A reader asked me this question while working for a reseller. The manufacturer kept withdrawing references, the matching product pages still brought in a trickle of organic traffic, and nobody knew whether to delete them, redirect them or leave them online. He also asked about the pages that brought nothing at all — is a well-built 404 page enough on its own?
I answered in a few lines, splitting the problem in two: permanently discontinued on one side, temporary stock-out on the other. Correct, and far too short. On a catalogue that moves, this decision gets taken dozens of times a year, and the cumulative effect shows up in index coverage, in internal linking and in the way Google crawls the site.
Here is the long version: the criteria I use in an audit, and what Google Search Central documentation says about 404, 410 and permanent redirects.
Three situations, not two
The first sort is made on the nature of the withdrawal, not on the traffic the page receives. Three cases, three different treatments.
- Temporary stock-out. The product returns in days or weeks. The page should neither disappear nor be redirected: keep it on a 200, state the expected restock date, offer a back-in-stock alert.
- Permanently discontinued, obvious successor. The manufacturer replaces the reference with an equivalent model. The search intent is still satisfied elsewhere on your site, so a permanent redirect is justified.
- Permanently discontinued, no replacement. This was the reader's case, and the awkward one. Nothing equivalent exists, so the question becomes: does this page still hold value of its own?
Catalogue churn is not marginal. For one order of magnitude — French market data, quoted as such — Fevad's annual review put online sales in France at €196.4 billion in 2025 across 3.2 billion transactions, as reported on 11 February 2026 by E-commerce Mag. In any market, the faster an assortment turns over, the more dead pages pile up behind it.
What Google actually does with 404, 410 and 301
Much of the "404 or 410" debate rests on a belief the documentation does not support. On its page about HTTP status codes, updated on 4 February 2026, Google states that all 4xx errors except 429 are handled in the same way: the content is treated as non-existent, the URL is dropped from the index, and crawling of it tails off. A 410 does not meaningfully speed up de-indexing. It documents an intention, which is useful for your team and your tooling, but it is not an SEO lever.
On redirects, the same page notes that crawlers follow up to ten redirect hops by default, and that a 301 is a strong signal in favour of the target, where a 302 is only a weak one. The dedicated redirects page, updated on 14 April 2026, adds that the indexing pipeline uses a permanent redirect as a canonicalisation signal: it is the target that subsequently appears in results.
| Situation | Treatment | Why |
|---|---|---|
| Temporary stock-out | 200, page kept, availability declared | The page keeps its history and its rankings while you restock |
| Discontinued, direct successor exists | 301 to the replacement product | Search intent is met by a page that really is equivalent |
| Discontinued, no successor, page holds no value | 404 or 410 plus a useful error page | Clear signal, no ghost page left in the index |
| Discontinued, page holds real value (links, brand, documentation) | 200 kept, content repurposed | Inbound links and informational traffic still do work for you |
| Reference replaced by a whole range | 301 to the precise sub-category | A relevant target, far closer than the home page |
Redirecting without manufacturing soft 404s
The most common mistake I meet in audits is the bulk redirect to the home page. Technically everything answers 200 and the dashboard looks clean. In practice the target does not answer the original query, and Google may classify those URLs as soft 404s — pages presenting themselves as valid without holding the requested content. Search Console help is explicit: return a genuine 404 for pages that really are gone, or make the page substantive enough to deserve its 200.
Three rules. First, the target stays in the same universe as the source: replacement product, failing that the sub-category, failing that the category. Second, never redirect "by default" to the home page — if the URL is worthless, a 404 is more honest. Third, clean up redirect chains as migrations pile up: ten hops is a limit, not a target. Broken URLs inherited from older catalogue versions belong in a technical SEO audit, not in a monthly fire-fight.
When a dead page deserves to stay online
A discontinued product page can still carry real value. Three signals push me towards keeping it on a 200: quality external links pointing at it, residual traffic on brand or manufacturer-reference queries, and a lasting after-sales need — manuals, spare parts, compatibility lists, consumables.
The page then changes nature. It stops being transactional and becomes a reference page: an explicit "no longer manufactured" notice, the withdrawal date, the specifications kept intact, links to the recommended replacements and to the documentation. The buy button goes, the content stays. This is the only setup that keeps the value of inbound links without misleading the visitor.
