The hardest content decision I ever institutionalized was deleting things. Every post feels like an asset because it cost money and effort to create. But cost is not value, and on a mature site, a meaningful share of the archive is not neutral dead weight. It is actively working against you.
I have run pruning exercises on my own properties and on client sites, and the pattern repeats: sites get healthier when they get smaller and denser. Here is the framework.
Why Old Content Hurts (Not Just Fails to Help)
Three mechanisms turn yesterday's posts into today's liability.
Sitewide quality evaluation. Search systems assess domains holistically. When 60 percent of your indexed pages are thin, outdated, or unvisited, that ratio is part of your site's identity. Your best pages are graded on a curve set by your worst ones.
Cannibalization by accumulation. Ten years of publishing produces five posts on overlapping topics, splitting links, splitting relevance signals, and forcing Google to guess which one to rank. Usually it guesses the old one out of inertia, and your better new page starves.
Trust erosion with humans and machines. A 2021 post with dead screenshots and obsolete advice, surfacing in 2026, damages the reader's trust in everything else you publish. And in an era where AI systems decide whom to cite, stale claims in your corpus are contradictions that lower machine confidence in your entity.
The Inventory: Grade Everything Before Touching Anything
Pruning without data is vandalism. My inventory pulls, for every URL: organic clicks and impressions over twelve months, inbound external links, internal links, conversions or assisted conversions, and publish/update dates. Then every page gets one of four grades.
Keep and update
Pages with traffic, links, or rankings that have decayed but whose topic still matters. Decay is usually fixable: refresh the facts, tighten the writing, add what you have learned since, improve the internal links. In my experience a serious update to a decayed page recovers rankings far faster than a new page earns them, because the URL already has history. This is the highest-ROI bucket, and it should be the biggest.
Merge and redirect
Multiple mediocre pages on one topic become one strong page. Combine the best material, 301 the losers to the winner, and reroute their internal links. This is where cannibalization dies. The merged page typically outperforms the sum of its parts because signals consolidate instead of splitting.
Redirect only
Pages with external links but no salvageable content purpose. The content is not worth maintaining, but the equity is worth keeping. 301 to the closest genuinely relevant page. Redirecting everything to the homepage is lazy and treated accordingly.
Delete
No traffic, no links, no conversions, no strategic role, no closely related target. Let it 404 or 410. This feels reckless the first time. It is not. A page nobody visits and nothing links to is contributing only to your thin-content ratio.
The Decisions People Get Wrong
A few judgment calls come up in every pruning project.
"But it might get traffic someday." Twelve months of zero impressions is the internet telling you something. Someday is not a strategy, and the page can always be rewritten properly if the topic becomes relevant, ideally as part of a deliberate topical authority map instead of an orphaned one-off.
"It still gets some traffic." Traffic to a page that misinforms visitors or represents you badly is negative traffic. I have deleted pages with real clicks because what they said no longer reflected reality. Brand damage does not show up in Search Console.
Old dates on evergreen pages. Updating a page means actually updating it, then adjusting the visible and structured dates. Changing dates without changing substance is a trick, readers notice, and it burns exactly the E-E-A-T credibility you are trying to build.
Seasonal and campaign pages. Do not delete annually recurring pages; update them on a stable URL. A single evergreen URL that improves every year beats a graveyard of dated variants splitting equity.
The Aftermath: What to Expect
On sites where I have run full pruning cycles, the pattern I have repeatedly observed: crawl stats improve within weeks as bots stop wasting budget on junk, then over the following two or three months the surviving pages tend to strengthen. I will not dress my observations up as a controlled study, but I have seen the shape enough times, across enough unrelated sites, that pruning is now a standing part of how I operate rather than an emergency measure.
Two operational notes from scars:
- Reroute internal links before removing pages, not after. Every deleted URL with inlinks becomes a batch of broken links unless you fix them proactively. Pruning and internal linking are one project wearing two names.
- Log everything. URL, action taken, redirect target, date. When something unexpected happens in the rankings three weeks later, this log is the difference between diagnosis and superstition.
Make It a Cycle, Not an Event
The real failure is treating pruning as a one-time cleanup. Decay is continuous, so the response has to be. My cadence: a light quarterly pass catching fresh decay on important pages, and a full annual inventory with the four-grade triage. On a well-run site the annual exercise gets smaller every year, because the publishing side stops producing future prunees once the team internalizes that every page must earn its place.
A content archive is a garden, not a museum. Nothing in it is sacred because it is old, and everything in it is either feeding the whole or feeding on it. Prune accordingly.