Content Pruning for Affiliate Sites: What to Delete, Merge or Leave Alone
Pruning gets sold as a cure: delete your thin pages and Google will reward you. The honest version is smaller and still useful.
Google doesn't fine a site for having weak pages; what weak pages do is blur what the site is about, compete with your better pages for the same queries, and absorb crawl attention. Removing the ones that do that helps the rest. Removing pages because a tool called them thin helps nothing.
This post covers why pruning works when it works, a three-bucket test you can run from Search Console, the four actions (merge, redirect, noindex, delete) and when each applies, the pages you should never prune, and a once-a-year routine. It builds on the content audit post, which finds the candidates; this one decides what to do with them.
The short answer
Export every page's impressions, clicks and position for the last twelve months from Search Console. Sort pages into three buckets: earning or ranking (leave alone), competing with a sibling (merge into the stronger one and redirect), and dead (no impressions in twelve months, no links, off-topic: redirect to the nearest relevant page or delete).
Never prune a page with external links, a page under a year old, or a page that supports the topic even if it earns nothing itself.
Why pruning works, when it does
- Topic clarity. A bike-lights site with forty posts about recipes from a previous owner looks like neither. Removing the recipes makes the remaining forty read as a site about bike lights.
- Cannibalisation. Two guides targeting the same keyword split the signals and often both sit on page two. Merging them puts one on page one.
- Crawl focus. On a big site, hundreds of dead pages mean Google spends its visits re-checking pages nobody wants and finds new guides more slowly. On a fifty-post site this doesn't apply.
What pruning doesn't do: lift a site out of a helpful-content problem by deleting the thinnest 20%, or make the remaining pages better than they were. If the remaining pages are the problem, pruning the others changes nothing.
The three-bucket test
Search Console → Performance → Pages, last 12 months, export. Add a column for external links (from the Links report) and one for the page's age. Then:
| Bucket | Test | Action |
|---|---|---|
| Keep | Any clicks, or impressions with position under 30, or any external links, or under 12 months old, or a page the cluster needs (hub, how-to that guides link to) | Leave alone; refresh if stale |
| Merge | Two or more pages appearing for the same queries, neither winning | Combine into the stronger URL, 301 the other |
| Remove | Zero impressions in 12 months, no external links, over 12 months old, and off-topic or superseded | 301 to the nearest relevant page; delete only if there isn't one |
Most affiliate sites find the Keep bucket is 80% or more of the site, and the Remove bucket is a handful of pages from an old direction. That's the normal result; a pruning pass that wants to remove half the site is usually a sign the test was run on clicks instead of impressions.
The four actions
Merge
For two pages on the same query. Take the stronger page (more impressions, better position, more links), fold in anything the weaker one had that it doesn't, and 301 the weaker URL to it. Update internal links to point at the survivor. The combined page usually ranks better than either did within a few weeks.
Redirect (301)
For a page you're removing that has links, history, or a close relative. Send it to the most relevant page that remains: the merged sibling, the hub for its topic, or a guide on the same subject. A redirect to the homepage is treated as a soft 404 by Google when the target isn't relevant, so pick a real destination or delete instead.
Noindex
For pages that must exist but shouldn't rank: tag archives, author archives on a one-author site, paginated lists, search result pages, thin category archives where a hub page does the job. The page stays, visitors can reach it, Google stops treating it as a candidate.
Delete (410 or 404)
For pages with nothing to pass on and nowhere sensible to go: the recipes on the bike-lights site. A 410 tells Google it's gone on purpose; a 404 gets there slightly slower. Either is fine. Remove the internal links to it first so you're not leaving dead ends.
Pages never to prune
- Anything with external links, however little traffic. The links are the asset; redirect if you must change the page.
- Pages under a year old. They haven't had time.
- Supporting pages that earn nothing. The how-to that three guides link to, the explainer that gets cited. The topical authority post covers why these count even at zero clicks.
- Seasonal pages out of season. A gift guide in March has no impressions by design.
- Pages with impressions and no clicks. Google is showing them; they need a title, not a redirect.
Once a year
- Run the content audit; export the twelve-month page data.
- Bucket every page. Expect to keep most.
- Merge the competing pairs first; they're the biggest wins.
- Redirect or delete the Remove bucket; fix the internal links that pointed at them.
- Note what you removed and why. Check in three months that nothing you kept lost ground.
Done once a year, after the site has history, it keeps a growing site readable as one site. The refresh post covers what happens to the Keep bucket; most of the work on a healthy site is there, not in the removals.
Doing it with AffSERP
The Merge bucket is where AffSERP earns its place in a pruning pass: when two of a site's guides compete, regenerating one with the X Best Review tool for a clearly different keyword (the "under $50" version, the "for commuting" version) gives it its own query to rank for, which is often better than merging.
The Refresh option handles the Keep bucket's stale guides without changing their URLs. Deciding which pages go in which bucket is the Search Console work, and it stays with you.
FAQ
Does deleting old posts help SEO?
Sometimes, and not for the reason usually given. There's no penalty for having thin pages; what pruning fixes is a site whose topic is blurred by off-topic posts, or pages competing with each other, or crawl time spent on pages nobody wants. Delete for those reasons and rankings on the rest tend to improve. Delete because a page is 'thin' and it's just gone.
Should I delete posts with zero traffic?
Not on that basis alone. A zero-traffic post that has links pointing at it, supports a topic, or is simply young is worth keeping. A zero-impression post after twelve months, off-topic, with no links, is the one to remove. Check impressions, not clicks: a page Google shows and nobody clicks is a title problem, not a pruning candidate.
Is it better to redirect or delete a pruned page?
Redirect if the page has any links from other sites or if there's a page on your site that covers the same subject; the links and whatever history it had pass along. Delete (a 410 or 404) only if there's nothing to pass and nowhere sensible to send a visitor. Noindex is for pages you need to exist but don't want ranked, like tag archives.
How often should an affiliate site prune?
Once or twice a year, after the site has at least a year of Search Console history. Pruning a site under twelve months old removes pages that haven't had time to rank. Do it as part of the content audit, with the same spreadsheet.
Keep reading
