Only 3.7% of Paid Backlinks Survive in Google’s Index After a Year
What the Study Measured and Why Indexing Matters
SpeedyIndex set out to answer a simple question that many link buyers never test: after paying for a backlink, does Google actually see it? The company bought 15,389 links itself from freelance marketplaces, Fiverr and Kwork, spreading purchases across different sellers and link tiers so the results would reflect the whole supply chain rather than a hand-picked sample. Each link was checked in two passes, an initial crawl in August 2026 and an index verification in late October 2026.
Two terms matter here. A backlink is a hyperlink on another website pointing to yours, and search engines treat these links as votes of confidence. Indexing is the step where Google stores a page in its database and can show it in results. A link on a page Google has never indexed passes no value, however expensive it was. Many failures in the study came before indexing even began. Dead links are domains that no longer resolve, pages returning 404 or 410 errors, and parked domains or redirects that send visitors to a homepage. Other failures came from blocking signals such as noindex tags, robots.txt disallows, or canonical tags pointing elsewhere.
The study also used a category called unverified, applied to URLs hidden behind Cloudflare challenges that automated checks could not pass. Readers evaluating any audit should look closely at how each category is defined, because a page that is technically alive but unverifiable is not the same as a page that is confirmed indexed.
Where Paid Links Fail and What Surprised the Researchers
The headline figure is that 562 of 15,389 links, about 3.7 percent, were live, open and indexed a year later. The failure pattern was more revealing than the total. Dead pages were the largest single category at roughly a third of all links. Only 21 percent of links returned a clean HTTP 200 response with no indexing block, and of those technically clean pages, just 8.6 percent were actually in Google’s index. The gap between looking healthy and being indexed is the central lesson.
Domain-level results were even harsher. Of 5,088 linking domains, 93.1 percent had no indexed link at all, which suggests that buying from many sellers does not diversify risk if the underlying placements share the same weaknesses.
Two findings challenge common assumptions. First, a high Domain Rating does not guarantee indexability. Many large platforms keep their homepages open while applying noindex directives to user profiles, posts and shared files, which are often the pages sellers offer as placements. SpeedyIndex listed hundreds of such closed page types across hundreds of sites. Second, link age changed the failure mode. Among links older than 12 months, 18.5 percent no longer resolved at the DNS level, while in the six-to-twelve-month group, deleted pages returning 404 errors were the main cause. Kwork links were live more often than Fiverr links, but indexing rates were nearly identical, around 3.5 percent for both.
Practical Steps for Buyers, Agencies and Tool Users
The study is most useful as a checklist. Before paying for any placement, confirm that the exact URL returns a 200 status, that it has no noindex meta tag or X-Robots-Tag header, that its canonical tag points to itself, and that it does not redirect to a different page. Checking the homepage alone tells you very little. Test the specific page the seller offers, ideally from a logged-out browser and a crawler-style tool.
After publication, verify indexing rather than trusting a screenshot. Search Console’s URL Inspection tool and a site: query are simple starting points. Re-check at six and twelve months, since the study shows that links decay over time through deleted pages and expired domains. Ask sellers for written refund terms tied to indexing, not just to delivery.
This matters more as automation grows. An auto backlink tool, an AI content generator or an AI Post Images Generator can produce large volumes of content quickly, but volume does not solve indexing. Automated campaigns should include monitoring for dead pages and blocked placements, and results should be judged by indexed links, not link counts.
Finally, keep perspective. SpeedyIndex is a vendor with its own commercial interest in showing that paid links fail, and its sample reflects its own purchases on two marketplaces. The data is still a useful warning: verify before you pay, and measure what Google actually indexes.
