It sounds odd coming from someone who makes a living growing websites, but one of the recommendations I repeat most often when auditing blogs with a few years of history is to delete content. The pattern repeats so much that I can spot it five minutes into opening a Search Console account: a blog that has been publishing for five or more years accumulates hundreds of pages that no longer serve anyone, and that dead weight ends up getting in the way of the content that does perform.
That cleanup is called content pruning, and in this guide I explain what it is, why it works, when it makes sense to do it, and the exact method I use to run it: a 16-month Search Console export and four buckets where every URL lands. The full recipe is here, example numbers included.
What is content pruning?
Content pruning is the process of reviewing your site's complete page inventory and making an explicit decision for every URL: keep it and update it, optimize it, consolidate it with another page, or delete it. The garden analogy is fairly literal: you cut the branches that bear no fruit so the tree's energy concentrates on the ones that do.
The important part of the definition is "explicit decision for every URL". Serious pruning runs on data, never on deleting old posts by gut feeling, and deleting is only one of four possible destinations. In well-maintained sites it tends to be the least frequent one.
Why can deleting pages improve your rankings?
Because Google doesn't evaluate your pages one by one, in isolation. It crawls your entire site, indexes what it can, and draws quality conclusions by looking at the full inventory. If you want the detailed mechanics, I wrote a guide on how Google's algorithm works, but the short version is that your domain carries the average of what it publishes.
That has two practical consequences. The first is crawl budget: Google dedicates a finite amount of resources to crawling your site, and every visit it spends on a 2019 article with zero impressions is a visit it didn't spend on your new content. On small sites this doesn't matter, but above a few thousand URLs it starts to show, and one of the typical symptoms is that new content takes weeks to get indexed.
The second is the aggregate quality evaluation. Google's systems for measuring how useful a site is operate at the site level, and an inventory where 700 out of 1,000 pages haven't registered an impression in over a year is telling the evaluator something. Cleaning up that inventory improves the overall picture you're judged on.
When does pruning make sense?
Not always. If your blog has 40 posts, your problem lies elsewhere and your time returns more invested in a content plan than in pruning. The signals that the moment has arrived:
- Your blog has several hundred posts and more than three years of publishing.
- A large share of your URLs hasn't registered impressions in over a year.
- New content takes weeks to show up in the index.
- You have several pages competing for the same search and none of them quite takes off.
- A migration or redesign is coming, which is the best moment to stop dragging dead weight. If that's your case, I have a full checklist for redesigning without destroying your rankings.
How do you prune content step by step?
1. Export 16 months of Search Console data
In Search Console, go to Performance, filter by pages and export clicks and impressions per URL. The window is 16 months, not 12, for a concrete reason: exactly 12 months can cut a season in half and make you believe a seasonal page is dead. With 16 you cover the full annual cycle with margin, and it also happens to be the maximum history Search Console's interface keeps, so you're not leaving data on the table. If your site exceeds the standard export's 1,000-row limit, use the API or the Looker Studio connector to pull the complete list.
2. Cross it with business and link data
The clicks-and-impressions sheet is the base, but two columns are missing before deciding anything. The first is business: if you have conversions or assists measured per page, crossing them here saves you from deleting a page that converts little traffic but very qualified traffic. The second is backlinks: check which URLs have external links pointing at them, with Ahrefs or with Search Console's own links report. A page with no traffic but good links holds authority you don't want to throw away.
3. Classify every URL into one of four buckets
With those columns in place, every URL falls into exactly one bucket:
- Bucket 1: gets clicks and contributes to the business. It stays, with periodic updates on the calendar. This is your living inventory.
- Bucket 2: has impressions but few clicks. Google shows it and people don't click. The first things to review are the title and the search intent, which is SEO writing work more than strategy work. If another page of yours targets the same intent, consolidate them into one.
- Bucket 3: no impressions, but it has backlinks or usable content. Don't delete it. Consolidate it with a 301 redirect to the page that does perform on that topic, so the links and the useful content add up where it matters.
- Bucket 4: no impressions, no backlinks, no business role. Delete it. And deleting means deleting: the URL returns a 404 or 410 and comes out of the sitemap. There's no need to redirect a page that nobody visits or links to anywhere.
To give you a sense of real proportions: in an 1,800-post blog I reviewed with this method, 240 posts concentrated 92% of the clicks and landed in bucket 1. Another 610 had impressions with minimal CTR and went into the optimization queue. The remaining 950 hadn't registered a single impression in 16 months: 130 had backlinks and were consolidated with 301s, and the other 820 were deleted. Nearly half the blog went offline, and traffic didn't move, because those pages weren't contributing any.
4. Execute in batches and document
Don't delete 800 pages on a Friday afternoon. Execute in batches, keep a record of which decision you made for each URL and why, update the internal links that pointed to deleted pages, and remove dead URLs from the sitemap. That record becomes gold later: when someone asks in six months "what happened to that article?", the answer lives in a spreadsheet instead of in the memory of whoever did the work.
Which pages should you leave alone?
Four categories stay out of this process without a second thought. Legal pages, because their job was never to bring traffic. Seasonal content, which can be in a legitimate pause (which is exactly why you pull 16 months of data). Pages with a commercial role, like pricing or case studies, which can bring zero organic visits and still close deals. And content published less than six months ago, which simply hasn't had time to mature in the index.
What results can you expect (and which ones you can't)?
Here's the part where I'd rather lower your expectations than inflate them. The pages you deleted weren't contributing clicks, so your traffic curve shouldn't move the day after, and be suspicious of any case study promising immediate jumps from deleting content.
What changes is something else. Google spends its crawl on fewer, better URLs, and on large sites that usually shows up as new content getting indexed faster. The 301 consolidations concentrate signals that were scattered, and it's common for the consolidated cluster to improve positions in the following weeks. And the average quality of your inventory goes up, which is one of those improvements that's hard to attribute to a single metric but pushes your whole organic visibility in the right direction. Measure those three things: indexing time for new content, coverage in Search Console, and performance of the pages that received consolidations.
What does AI have to do with all this?
More and more. When ChatGPT, Perplexity or Gemini look for sources to answer a question, they run into your complete inventory, just like Google does. A site where most pages are thin, outdated content gives those systems more opportunities to cite you wrong, summarize an old version of your position, or simply pick a tidier source.
That said, the mechanism here is less measurable than in Google, so treat it as a secondary benefit of pruning rather than the main reason to do it. The main reason is still letting your best content compete without dead weight on top.
To wrap up
The diagnosis takes one afternoon and no budget. Export your 16 months of Search Console data, filter the URLs with zero impressions and count how many there are. Then open 20 of those pages at random and ask yourself, honestly, whether they serve anyone today. With those two numbers you already know the size of your bucket 4 and whether the full project is worth it. The deleting decision comes later, calmly and in batches: the first step is simply looking at the inventory you've spent years not looking at.