We Ran a 90-Day Crazy Backlink Experiment: Parasites, Redirects, and Anchor Overkill in 2026
Photo by Photo by Agence Olloweb on Unsplash on Unsplash
The setup: three sites, three bad ideas, one spreadsheet
In January 2026, we set up a 90-day test to answer a question every aggressive SEO team argues about: which controversial link tactic actually moves rankings, and which one just makes a lot of noise? We used three separate US-focused sites in different niches. Each got one tactic, a baseline audit, weekly rank tracking, and a kill switch we agreed on before launch. No client sites. No fake brands pretending to be real businesses. Just controlled tests with clear limits.
This is not a universal verdict. It is a lab report. Ninety days is enough to see patterns, not enough to declare winners forever.
The three sites and what we did
- Site A - local home services, parasite pages. We published 12 thin-but-useful pages on high-authority platforms, each targeting a city plus service keyword. We pointed exact-match anchors from those pages to one money page and added a few internal links. The goal was to see whether parasite rankings would lift the main site or just cannibalize attention.
- Site B - ecommerce, expired-domain redirects. We bought six expired domains with relevant-ish histories and redirected them to category pages. We audited each domain for spam, checked archive history, and avoided anything with obvious casino or pharma links. The goal was to test whether redirected authority produced durable ranking movement.
- Site C - B2B SaaS, anchor-text overkill. We built 40 links over 60 days through guest posts and paid niche edits, with roughly 80% exact-match anchors. That is not a normal profile. That was the point. We wanted to see how much anchor manipulation the site could absorb before something looked off.
Baseline metrics
| Site | Niche | Baseline organic clicks/mo | Keywords in top 10 | Referring domains | Notes |
|---|---|---|---|---|---|
| A | Home services | 1,900 | 34 | 61 | Local pack dependence |
| B | Ecommerce | 8,400 | 112 | 184 | Seasonal demand |
| C | B2B SaaS | 3,200 | 57 | 96 | Long sales cycle |
We tracked rankings weekly, organic clicks daily, and new referring domains. We also logged indexation status, manual actions, and any weird volatility that did not match the niche. We set kill switches: any manual action, a 30% drop in organic clicks for two consecutive weeks, or a clear spam footprint warning from our audit tools.
Week-by-week: what actually happened
Weeks 1-2. Site A's parasite pages indexed fast. Some ranked on page two for city-plus-service terms. The money page did not move. Site B saw a small lift in category rankings, maybe 3 to 7 positions on a handful of terms. Site C's new links started indexing, but rankings were flat.
Weeks 3-5. Site A's parasite pages picked up a little referral traffic, but the main site stayed flat. Site B moved more, especially on long-tail category terms. Site C began to see movement on exact-match keywords, which was expected and also a little suspicious.
Weeks 6-8. Site A hit a ceiling. The parasite pages were ranking for the parasite URL, not the brand. Site B got volatile. Some pages held gains, others bounced back to baseline. Site C saw its biggest ranking jump, then a strange dip on branded queries that made no sense.
Weeks 9-12. Site A lost several parasite rankings after platform-level cleanups. Site B settled into a mixed pattern: two redirected domains looked useful, four looked like noise. Site C's rankings stayed higher than baseline but the link profile looked unnatural, and we started seeing anchor-level warnings in third-party tools.
What moved
The expired-domain redirects produced the clearest short-term movement on Site B. Two of the six domains had genuinely relevant link profiles and pushed category pages up for long-tail terms. The effect was not dramatic, but it was measurable. The parasite pages on Site A ranked themselves, not the money site. That is the trap. You can win the parasite SERP and still lose the actual business goal. Anchor overkill on Site C moved exact-match keywords, but the pattern was ugly and the gains were fragile.
What got ignored
The parasite pages did not transfer authority the way the tactic's fans claim. The money page barely moved. The four weak expired domains did nothing except add redirect clutter. And most of the anchor-overkill links passed no meaningful value beyond the first few weeks. Google did not need to penalize anything to make the tactic a bad investment. It just ignored the noise.
What triggered suspicious signals
Site C was the obvious one. An 80% exact-match anchor profile on a site with a previously natural link mix is a fingerprint. We saw third-party tools flag the profile, and we saw ranking volatility that did not match content changes or seasonality. Site B's redirects also created a pattern: six redirects in 30 days, all to category pages, all with similar anchor text. That is not how organic links arrive. Site A's parasite pages were less about link risk and more about platform risk. When the host platform cleans up thin content, your rankings disappear with it.
Which tactics survived enough to test again
Expired-domain redirects survived, but only in a much smaller, slower form. We would test one or two carefully audited domains per quarter, not six in a month. Parasite pages did not survive as a link tactic. They can work as a distribution tactic if the platform page itself is the destination, but they are a poor way to lift a money site. Anchor-text overkill did not survive. The rankings moved, but the risk profile was not worth it. We would rather build a natural anchor mix and spend the budget on original data.
The kill switches we used
- Manual action in Search Console: stop and audit immediately.
- Organic clicks down 30% for two straight weeks: pause all new links.
- Third-party spam score above our threshold: remove or disavow after review.
- Host platform cleanup or deindexation: stop publishing parasite pages.
- Redirect domain with unexpected toxic links: remove the redirect.
What we would do differently
We would baseline for 60 days instead of 30. We would isolate one variable per site instead of bundling tactics. And we would track branded search volume, because Site C's weird branded dip was the signal we almost missed. We would also spend less time arguing about DR and more time looking at whether the page had real users.
The bottom line
Aggressive link tactics are not binary. They exist on a spectrum from 'boring but effective' to 'expensive noise' to 'please call your lawyer.' In our 90-day test, the only tactic that earned a second round was a slower, cleaner version of expired-domain redirects. Parasite pages ranked themselves. Anchor overkill moved keywords and wrecked the profile. If you are going to run experiments like this, set a kill switch, document the baseline, and be honest about what the data actually says. CrazyBacklink is fine with aggressive. We are not fine with pretending a footprint is a strategy.