoogle reports a soft 404 and keeps crawling the URL.Fetch the URL and check the status code. It should be 404 or 410.
Everything redirected to the homepageVisitors who wanted a specific answer land on a generic page and leave, and the old page’s topic is lost.Redirect only to a close topical match. If none exists, let the page 404.
Internal links and sitemap still point at old URLsCrawlers and visitors go through redirect hops or hit 404s.Re-crawl after each batch and update links and the sitemap to the final URLs.
Redirect chains after repeated mergesA to B to C slows visitors and wastes crawl requests.Point every old URL directly at the final target.
Dates changed without real changesGoogle lists this as a warning sign of search-engine-first content.Only update the date when the page changed substantially.
Big-bang releaseHundreds of changes at once make any drop impossible to diagnose.Release in batches by type, with a note of what went live and when.

The robots.txt and noindex rule is stated in Google’s noindex documentation, and the soft 404 and 404/410 guidance comes from the crawl budget page linked above. That page recommends a 404 or 410 for permanently removed pages, so you do not need to agonize over which one to use.

Measuring the Result in Traffic and Revenue

Pruning is measured on what is left, not on what was removed. Removed pages lose their clicks by design. Compare a fixed window before and after each batch (for example eight weeks each), and compare against the same weeks last year to spot seasonality.

  • Search clicks to kept and target pages in Search Console. For a merge, add up clicks of the old pages plus the target before, and compare with the target alone after.
  • Indexing in the Page indexing report: are the pages you care about indexed, and are removed URLs dropping out?
  • Organic and AI assistant conversions and value by landing page. This is the number that tells you whether the prune helped the business, not only the traffic chart.
  • 404s with real traffic in your analytics: any removed URL that still gets visits needed a redirect.

Pruning usually ships alongside other SEO work, so you will rarely prove cause and effect cleanly. Look for changes that start right after a batch and are not explained by anything else that shipped. SEOConversion reports conversions and their value by landing page for Google, Bing and AI assistants, which makes the before and after comparison a single report. The wider setup is in our SEO conversion tracking guide.

How Often to Prune

Match the review rhythm to how fast your content library grows:

  • Publishing weekly or more: review one topic or subfolder each quarter, so the whole site is covered over a year.
  • Publishing a few times a month: one full review twice a year.
  • Publishing occasionally: one review a year.
  • Any size: an extra review after a large traffic drop, a migration, a rebrand or dropped product lines.
Prune less by planning ahead

Avoid years in URLs for content you will update (/best-crm instead of /best-crm-2024). Keep a list of pages with a known expiry date, such as pricing, events and annual roundups, and schedule their update or removal when you publish them. Check for an existing page on the topic before writing a new one, and expand that page instead.

Frequently Asked Questions

What is content pruning?

Content pruning is the practice of auditing the pages on a website and then updating, merging, hiding or removing the ones that no longer earn their place. Most pruned pages are kept and improved; deletion is the last option. The goal is a site where every indexed page is accurate, distinct and useful to the people who land on it.

Does content pruning hurt SEO?

It can. Deleting pages that still get search clicks, hold backlinks or drive conversions loses that value, and removing many pages at once makes it hard to see what went wrong. Pruning in small batches, redirecting anything with traffic, links or conversions to a close match, and measuring each batch keeps the risk low.

What is the 80/20 rule in SEO?

It is the Pareto principle applied to a website: a small share of pages usually brings most of the traffic and conversions. For pruning, it means you protect and improve the few pages that carry the site and treat the long tail with clear rules. Check the actual split in your own Search Console and conversion data rather than assuming 80/20.

What does pruning mean in AI?

In machine learning, pruning means removing weights or connections from a trained neural network to make the model smaller and faster with little loss of accuracy. It is unrelated to content pruning on a website. Large language models such as the one behind ChatGPT are neural networks, which is where the term shows up in AI discussions.

Is SEO dead now with AI?

No. AI assistants and Google’s AI Overviews still pull answers from web pages, and they cite pages that are clear, current and specific. A pruned site with one strong page per topic is easier for both search engines and assistants to understand. What has changed is that you should also check AI assistant referrals before deleting a page.

How often should you prune content?

Tie it to how fast you publish. Sites that publish weekly or more can review a slice of content every quarter, while small sites that publish a few times a year can run one review a year. Also run an unscheduled review after a large traffic drop, a site migration or a big merge of product lines.

Know what a page earns before you prune it.

SEOConversion shows the conversions and value each landing page brings from Google, Bing and AI assistants, so you can see which pages to protect and whether your prune paid off.

Start free