Content Pruning: When Less Content Ranks Better
Old, thin, and overlapping pages drag down the rest of your site. Here's how to decide what to improve, merge, redirect, or delete.
Most sites that have published content for a few years are carrying dead weight: posts that never ranked, pages that overlap with three other pages, and articles whose information quietly went stale. The instinct is that more content is always better. It is not. Quality evaluation happens at the site level as well as the page level, and a large mass of low-value pages drags on everything else. A page that earns nothing still costs something. It dilutes crawl budget, it splits link equity across URLs that should have been one page, and it gives both Google and the AI assistants more chances to surface a version of your brand that is no longer true.
Content pruning is the discipline of regularly deciding, page by page, whether content should be improved, merged, redirected, or removed. Done annually, it is one of the highest-leverage content projects available. It costs no new writing, no new links, no new promotion. It costs the labor of looking honestly at what you already published and admitting which of it is not earning its keep.
Build the inventory first
Export every indexable URL and, for each one, pull organic sessions and impressions (last 12 months), clicks and average position from Search Console, backlinks pointing at the page, conversions or assisted conversions if you track them, and publish date plus last meaningful update.
Any crawler plus Search Console plus your analytics gets you there. The goal is one spreadsheet where every page has evidence attached. A page that has impressions but no clicks tells a different story than a page with clicks but no conversions, and both tell a different story than a page with backlinks and no traffic. Without the evidence next to the URL, the verdict becomes an opinion. With it, the verdict becomes a decision you can defend in a QBR.
The inventory is the part teams skip, and it is the part that makes the rest of the work defensible. If you walk into the pruning meeting with a list of URLs and a gut feeling, the meeting becomes an argument about taste. If you walk in with one row per URL and columns for sessions, impressions, clicks, position, backlinks, conversions, and last updated, the meeting becomes a sorting exercise. You do not have to agree on whether a post is good. You have to agree on what the columns say.
The four verdicts
For each underperforming page, pick one.
Improve. The topic is right, the execution is behind. Signs: it ranks on page two, it has impressions but a weak click-through rate, or it ranked once and faded. Update the information, tighten the intro, add what competing pages cover that you do not, and republish with a current date. This is usually the highest-ROI bucket. Pages with existing history respond faster than new ones, because they already have crawl history, some links, and a rank signal Google remembers.
Merge. You have three mediocre pages circling the same intent, splitting links and confusing search engines about which to rank. Combine them into one strong page, 301-redirect the losers into it, and update internal links. Cannibalization is one of the most common problems we find on mature blogs. Three pages on the same topic usually means none of them rank, because the link equity and topical authority that should have gone to one URL is smeared across three.
Redirect. The page has backlinks or traffic but no longer fits your strategy: a discontinued product, an event that passed. 301 it to the closest relevant living page so the equity is not wasted. Do not redirect everything to the homepage. Irrelevant redirects get treated as soft 404s, which means the equity you were trying to save is discarded anyway. Pick the closest living page even if it is not a perfect match, and accept that a partial match is better than a homepage dump.
Delete. No rankings, no links, no traffic, no strategic value. Let it 404 or 410. This feels reckless the first time. It is routine hygiene. If nobody visits a page and nothing links to it, its absence changes nothing except the average quality of what remains. The fear is usually that someone, somewhere, has bookmarked it. The evidence in your spreadsheet says otherwise.
What pruning does for AI visibility
There is a second reason to prune now: retrieval. AI assistants pulling from your site work from the same index Google does. Stale pages with outdated claims are candidates for being retrieved and quoted, meaning an assistant can confidently cite your 2022 pricing page at a customer in 2026. Keeping your corpus current is not just a rankings play anymore. It is brand accuracy in AI answers. A wrong price quoted by ChatGPT is worse than no price at all, because the customer treats it as your word.
Google's AI optimization guidance is the official page we point clients at when they ask whether pruning "hurts GEO." Helpful, current pages are the ones models can quote. A stale archive is a liability.
On retainers we watch the after-effect in Promptwatch, the platform we run client programs on. If a URL we kept still gets cited for an outdated claim, that is an improve ticket. If a URL we redirected still appears in a citation, we check crawler logs on Professional, Business, or an agency plan before assuming the 301 failed. Essential has no listed crawler-log allowance, so on that tier we are reading citations without the fetch evidence to explain them. The crawler log is what tells us whether the bot is still picking up the old URL because the redirect has not propagated, or because a third party copied the content.
Cadence and guardrails
Run the full exercise once a year. Spot-check your top 100 pages quarterly.
Log every verdict and redirect. You will need the paper trail when someone asks where a page went. Without the log, the answer is "we deleted it in March," which sounds arbitrary. With the log, the answer is "we deleted it in March because it had zero sessions, zero links, and a 2021 publish date with no update," which sounds like a decision.
Watch Search Console for a few weeks after a large prune. Some fluctuation is normal. Sustained drops on pages you kept usually mean you deleted something that was quietly supporting them through internal links. That is the one mistake pruning can cause, and it is fixable: restore the internal link from a surviving page, or reinstate the deleted URL as a supporting page rather than a standalone one.
Publishing less, better, and maintaining what you have is a strategy. An ever-growing archive nobody tends is just a liability with a CMS.
FAQ
Do we delete every page that is not ranking?
No. Improve if the topic is right and the execution is behind. Merge if several pages split the same intent. Redirect if the page has links or traffic but no longer fits. Delete only when there are no rankings, no links, no traffic, and no strategic value.
Will pruning hurt AI citations?
Stale pages with outdated claims are the ones assistants can quote by accident. Keeping the corpus current is brand accuracy in AI answers, not a rankings-only play. Google's AI optimization guidance still applies: helpful, current pages.
How often do we run this?
The full exercise once a year. Spot-check the top 100 pages quarterly. Watch Search Console for a few weeks after a large prune.
If you want a prune pass on an existing archive, hello@1001seomedia.com.