It also stalls fast on one question: do you redirect the dead page or just delete it? There's no blanket answer. It comes down to whether the page still holds value worth preserving, and whether there's a genuinely relevant place to send the people and bots landing on it. Get either wrong and you either leak authority or create a mess of soft 404s.
Why it's worth doing
Search engines have finite attention for your site, and every thin page spends some of it. On a 5,000-page site where 3,000 pages are junk, Googlebot wastes crawl budget revisiting the junk. Index bloat also dilutes your topical signal. Forty mediocre pages loosely about "email marketing" means none of them reads as the authoritative one. Consolidate that into five strong pages and the story gets clearer for crawlers and humans alike.
The trap is treating "underperforming" and "worthless" as the same thing. They aren't, and confusing them is how pruning projects go wrong.
The three options
Every page lands in one of three buckets. It's worth being clear on what each actually does:
- Redirect (301): a permanent forward that passes most of the old page's ranking signals to the destination, provided the destination is genuinely relevant. That caveat is the whole game.
- Delete: the URL returns a 404 (not found) or, better, a 410 (gone).
- Keep: you leave it live, ideally because you've checked it earns its place, not because you're nervous about touching it.
When to redirect
Redirect when the old URL still carries something worth forwarding and you have somewhere sensible to send it:
- A better page already covers the topic. 301 the old URL to it, but only if the topics genuinely match. Redirect a discontinued product page to a vaguely related category page and Google reads the mismatch as a soft 404, passing nothing. No close match? Delete cleanly instead.
- You've merged several articles into one. Redirect all the old URLs to the new resource. You keep their equity and stop competing against yourself — see our guide to bulk 301 redirects for mapping many URLs at once without creating chains.
- A migration changed the URLs. Every old URL needs a 301 to its new home. Missing these is one of the most common causes of traffic loss after a migration we get called in to fix.
- The page has backlinks, even if the page itself is now weak. One link from a high-authority domain can justify keeping a page alive via redirect when everything else says delete. Check Ahrefs, Majestic, or Semrush before you touch anything.
Our redirect mapping and implementation service covers exactly this: building the map, choosing the right destinations, and verifying nothing 404s after launch.
When to delete
Delete when a page has nothing left to give and nowhere relevant to point: expired events, last year's campaign, an announcement for a product you no longer sell. Thin AI pages with no traffic and no links fall here too, because there's no value to forward, so redirecting doesn't help. For duplicates, keep the strongest version and redirect the rest to it.
One detail people skip: prefer a 410 (Gone) over a 404 (Not Found). A 404 tells Google "maybe temporary, I'll check back." A 410 says "gone on purpose," and Google drops those from the index noticeably faster, which is exactly what you want. Most CMSs default to 404, so this needs setting up deliberately.
When to keep
Some pages look like pruning candidates on a spreadsheet and shouldn't be touched. Keep a page if it still pulls organic traffic, even modestly, since 30 sessions a month across 200 pages is real traffic. Keep it if it has backlinks worth holding, targets an intent nothing else covers (the fix there is improvement, not removal), or plays a supporting role in a topic cluster that helps your pillar page rank. The mistake is treating low traffic as the only signal. It's one input. Links, intent, and topical role routinely override it.
Decision matrix
| Situation | Action | Why |
|---|---|---|
| A better page covers this topic | Redirect (301) | Forwards equity to the page you want ranking |
| Several pages merged into one | Redirect (301) | Consolidates signals, stops self-competition |
| URL changed in a migration | Redirect (301) | Non-negotiable; missing these loses traffic |
| Weak page with good backlinks | Redirect (301) | Preserves links you can't easily replace |
| Expired / no relevant destination | Delete (410) | Nothing to forward; 410 drops it faster |
| Thin AI page, no traffic or links | Delete (410) | No value to preserve |
| Duplicate of a stronger page | Redirect to survivor | Consolidates onto the version you keep |
| Earns traffic, links, or unique intent | Keep | Already pulling its weight |
What AI search changes
Skip the hand-waving, because the practical point is narrow. Answer engines are built to find the most authoritative source on a topic and cite it. A site that consolidated its email-marketing content into one strong, well-structured page is far easier for a model to identify and cite than one where that topic is smeared across forty half-finished posts. Fragmentation that merely diluted your classic rankings actively hurts you when the goal is being the source a model reaches for. The mechanics are the same as always, namely fewer, better pages and clean structure, but the upside is larger than it used to be. That's the core of what we do at Krawl.
The mistakes that undo the project
- Redirecting everything to the homepage. An irrelevant target gets treated as a soft 404 and passes nothing, so a mass redirect to
/is just a messier delete. Redirect to a relevant page or don't redirect — for large cleanups, redirect management at scale covers doing this systematically. - Deleting pages with live backlinks. The page might be a dud. The links usually aren't.
- Ignoring internal links. After you remove or redirect a page, update the internal links pointing at it to hit the final destination directly, rather than leaving redirect chains.
- Hoarding zombie pages out of fear. If a page has no traffic, no links, and no unique intent, deleting it is the right call, not a risk.
A workflow that holds up
- Crawl the full site (Screaming Frog or similar) for every URL, status, word count, and internal link count.
- Join Search Console traffic and Ahrefs/Majestic/Semrush backlink data to the crawl, so every page has the three numbers you need.
- Sort into Keep / Redirect / Delete using those numbers plus a manual look, never a pure traffic threshold.
- Map redirect destinations, checking relevance for each.
- Update internal links to final destinations.
- Implement (301s for redirects, 410s for clean deletes). Our bulk redirect tool handles the 301 side at volume.
- Monitor Search Console for a few weeks and watch for 404 spikes, crawl anomalies, or ranking drops.
The point
Pruning isn't about removing pages. It's about making everything that's left clearer and stronger, for the people reading it and the machines deciding whether to surface it. When a page holds value, a relevant 301 keeps it working. When it doesn't, a clean 410 gets it out of the way. The sites that come out of this well aren't the ones that deleted or kept the most. They're the ones that made a deliberate call on every page.