How to find the dead Amazon products quietly costing you commission.
Your broken-link checker says every link is fine. It is right — and it is measuring the wrong thing. The links that cost you money return a perfectly healthy 200 and sell nothing.
The short version
- A dead Amazon product is not a broken link. The page loads, returns 200, and looks completely normal. No link checker will ever flag it.
- There are six ways a product dies and only one of them produces a 404.
- The real signal is the absence of a buyable offer — via the API, an item with no offers returned cannot be bought, whatever the page looks like.
- Audit by traffic, not alphabetically. Your top 20 pages are almost all of your revenue; start there and you find most of the loss in an hour.
- This is the biggest invisible leak on any site over a year old, and it compounds silently because nothing ever alerts you.
Here is the situation almost every established affiliate site is in and almost none of them know about.
You wrote a review two years ago. It ranks, it gets traffic, readers still click the button. But the product was discontinued eleven months ago. Amazon still serves the page — it shows the old listing, maybe a “currently unavailable” notice, maybe a third-party seller at triple the price. Your reader lands, sees they cannot buy it, and leaves.
Your link checker reports no problems, because the URL resolves. Your analytics show clicks, because they clicked. Your Amazon report shows nothing, because nothing sold — and nothing is exactly what a missing sale looks like in a report. There is no alert, anywhere, ever.
Why your broken-link checker cannot help
The six ways an Amazon product dies
Knowing which one you are looking at determines the fix, so it is worth being precise.
| State | What the reader sees | HTTP | Fix |
|---|---|---|---|
| Discontinued | “Currently unavailable. We don’t know when or if this item will be back in stock.” | 200 | Replace with the successor product |
| Out of stock | Listing intact, no Buy Box, no add-to-cart | 200 | Wait, or swap temporarily |
| No Buy Box winner | Only third-party sellers, often at absurd prices | 200 | Replace — these convert terribly |
| Variation removed | A different size or colour than the one you reviewed | 200 | Repoint to the surviving variation |
| Merged or replaced ASIN | Redirects to a newer model | 301 → 200 | Update the ASIN and the review |
| Listing removed | “Sorry, we couldn’t find that page” | 404 | The only one your link checker catches |
The third row is the one that costs the most. A product with no Buy Box still looks alive — there is a price, there are sellers, the page is full of reviews. But the price is often two or three times the original, and conversion collapses. It reads as a working link in every audit and it earns almost nothing. If you check only one state, check this one.
Three ways to find them
The manual audit
Free · about an hour · finds most of the moneyIf your site is under about fifty posts, this is genuinely the right answer and you do not need a tool.
Do it in traffic order. This is the part people get wrong — they start at the oldest post or work alphabetically and give up halfway. Affiliate revenue is heavily concentrated: your top twenty pages are usually most of your income, so checking those twenty finds the large majority of the loss in the first hour.
- Open Search Console or your analytics and sort landing pages by clicks, last 12 months.
- Take the top 20. Open each one.
- For every Amazon product on the page, open it in a private window — you want to see what a stranger sees, not what your cookies show you.
- Ask one question per product: can I add this to a cart right now, at a sane price, sold by Amazon? If no, note the page and the product.
- Repeat quarterly. Put it in a calendar or it will not happen.
Use a private window, always. Amazon personalises heavily. A logged-in session can show you a regional listing, a cached price, or a variation that a first-time visitor never sees. Auditing while logged in produces false clean results.
Extract every ASIN, then check them in bulk
Free · better for 50+ postsAbove fifty posts, the manual pass stops being practical. The middle option is to get a complete list of what you are actually linking to — which most publishers have never seen.
Amazon product URLs contain a ten-character ASIN in a predictable position:
amazon.com/dp/B08N5WRWNW/ · amazon.com/gp/product/B08N5WRWNW
Export your posts, pull every Amazon URL, and extract the ASINs. You will typically find three things immediately, before checking a single product:
- Duplicates — the same product linked from fifteen posts. One dead product then costs you fifteen pages at once, which is why this list is worth having.
- Untagged links — links with no
tag=, earning nothing regardless of whether the product is alive. See why Amazon affiliate links stop tracking. - Wrong-marketplace links — a UK link on a US-tagged site.
Then check the unique ASINs rather than every link. A site with 300 Amazon links usually has 80 to 120 unique products, which is a much smaller job.
Ask the API, which is the only reliable signal
Needs API access · the accurate methodReading Amazon’s HTML to determine availability is both unreliable and prohibited — the licence excludes “data mining, robots, or similar data gathering and extraction tools”. The supported route is the API.
The signal is simpler than people expect: look for the absence of a buyable offer. Request the item and examine what comes back with it. An item returned with no offers attached is not currently purchasable, whatever its product page looks like to a human. That single check catches discontinued items, out-of-stock items and no-Buy-Box items at once — the three states that account for nearly all of the loss.
