How Important Are Backlinks for SEO? Results From 6 Studies

I analyzed seven SEO backlink studies covering more than 11.8 million Google search results, 350,000 backlink placements, 1,000 domains, and multiple real-world SEO experiments to determine how backlinks influence rankings in Google Search and AI search.

Across all seven importance of backlinks for SEO studies, several consistent patterns emerged:

• High-quality, relevant backlinks correlate with stronger search visibility.
• Referring domain diversity has a stronger relationship with rankings than simply increasing backlink count.
• Authoritative, editorial backlinks produce the largest SEO gains.
• Cheap and spammy backlinks provide little value and may create SEO problems.
• High-authority and natural backlink profiles are associated with greater visibility in AI-powered search engines.

This article summarizes the findings from each study, compares their methodologies, and identifies the backlink factors that consistently correlate with higher rankings in Google Search, Google AI Overviews, ChatGPT, Gemini, and Perplexity.

TL:DR

• Google continues to use PageRank as part of its core ranking systems, although its influence has decreased over time.
• Gary Illyes stated that backlinks are important but are no longer among Google’s top three ranking signals.
• Google removed the word “important” when describing links in its spam documentation during the March 2024 core update.
• Google continues to discourage SEOs from focusing on backlink quantity and instead recommends improving overall website quality.
• Ahrefs found that top-ranking pages gain between 5% and 14% more followed backlinks every month.
• Backlinko found that the #1 Google result has an average of 3.8x more backlinks than positions two through ten.
• Backlinko also reported that referring domains correlate more closely with rankings than backlink quantity.
• One authoritative backlink can outperform dozens of low-quality backlinks.
• Cheap backlinks from Fiverr and similar services produced no measurable ranking or traffic improvements in a real-world experiment.
• Toxic backlink profiles can suppress organic performance and should be included in every backlink audit.
• Backlinks continue to influence AI search visibility across ChatGPT, Google AI Overviews, Gemini, Claude, Grok, and Perplexity.

Study 1: Backlinko’s 11.8 Million SERP Study (2025)

Brian Dean published Backlinko’s 11.8 Million SERP Study in 2025 using backlink data from Ahrefs, content analysis from Clearscope, and page speed and engagement data from Alexa.

Backlinko analyzed over 11 million Google search results to identify which ranking factors appeared most frequently among first-page rankings. The dataset included backlinks, referring domains, Domain Rating (DR), URL Rating (UR), content length, page speed, title tags, H1 headings, Schema markup, HTML size, URL length, and user engagement metrics.

Key findings from Backlinko’s SERP Study

• The #1 Google result had an average of 3.8x more backlinks than pages ranking between positions 2 and 10.

• 95% of all pages had zero backlinks, so Brian Dean excluded those URLs from part of the backlink correlation analysis.

• Pages with backlinks from more referring domains are generally ranked higher than pages with links from fewer websites.
• Domain Rating (DR) correlated more closely with rankings than URL Rating (UR).
• The average first-page result contained 1,447 words, although content length alone showed little relationship with rankings.

• Schema markup, page speed, HTML size, title tags, and H1 keywords showed little or no relationship with ranking positions among first-page results.
• Websites with higher time on site generally ranked higher than websites with lower engagement.

What I Liked About Backlinko’s Backlink Study

Most people quote the 3.8x more backlinks statistic. I think the more valuable insight is that 95% of published pages never earned a single backlink. That statistic highlights how difficult it is to attract editorial links in the first place.

I also like that Brian Dean analyzed far more than backlinks. Testing multiple ranking signals in the same dataset gives readers useful context instead of pushing a single narrative.

The referring domain finding also deserves more attention. Backlink quantity is easy to inflate. Referring domains are much harder to manufacture because they require earning links from additional websites. If I had to monitor one backlink metric, it would be referring domains.

Limitations of Backlinko’s Search Engine Ranking Analysis

Backlinko identified correlations across millions of search results, but correlation does not establish cause and effect. The data cannot determine whether backlinks improved rankings or whether highly ranked pages naturally attracted more backlinks.

