899 Backlinks to Nowhere: Finding SEO Gold in Dead Websites

👁 10 views

I found 899 backlinks pointing to pages that no longer exist. Not on some abandoned domain from 2005—on a live business website. This happens more than you’d think, and it’s one of the most overlooked opportunities in SEO.

The Setup: A Website That Forgot Its Past

While doing competitor research today, I pulled historical data from the Wayback Machine for a site in a niche B2B industry. What I found:

• 2,645 unique URLs archived over the years
• 1 page currently live on the actual website
• 899 backlinks still pointing to content that returns 404

Someone rebuilt this site and nuked 99.96% of their content. The backlinks didn’t get the memo.

Why This Matters

Backlinks are still the currency of SEO. When a page disappears but the links remain, that link equity doesn’t transfer—it just… evaporates. Those 899 links? Worthless to the site owner now.

But here’s the opportunity:

• If you’re the site owner: You can reclaim that equity with redirects or by recreating the content
• If you’re a competitor: You can reach out to those linking sites with better, actually-existing content
• If you’re an SEO: You just found a goldmine of broken link building targets

How I Did This (With MCP)

Here’s where it gets interesting. I didn’t manually crawl the Wayback Machine or write one-off scripts. I used an MCP server for the Wayback Machine—which means my AI assistant could query historical data conversationally.

The workflow looked like this:

Me: “Pull all archived URLs for example.com from Wayback Machine”

Agent: Queries the CDX API, returns 2,645 unique URLs sorted by date.

Me: “Check which ones are currently 404”

Agent: Tests each URL, flags the dead ones.

Me: “Get backlinks pointing to those dead URLs”

Agent: Queries backlink API, finds 899 links from 58 domains pointing to pages that no longer exist.

Total time: about 20 minutes, including uploading everything to a spreadsheet for analysis.

The Technical Bits (If You Want Them)

Under the hood, the Wayback MCP server wraps the Internet Archive’s CDX API:

# Get all archived URLs for a domain
wayback urls "example.com" --text-only

# Check snapshots for a specific page
wayback snapshots "https://example.com/old-page"

# Get content from a historical snapshot
wayback content "https://example.com/old-page" --timestamp 20230101

You can also hit the API directly:

curl "https://web.archive.org/cdx/search/cdx?url=example.com/*&output=json&fl=timestamp,original&filter=mimetype:text/html&collapse=urlkey"

This returns deduplicated URLs with timestamps. Combine this with any backlink API (Ahrefs, DataForSEO, Moz) to find links pointing to dead pages.

What I Learned Today

The site I analyzed competes in a space where the top players get 15,000-30,000 monthly organic visits. This site? 199 visits. They’re sitting on the skeleton of what was once a content-rich resource, and they apparently have no idea.

Or maybe they do, and rebuilding isn’t a priority. Either way, those 899 backlinks are up for grabs.

The Bigger Picture

Every industry has sites like this—companies that redesigned without redirects, migrated platforms badly, or just abandoned content strategies. The Wayback Machine remembers everything.

What makes this powerful with MCP: the research that used to require multiple tools, browser tabs, and manual data wrangling now happens in a single conversation. I asked questions, the agent did the work, and I got a spreadsheet with 2,645 historical URLs and 899 broken backlinks ready to act on.

That’s the point of connecting AI to real tools. Not generating content—doing actual research.

---

This is what I actually did today. Real research, real numbers, real opportunity. The wayback-mcp-server is open source if you want to try it yourself.