Declaring availability in structured data
What the visitor sees and what the crawler reads should agree, and that agreement runs through the availability property. Google's merchant listing documentation lists the accepted schema.org values, among them InStock, OutOfStock, BackOrder, PreOrder, SoldOut and Discontinued; the full list sits on the merchant listing structured data page. The Discontinued value in the schema.org vocabulary exists for exactly this case.
<script type="application/ld+json">
{
"@context": "https://schema.org",
"@type": "Product",
"name": "Reference XYZ-100",
"sku": "XYZ-100",
"offers": {
"@type": "Offer",
"availability": "https://schema.org/Discontinued"
}
}
</script>Two things to watch. Mark-up announcing InStock on a page displaying "unavailable" is an inconsistency that eventually shows. And if the page stays on a 200 with no way to buy, an active offer and price make little sense: remove them, or switch to Discontinued without a live price.
Internal linking and crawl budget
Deleting a page without touching the internal links leaves those links pointing at nothing. Before switching a URL to 404, I remove the links from categories, related-product blocks, filters, product feeds and the XML sitemap. A URL absent from the sitemap and unlinked internally stops being advertised as important.
On crawl budget, keep a sense of proportion. In its crawl budget guide, updated on 22 July 2026, Google says the subject mainly concerns sites with more than a million pages updated weekly, or more than 10,000 pages changing daily. The same guide notes that a 404 is a strong signal to stop crawling a URL, while soft 404s keep being crawled and keep consuming budget. For a shop with a few thousand references: crawl budget is not your main problem, but a steady supply of soft 404s is. It helps to know how Google discovers, indexes and refreshes pages before you touch anything.
One last recurring question: the Search Console removals tool. It helps in an emergency, but Google is clear that requests made in the tool last around six months. It is not a catalogue management method. It is a plaster.
The procedure I apply to a catalogue
- Export the discontinued product URLs, each with its organic traffic over the last twelve months, its external inbound links and any residual conversions.
- Decide page by page using the table above: stock-out, successor, standalone value, or nothing.
- Put the 301s in place towards relevant targets, checking that no chain is created.
- Switch the valueless pages to 404 or 410, with an error page offering internal search and the main categories.
- Clean the linking: internal links, sitemap, product feeds, faceted navigation.
- Watch index coverage in Search Console for eight to twelve weeks, soft 404s in particular.
A simple Looker Studio dashboard connected to Search Console is enough to see whether the traffic recovered by the target pages offsets what the removed pages lost. When the shop is young and structural decisions are not yet frozen, setting these rules at design time costs far less, and it is one of the things worth settling when you scope an SEO audit.
And to answer the reader's last question: a custom 404 page is useful, but it does not replace the decision upstream. A good error page keeps the visitor on site; it says nothing to Google about what you meant to do with the URL. That message is carried by the HTTP status code and by your internal links.
Common questions
Should a discontinued product return a 404 or a 410?
Either works. Google's documentation states that all 4xx errors except 429 are handled the same way: the URL is dropped from the index and crawling of it tails off. A 410 expresses a deliberate removal, useful for your own monitoring, but it does not accelerate de-indexing.
Can I redirect every deleted product page to the home page?
It is a poor idea. When the target does not match what the visitor was after, Google may treat the page as a soft 404 — a URL answering 200 without the expected content. Redirect to the replacement product or the precise sub-category, and return a genuine 404 when no relevant target exists.
What should I do with a product page that is only out of stock?
Keep it online on a 200 and do not redirect it. State the unavailability clearly, offer a back-in-stock alert and suggest alternatives without forcing anyone off the page. Declare availability in the structured data too, with OutOfStock or BackOrder as appropriate.
How do I keep the value of external links pointing at a product that no longer exists?
Two options. If a successor exists, a permanent redirect passes the signal to that product, which Google then treats as canonical. If nothing equivalent exists but the links are numerous and good, keep the page on a 200 and turn it into a reference page with no buy button.
Which availability value should structured data use for a discontinued product?
The schema.org vocabulary provides Discontinued, which Google recognises among the accepted values for the availability property. It says the product is no longer sold, unlike OutOfStock, which signals a temporary gap. If no purchase is possible, remove the offer and the price rather than leave commercial mark-up active.
Does the Search Console removals tool settle the question?
No. It hides a URL from search results temporarily. Google states that these requests last around six months, after which the URL can reappear if the page still exists. The durable answer is the right HTTP status code, a noindex tag, or actually removing the content.