Content pruning SEO is the practice of auditing every indexed URL on your site and then deleting, redirecting, merging, or rewriting the ones that earn nothing. It is not a purge. Roughly 60–70% of the pages that fail an initial screen turn out to be fixable rather than deletable, and knowing which is which is the entire skill.
The reason to do it is straightforward. Crawl budget, internal link equity, and your own attention are finite. A hundred dead pages dilute all three, and on a site over a few thousand URLs the effect on how quickly Google recrawls your good pages becomes measurable.
What counts as a prunable page
A page is a pruning candidate when it meets all four of these conditions over a 6-month window:
- Fewer than 10 organic sessions total.
- Fewer than 100 Search Console impressions.
- No ranking keyword inside the top 20.
- Zero conversions, form fills, or assisted conversions.
All four, not any one. A page with 8 sessions and 4 conversions is your best-performing asset per visit. A page with 3,000 impressions and no clicks has a title and intent problem, not an existence problem. The four-condition test exists specifically to stop you deleting those.
Exclude from the audit entirely: pages under 90 days old, anything intentionally noindexed, legal and policy pages, contact and about pages, and anything that earns links or referral traffic regardless of search performance.
The data you need to pull
Build one spreadsheet with one row per URL. Six columns is enough:
- URL — from a full crawl, not a sitemap. Sitemaps lie by omission.
- Organic sessions, last 6 months — GA4.
- Impressions and clicks, last 6 months — Search Console, page-level report.
- Best ranking position and the query — Search Console or a rank tracker.
- Referring domains to the URL — a backlink index. This is the column that saves pages from deletion.
- Word count and publish date — from the crawl.
SEO Rocket runs the crawl side of this — an instant quick scan of about 25 pages, or a deep full-site crawl verified past 900 pages, returning actual URLs, titles, and H1 text rather than counts. Connect Search Console and GA4 in the same workspace and you have four of the six columns without a manual export. The backlink column comes from the site explorer.
The four decisions and when to use each
Every candidate URL gets exactly one of four outcomes. Do not invent a fifth.
1. Improve (default for most candidates)
Use when the page targets a keyword with real volume and the SERP shows the intent is still servable. The page failed because it is thin, outdated, or badly matched to intent — not because the topic is dead. Rewrite to match what page one actually contains, refresh the data, fix the title and meta description, then request indexing.
Give it 90 days before re-judging. Rewrites take one to two crawl cycles to be re-evaluated, and a rewritten page that has not moved after three weeks has not failed yet.
2. Merge and 301
Use when two or more pages target the same intent and split each other’s signals. This is the single most common finding on sites over 200 posts, and the easiest win. Pick the stronger URL — usually the one with more referring domains, not the newer one — fold the unique material from the others into it, then 301 the weaker URLs to it.
Check internal links before you redirect. Any link pointing at a merged URL should be updated to the target, not left to pass through a redirect. Chained redirects accumulate quietly and cost you crawl efficiency.
3. Redirect without merging
Use when a page has referring domains or historic value but nothing worth keeping on it. Discontinued products, expired events, retired features. 301 to the closest genuinely relevant page. Not the homepage — Google treats an irrelevant redirect as a soft 404, and you get nothing.
4. Delete and return 410
Use only when a page has no traffic, no rankings, no backlinks, and no relevant redirect target. Auto-generated tag archives, duplicate paginated listings, test pages, and imported junk. Return 410 Gone rather than 404 — it signals deliberate removal and Google drops it from the index faster.
How to run the audit in one afternoon
A workable sequence for a site of 200 to 2,000 URLs:
- Crawl the full site. Filter to indexable, canonical, HTML pages returning 200.
- Join Search Console and GA4 data by URL.
- Apply the four-condition filter. Everything failing all four goes into the candidate list.
- Sort candidates by referring domains, descending. Anything with links is never a delete — it is a redirect or an improve.
- Group the rest by target topic. Duplicates within a group become merges.
- Assign one of the four outcomes per URL. Cap the first pass at 50 URLs.
- Execute, update internal links, resubmit the sitemap.
- Note the date. You need it to attribute the result.
Cap the batch. If you prune 400 URLs in one week and traffic moves, you will never know which change did it. Batches of 30 to 50, two weeks apart, keep the signal readable.
What to expect afterwards, honestly
Do not expect a traffic spike from deletion alone. Removing dead pages removes dead traffic — which is zero — so the direct gain is zero by definition. The gains are indirect and slower: faster recrawl of pages that matter, consolidated signals on merged URLs, and less of your own attention spent on pages that were never going to work.
Merges are where measurable lifts show up, and typically four to eight weeks out. Expect the merged URL to wobble first — often down for two or three weeks — before settling above where either original sat. That dip is normal reindexing, not a mistake, and daily jitter of two or three positions on top of it is standard noise. Read the trend at week six, not day four.
Mistakes that cost real traffic
Five failures account for nearly every pruning disaster:
- Deleting pages with backlinks. You throw away authority you cannot buy back cheaply. Always check referring domains first.
- Redirecting everything to the homepage. Treated as a soft 404. Equity does not transfer.
- Using noindex instead of removing. Noindexed pages still get crawled, so you keep the crawl cost and lose the page.
- Pruning during a core update. You will not be able to separate your effect from Google’s.
- Judging by traffic alone. Low-traffic pages that convert, earn links, or support internal linking are load-bearing.
Making it a habit rather than an event
Run a full prune once a year and a lightweight review quarterly. The quarterly pass only needs one question: which pages published more than six months ago still have zero top-20 rankings? That list is short, and acting on it early is far cheaper than a thousand-URL cleanup two years later.
The discipline scales. The playbook behind SEO Rocket came from a site that passed 30,000 published, ranking pages and kept growing through core updates — and at that volume, pruning stops being optional. Publishing is only half of a content operation; deciding what stops existing is the other half, and it is the half most sites never build.