Never act on one call. Throttling, transient errors and regional gaps all produce a response that looks identical to “this product is dead”. Marking a healthy product as discontinued and pulling it from a ranking page is a worse outcome than the problem you were fixing. Check twice, with a gap between, and treat a single failed lookup as unknown rather than dead.
If your API calls are failing outright rather than returning empty offers, that is a different problem — see Amazon API access denied.
What to do with what you find
Finding them is half the job and the half people stop at. The fix depends on the state.
Discontinued → replace and rewrite
Find the successor model, update the ASIN, and rewrite the section so it describes the new product honestly. Swapping the link but leaving a review of the old model is worse than the dead link — it misleads the reader.
Out of stock → wait or swap
Genuinely temporary shortages resolve. If it has been weeks on a high-traffic page, swap in an alternative and note that you did.
No Buy Box → replace now
These convert worst of all. Do not wait for the Buy Box to return.
Whole roundup obsolete → rewrite
If three of five products in a “best of” post are dead, the post is stale rather than broken. Rewriting it usually improves rankings as well as revenue.
Update, do not delete. A page that ranks is an asset even when its products are dead — the ranking took months to earn and would take months to earn again. Keep the URL, keep the publish date, replace the products, and refresh the content around them. Deleting a ranking page to “clean up” throws away the only hard part.
Do it in the right order
Sort by traffic and fix downward. A dead product on a page with 4,000 monthly visits is worth more than fifty dead products on pages nobody reads. Most sites recover the majority of the lost revenue in the first afternoon and then find diminishing returns — which is fine, and is a good reason to stop rather than to grind through everything.
Making it not happen again
A one-off audit fixes today and guarantees nothing about next year, because products keep dying at a steady rate whether or not you are looking.
- Schedule it. Quarterly, in a calendar, treated as maintenance rather than a project.
- Link fewer, better products. A page recommending three researched products has a third the failure surface of one listing ten.
- Prefer products with a long life. Tools, cookware and furniture stay listed for years. Consumer electronics turn over annually, and phone accessories faster still.
- Watch for the silent drop. A page whose earnings fall while its traffic holds is the signature of a dead product. That divergence is the single most useful thing to monitor, and it is easy to spot once you know to look.
If you automate one thing on your site, automate this. Not because it is difficult, but because it is invisible and unurgent — exactly the combination that means it never gets done. A scheduled check that emails you “four products on your three highest-traffic pages are no longer buyable” is worth more than any amount of new content, because it protects revenue you have already earned.
Part of The Amazon Affiliate Site Playbook — the whole job in six stages, from choosing a niche to keeping a mature site earning.
Have this checked for you, on a schedule
The Product Health Monitor in Ama Affiliate Pro scans every Amazon product on your site, flags the ones that are discontinued or unbuyable, and uses a two-pass check so a throttled API call is never mistaken for a dead product. Per-product links, CSV export, and scheduled re-checks.
Start a 14-day trialCommon questions
Why doesn’t my broken link checker find dead Amazon products?
Because the link is not broken. Amazon serves the product page normally with a 200 response even when the item is discontinued, out of stock or has no Buy Box. Link checkers test whether a URL resolves, and it does. Only one of the six ways a product dies produces a 404.
How often should I audit for dead products?
Quarterly for most sites. Monthly if you are in a fast-turnover category like consumer electronics or phone accessories. If you have never done it and your site is more than a year old, do one now — the first audit almost always finds more than you expect.
Can I check availability by scraping Amazon product pages?
No, on two grounds. Amazon’s licence excludes “data mining, robots, or similar data gathering and extraction tools”, so it puts your Associates account at risk. It is also unreliable — Amazon’s pages vary by region, session and test group, so the same URL can look available to one request and not another.
Should I delete posts where the products are all discontinued?
Almost never. A ranking page is an asset that took months to earn. Keep the URL, replace the products, and refresh the content around them. Deleting it throws away the ranking and gains you nothing.
What is the fastest way to find the ones that matter?
Sort your landing pages by clicks over the last 12 months and check the top twenty. Affiliate revenue is heavily concentrated, so those twenty pages usually contain the large majority of the recoverable loss, and you can work through them in about an hour.
How do I spot a dead product without checking every link?
Watch for pages where earnings fall while traffic holds steady. That divergence is the signature of a product that has stopped being buyable, and it is visible in your own analytics without touching Amazon at all.
A product came back in stock after I replaced it. Did I waste the effort?
No — an out-of-stock product earns nothing while it is out of stock, and high-traffic pages should not sit idle waiting. This is why the two-pass check matters though: you want to be confident a product is genuinely unavailable rather than reacting to one failed API call.
Do dead products hurt my SEO as well as my revenue?
Indirectly, and it can matter. Readers who land on a recommendation they cannot buy return to the search results quickly, and a page full of unavailable products is a poor experience by any measure. The revenue loss is the immediate cost; the ranking risk is the compounding one.