Did your content pruning work? How to separate the pruning from everything else

KEY TAKEAWAYS

1. A before and after total mixes pruning with paid shifts, branded demand, updates and reporting changes.
2. Prune one section and hold a matched one back. The difference between them is the pruning effect.
3. Measure non-branded clicks, outside core and spam update windows, at two, six and twelve weeks.
4. Redirect only to close matches and return 404 or 410 otherwise. Loose redirects leave debt behind.

On 19 October 2026, Search Engine Land published a content pruning case study by Anna Crowe. A small B2B healthcare site removed or hid a third of its pages, which had earned 3.1% of its clicks. Four months later, Search Console clicks were up 37%. The author is careful to say she cannot prove the pruning caused it. Her caveats are the useful part, because they list almost every reason a pruning result is hard to read.

What did the case study actually show?

In December 2024 the team crawled 600 URLs and marked 198 for removal. Those pages had earned 144 clicks, and 140 of them had never earned one. Old event posts and dated news were redirected to hub pages and then set to noindex. Thin customer logo pages were unpublished and redirected. Podcast stubs and syndicated reposts were set to noindex.

From December 2024 to April 2025, Search Console clicks rose from 5,710 to 7,840. GA4 organic sessions rose from 10,404 to 16,103. Over the same period, paid search sessions fell from 11,863 to 182, because the paid budget was switched off. Total search sessions, organic plus paid, fell 27%.

The author lists the other changes in that window. More than 80% of organic clicks were branded searches. Page speed slowed from about 2 seconds to 6 seconds in January. Rewritten content went live in April. Google ran updates in December and March. Any of these could move the organic line without a single page being pruned.

The case is honest about this, and that is what makes it a good lesson. A before and after total cannot tell you what the pruning did.

Why do Google and SEOs disagree about pruning?

Google's position has been steady. When CNET deleted thousands of articles in 2023, Danny Sullivan said Google's guidance does not encourage removing content for being old, and that deleting a page does not make the rest of a site better. John Mueller called CNET's plan reasonable but said the SEO effects were "unlikely to be noticeable."

Google's current core updates guide is more open. It calls deletion a last resort for content that cannot be salvaged, and says removing unhelpful content may help the rest of a site perform better. In September 2026 Mueller went further on one kind of site. He said Google's systems may have lost faith in a site built on low-value programmatic pages, and that recovery takes time and significant effort.

Practitioner cases report large gains. Ahrefs describes a vehicle valuation site cut from about 4.86 million pages to 1,500, with organic visits up 160%. Animalz reported QuickBooks traffic up 44% at peak season after removing about 2,000 posts.

The two views fit together once you look at scale. The big wins come from sites where low-value pages were a large share of the whole, often millions of them. Google's crawl budget guide is written for sites with over a million pages that change weekly, or over 10,000 pages that change daily. On a site of a few hundred pages, crawl is rarely the constraint, and the effect of removing 100 unvisited posts is likely small. Ahrefs also notes that pruning usually ships with other changes, which makes the effect hard to isolate. Both sides agree on one point: remove only what cannot be saved.

Which confounds hide inside a before and after total?

The update calendar matters more than it looks. Google's status dashboard shows the December 2024 core update ran from 12 to 18 December, followed by a spam update from 19 to 26 December. The March 2025 core update ran from 13 to about 27 March. A December 2024 prune and an April 2025 read straddle all three.

Reporting changes are the same trap. After Google stopped supporting the num=100 parameter on 12 September 2025, Search Console impressions and average position shifted without rankings changing. Search Engine Land advised treating data from 13 September 2025 as a new baseline.

How do you set up a pruning test you can read?

The guides from Semrush and SearchEngineWorld recommend staged rollouts and before and after metrics. None of the top guides we read uses a control group. This method adds one.

1. Snapshot first. Export clicks, impressions and position per URL and per query, plus the indexed page count. The Search Engine Land case lists the missing before snapshot as its first mistake.
2. Split the candidates. Group prune candidates by section. Pick two sections with similar traffic and content types. Prune one. Leave the other untouched as a holdout for at least twelve weeks.
3. Measure non-branded clicks. Search Console's branded queries filter splits branded from non-branded queries. Google expanded it to all eligible sites in March 2026. It uses an AI-assisted classifier, so spot-check it, and some low-volume properties do not qualify.
4. Time it outside updates. Check the Search Status Dashboard before each batch. Google's own method for core updates compares a full week after a rollout ends with the week before it began. Use the same discipline.
5. Log every other change. Paid budgets, speed, rewrites, template edits and migrations, with dates.
6. Redirect only to close matches. Google's redirect guide says a 301 makes the target canonical. Glenn Gabe's 2016 migration cases, and Mueller's comments he cites, say redirects to less relevant pages can be treated as soft 404s. Where no close match exists, return a 404 or 410. Google treats all 4xx codes except 429 the same way. Its crawl guide also advises against noindex as a way to save crawl capacity.
7. Read at two, six and twelve weeks. Google says some changes show within days and others take months. A single read at four weeks is a guess.

