Content pruning for stores: merge, redirect or leave

Pruning is not deleting pages that get no traffic. It is deciding, page type by page type, whether a URL should stay as it is, absorb another page, hand its value to a better page, or leave the index while staying on the site. Deleting is one of four outcomes and it is the least often correct.

Stores accumulate URLs faster than any other kind of site. Every discontinued product, every seasonal collection, every tag page a theme generated, every blog post written to fill a calendar. A catalogue of two thousand products can carry twenty thousand indexable URLs without anyone deciding it should. This is how to work through that without losing revenue.

Start by classifying, not by traffic

Pull every indexable URL from a crawl and join it to three data sources: clicks and impressions from Search Console over twelve months, sessions and revenue from analytics, and internal link count from the crawl. Twelve months matters, because a seasonal page with zero traffic in March may be the page that earns in November.

Then sort every URL into one of these types before looking at a single number:

• Product pages, in stock
• Product pages, out of stock or discontinued
• Collection and category pages
• Filtered or faceted URLs
• Tag and vendor pages generated by the platform
• Blog and guide content
• Landing pages built for campaigns
• Utility pages: policies, account, search results

The rule differs by type. Applying a single traffic threshold across all of them is how stores delete their seasonal collections and their newest product pages in the same afternoon.

The four outcomes

Keep and improve. The page has a job and does it badly. It gets rewritten, not removed.

Merge. Two or more pages compete for the same intent. One becomes the destination, absorbs whatever the others said that was worth keeping, and the others redirect to it.

Redirect. The page has no future but has links, history or a URL people still reach. A permanent redirect sends it to the closest equivalent page. Closest matters: a redirect to the homepage tells Google and the shopper nothing, and is treated as a soft error rather than a signal.

Leave the index, stay on the site. A noindex directive with the page still reachable and still functional. This is the answer for pages that serve shoppers but should not compete in search: internal search results, thin filter combinations, thank you pages, paginated tag archives.

Actual deletion, returning a gone status with no replacement, is right in one situation: the page should never have existed, has no links, and no equivalent page exists. That is rarer than it sounds.

Product pages

In stock, low traffic. Keep. A product page with no search traffic still converts visitors who arrive from a collection page, a campaign or a feed. Pages that are thin because they carry a supplier description are a rewriting job, not a pruning job.

Temporarily out of stock. Keep the page live and indexable, show the status honestly, and offer a back-in-stock signup. Removing it discards accumulated ranking that returns when the stock does.

Permanently discontinued. This is the real decision. If a direct successor exists, redirect to it. If a close equivalent exists, redirect to the category rather than to an unrelated product. If neither exists, keep the page live with the status clearly marked, links to alternatives, and let it fall in ranking naturally. Should out of stock product pages be deleted has the longer version, and out of stock page handling covers the on-page treatment.

Variant URLs. Where a theme generates a separate URL per size or colour, those should canonicalise to the parent unless each variant genuinely has its own demand. The canonical tag entry explains what canonicalisation does and does not consolidate.

Collection and category pages

Collections are where merging pays best, because stores routinely create several that mean the same thing. Blue dresses, dresses in blue and blue dress collection all target one query, and each holds a share of the internal links.

The merge process:

1. Group collections by the query they are trying to answer, not by their titles.
2. Within a group, pick the destination: the one with the most links, the most impressions, and the URL that reads best.
3. Move any unique product assignments and any genuinely useful copy into the destination.
4. Redirect the others.
5. Update the navigation and any internal links pointing at the redirected URLs, so the redirect is a safety net rather than the path.

Empty collections are the other common case. A collection with no products should either be filled, redirected to its parent, or removed from navigation and left out of the index. Leaving an empty collection indexable produces a page that ranks briefly and disappoints everyone who lands on it.

Before merging aggressively, check whether the pages actually compete. Two collections that each rank for different terms are doing separate jobs. Do subcategory pages help SEO covers where the split earns its place, and category page SEO is the wider treatment.

Faceted, tag and vendor URLs

These are the largest source of index bloat on most stores and the fastest to fix.

The rule: a facet combination earns indexing when it matches real demand and produces a page a shopper would want to land on. Waterproof hiking boots is a page. Waterproof hiking boots in size nine sorted by price is not. Everything that does not clear the bar gets a noindex, and any parameter that only reorders the same set should be handled so crawlers are not walking through thousands of near-identical results.

Platform-generated tag and vendor archives are usually the same story. Keep the ones that correspond to a real browsing intent, such as a brand a shopper searches by name, and remove the rest from the index. The decision framework is in faceted navigation rules, and if these URLs are already indexed, faceted navigation URLs indexed is the recovery path.

Blog and guide content

Sort store content by the job it does, then decide:

• Buying guides and comparisons that support a category. Keep, update, and link them from the category they support.
• Posts that answer a real customer question. Keep. These are what AI answer engines and search both draw on.
• Several short posts circling one topic. Merge into one thorough guide, redirect the others, and keep the strongest examples and details from each.
• News, launches and event posts. Leave them. They are a record, they cost nothing, and they occasionally earn a link. Redirecting old announcements to a category is worse than leaving them.
• Posts written for a keyword with no relationship to what you sell. These are candidates for merging into something useful or for noindex. Removing them from the index costs nothing if they never sent a buyer.

Does an online store need a blog covers what the content is for in the first place, which is the standard to prune against.

What pruning actually improves

Be honest about the mechanism, because the benefit is often overstated.

Pruning helps by concentrating internal links and topical clarity onto fewer, stronger pages, and by keeping crawlers on the pages that matter. On a large catalogue where crawling is spread across tens of thousands of low-value URLs, that second effect is real, and Google’s own guidance on crawl budget is aimed at exactly that situation. On a store with a few hundred pages, crawl efficiency is not the constraint and the benefit is mostly the clarity.

What pruning does not do is lift a site by removing pages that were simply not popular. If the page is good and serves someone, low traffic is not a reason to remove it.

Sequence and safety checks

1. Export the current state: all URLs, their status codes, their traffic and their links. Keep it. You will want it in three months.
2. Do the merges first, in one batch per group, and update internal links in the same change.
3. Apply noindex to the facet and tag sets next, and confirm the pages remain crawlable so the directive can be seen.
4. Handle discontinued products last, in batches by category, so any traffic effect is attributable.
5. Verify the redirect map: no chains longer than one hop, no loops, no redirects to pages that themselves redirect, no redirect to a page that returns an error.
6. Watch impressions by page group rather than site totals for eight to twelve weeks. Movement takes that long, and site totals will hide it.

Do not run a pruning project during peak season, and do not run one in the same month as a redesign or a replatform. When two changes land together, neither can be read. If a batch is followed by a decline that does not recover in a month, the export from step one is what lets you put a page back.

---

Sources

• Google Search Central: Redirects and Google Search
• Google Search Central: Block search indexing with noindex
• Google Search Central: Consolidate duplicate URLs
• Google Search Central: Large site owner's guide to managing crawl budget