What Is Content Pruning?

Content pruning is the systematic review of what has been published in order to decide what stays, what gets improved, what gets merged and what gets retired. The name comes from pruning plants, and the image fits: you do not cut for the sake of cutting, you cut what stops the rest from growing.

The need appears over the years. A site that publishes regularly accumulates articles for finished campaigns, landing pages from another era, old versions of guides, and pages created for a trade fair five years ago — and all of it is still there.

It is not the same as deleting. Most of the work is the deciding, and only a small part ends up actually retired; confusing the two is the origin of nearly every disaster told about this practice.

It deserves separating from two things it gets mixed with. It is not a technical clean-up: it is not about URLs the system generates — that is the territory of index bloat — but about texts somebody wrote and published on purpose. And it is not an update: refreshing an article is maintaining it, not deciding about it.

The difference matters because it changes who decides. Surplus URLs are handled by a rule; published texts have to be read, and no report does that. That is why pruning is slow, and why it almost always gets postponed.

Every page falls into one of four boxes, and the box decides the action. The two questions behind them are simple: does it receive visits, and is the content worth keeping?

What matters is that only one of the four ends in retirement. The two middle ones — good content with no visits, and visits landing on thin content — are the most numerous, and in both the right answer is to repair rather than remove.

Hence the order: measure first, decide after. A URL list sorted by visits is not enough; you also need to know what search intent whoever arrived had, and that shifts the box for many pages.

A fifth case does not fit the matrix and deserves separate treatment: content that cannibalises itself. Two or three articles answering the same search, textbook cannibalisation, are not bad individually, but together they take each other's place. The decision there is not which to delete but which keeps the topic — the rest are absorbed into it.

And a warning about the horizontal axis: "receives visits" means something different in January than in August. A tax return guide looks dead six months of the year, and measuring its traffic in the wrong month sends it to the retirement box for no reason. A full year of data is the sensible minimum.

Retiring does not mean the URL simply vanishes. If the content is merged into another page, a 301 redirect belongs there, because whoever arrives will find what they wanted elsewhere.

If there is no equivalent destination, the honest answer is a 404 or a 410. Redirecting to the homepage out of habit produces a soft 404 and leaves the visitor worse off than a clear error.

And there is a third route people forget: keep the page alive for those who need it, reachable through internal linking and take it out of the index with noindex. That fits anything with internal use, or for long-standing customers, that should not compete in results.

A precision about merges, where the most value is gained and the most is spoiled: merging is not gluing two texts one after the other. What is kept is whatever answered the search; the rest is surplus, and leaving it in turns the receiving page into a long, disordered text that no longer answers anything clearly.

And a detail forgotten in every case: the internal links that pointed at the retired page. A redirect keeps them working, but continuing to link to a redirecting URL leaves an unnecessary chain inside your own site. Links get updated to the new destination rather than left dangling.

You need three figures per page and none is enough alone: how many visits it brings, how many impressions it gathers, and how often it gets crawled. The first two come from Search Console and web analytics; the third from log file analysis.

A page with no visits but with impressions is not a retirement candidate: it appears in results and nobody picks it, which is a problem of title and description, not of existence. Separating those two cases prevents most unnecessary deletions.

And look at the links before touching anything. A barely visited page may be receiving backlinks, and retiring it throws that value out of the window instead of carrying it over through a merge.

There is a fourth signal that is not a figure and decides more than the three before it: whether the text is still true. A technical guide with screenshots of a changed interface, an article citing a repealed rule, or a comparison with prices from three years ago do not fail for being thin, they fail for being false — and that appears in no report.

So order the work by section rather than by the complete list. Reviewing everything published at once is unmanageable; reviewing a whole category lets you see the repetitions and decide with context — exactly what is missing when you look at a single page without its neighbours.

The first is bulk deletion by a single figure. A list of pages with no visits in the last ninety days throws a seasonal article, a technical guide the support team uses and web analytics never sees, and a genuinely dead page into the same bag.

The second is expecting immediate results. The effects of a clean-up show late, because they depend on what changed being crawled again — and in that interval it is easy to attribute to the pruning any movement that actually comes from elsewhere.

And the third is keeping no record. Without a list of what was retired, when and why, nobody can review the decision a year later, and any drop in traffic during that time becomes an argument without data.

A fourth mistake is doing it without telling anybody else. An article that looks dead may be linked from a live campaign, from an internal manual, or from the material the sales team uses; none of that appears in Search Console, and you find out when somebody complains.

And the last, the least intuitive: pruning content does not replace publishing it. A site with little good material does not improve by having less bad material, it improves by getting more good material. Pruning orders what is there; it creates nothing, and selling it as a growth strategy promises what it cannot deliver.