A worked example

Take a hypothetical software blog with two similar sections: tutorials and industry news. Each has about 120 prune candidates. The team prunes the news section and leaves tutorials alone.

Twelve weeks later, non-branded clicks to the news section are up 14%, from 2,000 a month to 2,280. Non-branded clicks to the untouched tutorials are up 11%, from 1,800 to 1,998.

The naive read is a 14% pruning win. The holdout says most of that rise happened anyway, from demand, an update or a change the team forgot. The pruning effect is the gap, about 3 points, or roughly 60 clicks a month on the news section. That may still be worth it. It is a very different number to report, and it tells the team whether to prune the next section.

If the holdout had risen more than the pruned section, that would be a signal to stop and check redirects before going further.

What does pruning leave behind?

Redirect debt. By September 2026, the Search Engine Land case site had 278 redirecting URLs among 923 crawled, and 3,390 internal links still pointed at them. Some old URLs had also gone to generic hubs rather than close matches. Unpublishing pages before setting up redirects produced 404s.

A cautious counterexample is useful here. Sunny Kumar at TheGuideX reviewed 201 legacy URLs one at a time and deleted only 5. The rest were redirected, rewritten or moved out of the main navigation. He worked in batches of 20 to 50 with a rollback trigger. His traffic rose from a low of 285 clicks in February 2026 to 1,573 in June, still about 96% below the site's former level. He declined to call it a recovery, which is the right instinct.

After each batch, recrawl. Fix redirect chains, update internal links to point at final URLs, and check that nothing you meant to keep returns a 404.

How does this fit with other diagnostics?

The same holdout logic applies to any traffic change you want to explain. Our spam update diagnosis uses update windows the same way, and rank tracker data needs the same scepticism before you act on it. When you do rewrite rather than remove, judge pages against Google's four main content attributes.

Pruning is a reasonable tool when it removes pages nobody needs. Report it with a control, or report it as unproven. If you want your search presence measured alongside how AI engines see your site, the waitlist is open.

Questions

Does content pruning improve SEO?

Sometimes, and mostly on sites with many low-value pages. Google says deletion is a last resort and that removing unhelpful content may help the rest of a site. John Mueller called the SEO effect of CNET's 2023 plan unlikely to be noticeable. The large gains in published cases come from very large sites or sit alongside other changes.

How long does it take to see results from content pruning?

Google says some changes show within days but its systems can take several months to confirm a site's content is helpful. Plan reads at about two, six and twelve weeks. Do not judge a batch during a core or spam update rollout. Compare full weeks outside those windows.

Should I delete, redirect or noindex pruned pages?

Redirect with a 301 only when a close match exists. Otherwise return a 404 or 410, which Google treats the same way. Mueller has said redirects to the homepage or unrelated pages can be treated as soft 404s. Google's crawl guidance advises against noindex as a way to save crawl capacity.

How do I know if traffic gains came from pruning or something else?

Use a holdout. Prune one section, leave a similar section untouched, and compare non-branded clicks for both over the same weeks. Search Console's branded queries filter splits the data. Record paid search changes, site speed, rewrites and migrations, since each can move organic numbers on its own.

Sources

1. Search Engine Land, Content pruning case study. Cutting 33% of pages that drove just 3% of clicks searchengineland.com
2. Google Search Central, Google Search's core updates and your website developers.google.com
3. Google Search Central, Optimize your crawl budget developers.google.com
4. Google Search Central, How HTTP status codes, and network and DNS errors affect Google Search developers.google.com
5. Google Search Central, Redirects and Google Search developers.google.com
6. Google Search Status Dashboard, Ranking incident history status.search.google.com
7. Search Engine Land, Google expands Search Console branded queries filter searchengineland.com
8. Search Engine Land, Why Google Search Console impressions dropped and how to interpret the data searchengineland.com
9. Search Engine Land, Google warns against content pruning as CNET deletes thousands of pages searchengineland.com
10. Search Engine Roundtable, Google Says CNET's Content Pruning Plans Were Reasonable But SEO Impact Would Be Unnoticeable seroundtable.com
11. Search Engine Roundtable, Google Can Lose Faith In Sites Based On Low Value Programmatic SEO seroundtable.com
12. Ahrefs, Content pruning ahrefs.com
13. Animalz, How QuickBooks Nearly Doubled Traffic by Deleting Half Its Content animalz.co
14. TheGuideX, Content Pruning in SEO. Why I Deleted Only 5 of 201 URLs theguidex.com
15. GSQi, Proof That 301 Redirects To Less-Relevant Pages Are Seen As Soft 404s To Google gsqi.com
16. Semrush, Content Pruning. A Step-by-Step Guide to Improving Your SEO semrush.com
17. SearchEngineWorld, Content Pruning Measuring Success searchengineworld.com

NEXT STEP

See where your brand stands.

Your Digital Authority Score, who AI recommends for your target questions, and the placements that close each gap. Your first report is free.

Get your free report →