Discontinuing a product feels like a purely commercial decision. It stopped selling, the supplier changed, the range was rationalised, so it comes off the site. The part nobody costs is what the page was doing while it was there.
That page had an address. Other pages linked to it, possibly other websites did too, and search engines had formed a view of what it was about and how well it answered a query. Some of those queries were about the product by name. Others were broader, and the page happened to be a good answer.
Remove it carelessly and all of that is discarded at once. Do it to two hundred products over two years, which is normal turnover in plenty of categories, and the store is quietly resetting a meaningful part of its search presence while its team is busy building new pages to replace what it just threw away.
The fix is not to keep everything forever. It is to retire products deliberately, with a destination.
What disappears with the address
Four things go at once, and only one of them is obvious.
The position itself. Whatever queries the page ranked for, it stops answering, and the traffic goes to whoever is next in line.
The links. Any external link pointing at that page now points at nothing. Those are the hardest thing in search to acquire and the easiest to lose in a bulk delete, and nobody notices because nobody audits inbound links to a product they stopped selling.
The internal structure. Collection pages, guides, blog posts and related product blocks all point at a page that no longer exists, so the store fills up with broken internal links, which degrade the experience and waste the crawling attention spent on them.
And the accumulated content. Reviews, questions, images, description work. On an older product that is often the richest content in the store, and it is deleted along with the row in the database.
None of that shows up in this month's report. It shows up as a slow decline in non branded traffic that nobody can attribute, which is precisely why it goes unaddressed for years.
A discontinued product is a merchandising decision with a search bill attached.
Three retirement routes, and when each fits
There are only three sensible destinations for a retired product, and choosing between them takes about a minute per product once the rule is clear.
Keep the page. Appropriate when the product still gets meaningful search interest, when it is a well known item people research, or when the page holds substantial reviews and content. The page states plainly that it is no longer available, keeps its information, and points clearly to what replaced it. This is the right answer more often than stores expect.
Redirect to the successor. Appropriate when there is a direct replacement: the next version, the same thing in a new format, the same function from a different maker you now stock. The visitor arriving gets something genuinely equivalent, which is the test.
Redirect to the parent collection. Appropriate when there is no successor but the category still exists. The visitor lands somewhere relevant and can choose, which is far better than a homepage or an error page.
What is never right is deleting with no destination, or redirecting everything to the homepage because it is one bulk action. Both throw away the value you are trying to move.
The successor is usually the best destination
When a replacement exists, sending the address to it does two jobs at once.
The visitor gets the modern version of the thing they were looking for, which is what they would have chosen anyway if they had known. And the relevance that the old page accumulated is passed toward a page that is still selling, rather than being spread across a category or lost.
A few practical points. The destination should be genuinely equivalent, not merely in the same category, because a redirect to something unrelated is treated as a poor experience and often behaves as if it went nowhere. It is better to redirect the old boot to the new boot than to redirect it to footwear.
If the old product is well known by name, consider keeping the page instead and linking prominently to the successor. That way you keep answering searches for the old name, which a redirect stops you doing, while still moving buyers forward.
And on the successor page, mention the lineage. A line saying this replaces the previous model, with what changed, catches the search for the old name and reassures the buyer who owned the old one.
Retiring a seasonal line, worked through
An invented example, to show the sequence.
Say a store retires a summer range of twelve products at the end of the season, four of which have direct equivalents in the new range, five of which are simply not returning, and three of which are the same product in a new colourway.
The three colourway changes are the easiest: keep the addresses if the platform allows a variant swap, or redirect each to its new colour, since the buyer wants the same thing.
The four with equivalents get redirected to their successors, with a line on the destination page acknowledging what it replaced.
The five with nothing to replace them get a decision each. Any with meaningful search interest or accumulated reviews keeps its page, marked unavailable, pointing to the collection. The rest redirect to the parent collection.
The whole exercise is perhaps an hour for twelve products, and it is done at the moment the decision is made rather than six months later when nobody remembers which product replaced which. That timing is most of the difference between a retirement plan and an archaeology project.
Keep a register of what you retired
The single most useful artefact here is a simple list, maintained as you go.
One row per retired product: the old address, the date, the reason, the destination chosen, and the successor if there is one. A spreadsheet is fine.
It pays for itself in several ways. When a product comes back, and they do come back, you know exactly which address it had and can restore it rather than creating a near duplicate. When traffic drops in a category, you can check what was removed and when. When a platform migration happens, the register is the map for the redirect file, and migrations are where these decisions get lost at scale.
It also makes the pattern visible. A store that discovers it retired four hundred pages in two years usually changes how it launches products, because it can see the cost of a catalogue that turns over faster than its content can accumulate.
The register takes about thirty seconds per product at the time. Reconstructing it later takes days and is never complete.
The feed side, and building retirement into launch
Two loose ends worth closing.
The feed. A discontinued product should leave the feed, not sit in it being rejected. Stale entries produce errors that cloud the account's real problems, and a product advertised while unavailable wastes budget and produces a poor experience. If the address now redirects, the feed entry should be gone rather than pointing through a redirect.
And the launch process. The cheapest way to make retirement cheap is to decide it at the beginning. When a product is added, note what it replaces and what would replace it. Give related products a stable collection home rather than only a seasonal one. Write the durable parts of the content on the collection page, where they survive, rather than only on individual products, where they do not.
The underlying principle is the same as everywhere else in search. The pages that accumulate value are the ones that stay, so put your effort where it stays, and when something has to go, hand what it earned to something that is still here.