Discussion starter: a content SEO question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- Content SEO
- Indexing
- Search Console
Has deleting or noindexing low-traffic posts ever produced a sitewide lift you could attribute with confidence, not just a tidier Search Console?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
The honest answer is that I am not aware of any public evidence that meets the bar you set. What exists are single-site before and after stories, and most of them overlap with a core update or with other work shipped at the same time (redirects, rewrites, template fixes). Google’s own documentation does leave room for a sitewide benefit, but it attaches that benefit to removing unhelpful content, not low-traffic content, and it calls deletion a last resort.
The bigger trap is measurement. Pruning mechanically improves the Search Console numbers most people look at, so a tidier Search Console can look exactly like a lift even when no surviving page gained anything. If you want to know whether pruning worked, you have to measure the pages you kept, against a counterfactual, over a window long enough to include the next core update.
What Google has said, and why it is narrower than it sounds
Google’s core updates guidance says: “Deleting content is a last resort, and only to be considered if you think the content can’t be salvaged.” It goes on to say that if you are considering deleting entire sections, that is likely a sign they were created for search engines first, and in that case “deleting the unhelpful content can help the good content on your site perform better.” The same page warns that it “could take several months” for Google’s systems to confirm a site-level change, and that it may mean waiting until the next core update.
In 2023, when CNET was reported to be deleting thousands of old articles, Google’s Search Liaison was blunter. Removing a page might help Google crawl other content on a massive site, “but it doesn’t mean we go ‘oh, now the whole site is so much better’ because of what happens with an individual page”. John Mueller made the related point in 2020 that low-traffic articles are not necessarily low quality, and that filtering should target content with real quality problems.
Those statements do not contradict each other. The lever Google describes is removing a body of content that makes the site look search-engine-first. Trimming the quiet long tail of an otherwise decent blog is a different intervention, and Google has not said it produces a sitewide effect.
Why low traffic is a poor proxy for unhelpful
A post can get little traffic for at least six reasons, and only one of them matches what Google’s guidance describes:
- the topic has small but real demand (a niche integration, a regional regulation)
- the post is new, or seasonal and outside its season
- a sibling post covers the same intent and wins (a consolidation case)
- it is outdated but the topic still has demand (an update case)
- it serves users who arrive from email, sales or internal links, not search
- it is thin, inaccurate or written for keywords rather than readers
Selecting by traffic sweeps up the first five and can leave the worst content in place, because thin pages that still rank for something survive a traffic threshold. If a sitewide quality signal exists in the form Google describes, a traffic-based cull is aimed at the wrong pages.
Why a tidier Search Console looks like a lift
If the heaviest members leave a slimming class, the class average drops without anyone losing a kilo. Search Console aggregates behave the same way. Average position and CTR are weighted across every impression, so removing pages that rank on page four with near-zero CTR raises both averages even if every remaining page is unchanged.
Hypothetical example:
A B2B blog has 1,400 posts and noindexes the 900 with fewer than five clicks a month. Before: 400,000 impressions a month, average position 28, CTR 1.2%, so about 4,800 clicks. After: 300,000 impressions, average position 19, CTR 1.55%, so about 4,650 clicks. The dashboard now shows a nine-place position gain and a CTR improvement, while clicks fell slightly. No surviving page moved.
Metric What pruning does mechanically What would indicate a real sitewide lift Indexed pages Falls by design Not evidence of anything on its own Average position Rises as low-ranking pages leave the average Surviving URLs improving on a fixed set of queries CTR Rises as low-CTR impressions disappear CTR on surviving queries rising at stable positions Total impressions Falls Survivors’ impressions above their forecast Total clicks Flat or slightly down Survivors’ non-brand clicks above a counterfactual Other reporting changes compound this. When Google stopped supporting the
&num=100parameter in September 2025, many sites saw desktop impressions fall and average position improve, while clicks were affected far less, according to Brodie Clark’s analysis. A prune shipped around then would have looked twice as successful as it was.Why clean attribution is structurally hard
The mechanism you are hoping for is the one that breaks your control group. SEO split testing works by changing one bucket of similar pages and comparing it with an untouched bucket. A genuine sitewide signal would lift both buckets, so the test would report no effect. Section-level holdouts have the same weakness.
