Link building mistakes: ten errors the evidence points to

The most expensive link building mistakes are buying from the low-quality pool, paying by Domain Rating, forcing exact-match anchors and never checking that links are indexed and still live. Each mistake below is tied to a study, a survey or Google's own wording.

In short

  • BuzzStream's June 2026 study of roughly 257,000 guest post sellers rated 96.2% as low quality and found 39.6% had no organic traffic.
  • Domain Rating correlated with rankings at only 0.131 in Ahrefs' January 2025 study of one million keywords, against 0.255 for referring domains.
  • Adsy's analysis of 37,542 sites, using 2024 data, found an average asking price of $406 against an average purchase price of $59 at DR 41 to 50.
  • Google's John Mueller said in September 2026 that the disavow tool does have effects, which includes harm when good links are disavowed.
  • A Digitaloft compilation updated in July 2026 reports that 68% of businesses lack a documented link building strategy.

Most link building money is lost quietly. The link goes live, the invoice is paid, and nothing happens, because the site was weak, the page was never indexed or Google ignored the link. Penalties are the rarer outcome. That is worth holding on to, because it changes what a mistake looks like. The costly errors in this field mostly do not announce themselves with a warning in Search Console. They show up as a budget that was spent and a ranking that did not move, and by then it is hard to tell which of several things went wrong. The ten mistakes below are ordered roughly by how much they cost, and each is tied to evidence.

Buying from the junk pool and paying by Domain Rating

The first mistake is buying from the junk pool, and it comes first because the supply of paid placements is mostly poor. BuzzStream’s June 2026 study of roughly 257,000 sites that sell guest posts rated 96.2% as low quality. Only 1.37% met a basic bar of DR 65 or more with 10,000 monthly organic visits, and 39.6% had no organic traffic at all. Its August 2026 study of link insertions rated 97.5% of offers as low quality. On those figures a buyer choosing without checks is far more likely to land on a poor site than a good one, simply because poor sites make up nearly all of what is for sale.

So an unvetted order lands in the weak pool by default. Marketplaces are tools for paying, not for judging quality. Run the checks in how to vet a publisher before any payment, whichever of the link marketplaces you use. The point about marketplaces deserves emphasis. A catalogue with filters and a checkout makes buying feel like a considered choice, but the platform’s job is to handle the transaction, and the judgement about whether a site has real readers stays with the buyer.

The second mistake is paying by Domain Rating, which can look like a way of avoiding the first and is not. DR is a third-party estimate, and it can be inflated with redirects and link networks. In Ahrefs’ study of one million keywords, published in January 2025, DR correlated with rankings at 0.131. Referring domains scored 0.255. So the score this mistake relies on has about half the relationship with rankings that a plain count of linking websites has, and, as noted, it can be inflated.

Price data points the same way. BuzzStream’s August 2026 study found that list prices for link insertions rise steadily with a site’s organic traffic and unevenly with DR. Sellers know buyers shop by DR, so that is the number they dress up. Domain Authority vs Domain Rating explains what each score measures. The practical lesson is to treat DR as a first filter at most, and to look at organic traffic, which the same study found tracks price more steadily, before deciding what a placement is worth.

The third mistake is forcing exact-match anchors. Over-optimised anchors are widely described as the classic trigger for Google’s Penguin update in 2012. Google publishes no safe ratio. An Ahrefs study of 384,614 pages found that most top-ranking pages have a low share of exact-match anchors, and that the share shows only a weak link with ranking. The data therefore offers little reward for pushing keyword anchors, and the history of Penguin shows where the risk lies, which makes this a mistake with a small upside and a well-known downside.

Google’s leaked documentation from 2024 contains fields that count spam-phrase anchors and measure spikes in them. Those fields are documented, and their weight is unknown. The safer method is to compare your anchor text profile with the pages that rank for the same term, and keep exact-match anchors for internal links. That comparison replaces a ratio nobody can source with a range you can observe, and it keeps the keyword wording on your own site, where you control every link and nobody expects the anchors to vary.

The fourth mistake is counting links instead of referring domains. A report that says “400 new backlinks” may describe four websites. Ahrefs’ 2025 study found referring domains correlate with rankings slightly better than raw backlinks, 0.255 against 0.248. Report unique referring domains, leave out syndicated copies and show nofollow links separately. The gap between the two correlations is small, so the stronger argument is about honesty in reporting: a count of websites cannot be inflated by one sitewide link or by a press release copied across a network, and a count of backlinks can.

The fifth is pointing every link at the wrong pages, and two opposite errors are common. The first is sending every paid link to a commercial page with a commercial anchor. Pages with heavy exact-match anchors are the ones Penguin-style devaluation tends to hit. The second is earning links only to blog posts and never passing the value on, which our industry research lists as a typical SaaS mistake. Law firms tend to make a third: every link to the homepage. In each case the links exist and the pages that earn money see little benefit, either because the links were discounted or because nothing connects them to those pages.

Ahrefs’ 2025 data shows the link correlation is weaker for commercial and transactional searches. The standard model is to earn links to assets people cite, then link internally from those assets to the pages that sell. Which pages need links covers the split. The model works with the grain of how people link. Writers cite guides, data and tools far more readily than they cite a sales page, so the asset collects the links and the internal links carry the value to where it is needed.