Timing adds a second problem. If Google’s guidance is right that site-level reassessment can wait for the next core update, any lift is likely to arrive on the same day the whole index shifts. This year alone included the March 2026 and May 2026 core updates, plus the unannounced smaller core updates Google documents. Without sites that did not prune, you cannot separate “pruning made the update favour us” from “the update favoured us anyway”.
Third, pruning rarely ships alone. Practitioner write-ups of recoveries, such as Glenn Gabe’s analysis of remove versus improve, describe sites that removed content while also fixing technical issues and publishing better work, and the author explicitly does not claim removal alone caused the turnaround. That is useful field observation, but it cannot answer your question.
Finally, the treatment date is fuzzy. A noindexed URL drops out only when Google recrawls it, which can take weeks for low-priority posts, so the deploy date is not the day the change took effect.
How to design it so the answer means something
- Hypothesis. Removing a defined set of posts, selected on quality criteria and not traffic alone, raises non-brand clicks to untouched URLs beyond normal variance within a set window that includes the next core update.
- Survivor set. Every indexable URL that was not pruned, is not a redirect target, and was not edited during the window. Redirect targets gain for page-level reasons and must be measured separately.
- Primary metric. Weekly non-brand clicks to the survivor set, using the branded queries filter so brand demand does not move the result.
- Secondary metrics. Survivors’ impressions on a fixed query set, crawl frequency of survivors from logs, and organic conversions from survivors.
- Counterfactual. A seasonal forecast from at least a year of history, plus external comparison: tracked competitors’ visibility, or, for agencies and portfolio owners, the same intervention rolled out to comparable sites at different times.
- Treatment tracking. Record when each pruned URL actually left the index, and log every other deployment in the window with Search Console annotations.
- Decision rule, set in advance. Credit pruning only if survivors beat the forecast, the gain is broad across sections rather than concentrated in a few URLs, comparable sites did not rise at the same time, and the gain persists beyond one update.
On a single site, the strongest honest conclusion is usually “consistent with a sitewide effect”, not proof of one. That is still worth knowing, as long as it is labelled that way.
When pruning is worth doing anyway
A sitewide lift is the weakest reason to prune, because it is the hardest to prove. Stronger reasons: outdated advice that creates accuracy, legal or brand risk; overlapping posts that split links and relevance, where consolidation with a 301 produces page-level gains you can measure; very large sites where crawl capacity is genuinely constrained (Google’s crawl budget guide is aimed at sites with a million or more unique pages changing about weekly, or 10,000 or more pages changing daily); and editorial maintenance cost.
Post condition Action Main trade-off Low demand, accurate, distinct Leave it None worth the risk Overlaps a stronger post Merge and 301 Some queries unique to the weaker post may be lost Outdated, topic still has demand Update Editorial time Useful to users, not a search landing page Noindex, keep live Still crawled, and slower to drop out Thin or wrong, no links, no conversions Remove (404 or 410) Irreversible for users Weak content with backlinks 301 to the closest relevant page A 404 would discard those links Watch for this:
Low search traffic does not mean zero value. Before removing a post, check assisted conversions, backlinks, internal traffic from sales and support, and AI citations. Bing Webmaster Tools’ AI Performance report (public preview) shows page-level citations across Copilot and Bing’s AI summaries, which a traffic export will not.
A practical position
If this were my site, I would not prune to chase a sitewide lift. I would classify posts by the six causes of low traffic above, act on the ones where the action has its own justification (consolidate overlap, update decayed content, remove what is wrong or embarrassing), and then measure the survivors against a forecast through the next core update. If the survivors clearly outperform and comparable sites did not, you have the closest thing to confident attribution a single site can get. If they do not, the pruning still earned its keep on accuracy and maintenance, and you have avoided reporting a composition effect as a win.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.