The sixth mistake is never checking that links are indexed and still live. A link on a page Google has not indexed passes nothing, and links disappear. An Ahrefs study is reported to have found that about two thirds of links to sites over nine years are now dead. We have not rechecked that figure. Even with that caution, the direction is not in doubt: a link profile is something that decays, and a link paid for last year cannot be assumed to exist today.

A working routine is to check each new link for indexing at day 7, day 14 and day 30, then monitor monthly. Start with what you already lost: link reclamation recovers earned links for the cost of a redirect. The routine costs very little next to the price of the links it protects, and an early check gives you dated evidence to put to a vendor if a placement was never indexed or has gone.

The seventh mistake is trusting the list price. Adsy analysed 37,542 sites using its 2024 data and compared asking prices with what buyers paid. At DR 41 to 50 the average asking price was $406 and the average purchase price was $59. At DR 71 to 80 it was $941 against $199. In both bands the price paid was a small fraction of the price asked, which tells you the asking price is an opening position and that buyers who pay it are paying several times what others paid for comparable sites.

Vendors add their own layer. BuzzStream’s 2026 data puts a guest post at $295 bought direct and $461 through a vendor, a markup of about 56%. If you accept the first quote, you are likely to overpay. A markup is not improper in itself, since a vendor may be doing the finding and the negotiating, but a buyer should know that it is there and decide whether that service is worth the difference.

Working without a plan and disavowing by reflex

The eighth mistake is working without a written plan. A Digitaloft compilation updated in July 2026 reports that 68% of businesses lack a documented link building strategy and that 70% are dissatisfied with their link building partners. The original samples behind those figures were not inspected, so read them as indicative. The two figures appear side by side in that compilation, and while nothing in it shows that one causes the other, it is hard to hold a partner to a standard that was never written down.

Without a plan there is no target, no quality bar and nothing to hold a vendor to. Habit fills the gap. In uSERP’s September 2026 survey of 116 senior SEOs, guest posting was named the tactic most overused relative to its results, by 40.5% of respondents. That is what habit looks like in practice: a tactic keeps being ordered because it was ordered last quarter, and nobody is asked to show that it is still the best use of the money.

The ninth mistake is disavowing by reflex. Google’s guidance is reported to limit the disavow tool to two cases: a manual action for unnatural links, or a large number of artificial links you created that are likely to cause one. John Mueller has called “toxic links” a concept made up by SEO tools. In September 2026 he said the disavow tool does have effects and that recovery can take many months. Taken together, those statements describe a tool meant for a narrow situation, and one whose use has consequences that take a long time to work through.

The tool works in both directions. Disavow good links and rankings can fall. Our lesson on backlink audits and toxic links explains when an audit is warranted. The reflex is to act on a tool’s warning about “toxic” links, and the risk is that a site owner who does so removes links that were helping, in order to solve a problem that Google’s own guidance does not recognise outside those two cases.

Using disposable tactics on a brand

The tenth mistake is using disposable tactics on a brand. High-risk tactics are built for sites the owner can afford to lose. With its March 2026 spam update, Google repeated that once the effect of spammy links is removed, the ranking benefit is lost and clean-up does not restore it. That statement removes the comfortable idea that a risky campaign can be tried and then tidied away if it goes wrong, because whatever the links gained goes when the links stop counting.

Research on gambling markets lists the related errors: treating a licensed brand domain as disposable, buying the same marketplace sites every competitor uses, and placing English anchors on foreign-language sites for a local-language market. The iGaming hub describes them. A rule of thumb covers all of them: never use a tactic on an asset whose loss you cannot afford. An operator with many throwaway domains and a business with one domain that carries its name are playing different games, and methods copied from the first do not suit the second.

Where to go next

Nearly every mistake on this list comes down to a decision that was never made explicitly, whether about quality, price, anchors, reporting or risk. Turn the list into a process with a link building strategy plan, which is where those decisions get written down before the money is spent. Unfamiliar terms are defined in the glossary.

Common questions

What is the most common link building mistake?

Buying placements without vetting the site. BuzzStream's 2026 data rated 96.2% of guest post sellers as low quality, so an unvetted purchase is far more likely to land in that pool than outside it.

Can bad backlinks hurt my site?

Random spam links are mostly ignored, according to repeated Google statements. Links you built or bought in manipulative patterns can lead to a manual action, and the usual cost of a bad link is the money spent on it.

Is using too much exact-match anchor text a mistake?

Yes. Google publishes no safe ratio, and an Ahrefs study of 384,614 pages found most top-ranking pages have a low share of exact-match anchors. Benchmark against the pages that rank for your term.

Should I disavow low-quality links every month?

No. Google's guidance is reported to limit the disavow tool to manual actions and to links you created that are likely to cause one. John Mueller said in September 2026 that the tool has real effects, so disavowing good links can lower rankings.

Is building links only to the homepage a mistake?

It limits you. Our industry research lists sending all links to the homepage as a common law firm mistake, and building links only to the blog without passing value to commercial pages as a common SaaS mistake.

Vendors to look at

  • BazoomEditor's pick

    Sponsored content and link marketplace, managed service

  • MotherlinkEditor's pick

    Backlink services, guest posts, niche edits, full SEO