I also would have liked to see separate findings for informational, commercial, local, and branded searches. Different search intents often rely on different ranking signals, and combining them into one dataset may hide meaningful differences.

My Review of Backlinko’s SERP Data

Category	My Rating	Comments
Sample size	10/10	11.8 million search results provide an exceptionally large dataset.
Methodology	8/10	Well executed, but still a correlation study rather than a controlled experiment.
Data quality	9/10	Ahrefs, Clearscope, and Alexa provide reliable third-party datasets.
Practical value	10/10	The findings are relevant for almost every SEO campaign.
Risk of bias	8/10	Uses proprietary SEO metrics, but the conclusions are balanced and not overly promotional.
Overall score	9.0/10	One of the most useful backlink studies available, provided you understand the difference between correlation and causation.

Study 2: Search Atlas’s Page-Level Backlink Study (2025)

Manick Bhan, Founder and CEO of Search Atlas, published the SEO Impact of Backlink Placement in August 2025. The study analyzed 350,159 backlink placements across 13,002 landing pages to measure how backlinks affected SEO performance after they were placed. Search Atlas used Google Search Console data, regression models, Wilcoxon Rank-Sum tests, semantic relevance analysis, and XGBoost machine learning to evaluate backlink quality and ranking improvements.

Key findings from Search Atlas’s Page-Level Backlink Study

Search Atlas measured every page before and after backlink placement and reported statistically significant improvements across every SEO metric.

Metric	Before Backlinks	After Backlinks	Improvement
Ranking keywords	12	30	+149.9%
Google Search impressions	~20 million	~75 million	+275%
Organic clicks	—	—	+14%
Average Google position	32.3	28.8	+3.5 positions (10.8%)

The study also identified which backlink characteristics produced the strongest SEO improvements.

• High Domain Power (DP) backlinks consistently outperformed backlinks measured only by Domain Rating (DR) and Domain Authority (DA).
• One backlink from a high-authority website delivered stronger ranking improvements than multiple backlinks from lower-authority domains.
• Topically relevant backlinks generated larger SEO gains than backlinks from unrelated websites.
• Anchor text relevance significantly strengthened backlink performance.

• The semantic similarity between anchor text and target keywords increased from 0.415 to 0.703, representing a 69.6% improvement.
• The correlation between anchor text relevance and Google rankings improved from -0.75 to -0.86 after backlink placement.
• The coefficient of determination (R²) increased from 0.564 to 0.745, showing that anchor text explained a much larger share of ranking variation after backlinks were acquired.
• Domain Power (DP) became the strongest predictor of organic traffic, impressions, and ranking improvements, outperforming both Domain Rating (DR) and Domain Authority (DA).

What I Liked About Search Atlas’s Backlink Study

The biggest strength is the methodology.

Rather than comparing websites that already ranked well, Manick Bhan measured what happened after backlinks were placed. That makes the findings much easier to apply because they show how rankings changed over time instead of simply identifying patterns among successful websites.

The statistical analysis also deserves credit. The paper includes regression models, effect sizes, statistical significance testing, machine learning, and semantic analysis instead of relying on a handful of correlation charts. That level of detail is rare in publicly available SEO research.

The anchor text section was another highlight.

Most backlink studies focus almost entirely on authority metrics. Search Atlas demonstrated that context matters just as much as authority. A backlink from a trusted website becomes substantially more valuable when the anchor text closely matches the destination page’s topic and search intent.

Finally, I appreciated that the study evaluated multiple authority metrics instead of assuming one measurement tells the whole story. Comparing Domain Power, Domain Rating, and Domain Authority produced a more complete picture of how different authority models relate to ranking improvements.

Limitations of Search Atlas’s Page-Level Backlink Study

The biggest limitation is Domain Power (DP).

Search Atlas created Domain Power and concluded that it predicts SEO performance better than Domain Rating and Domain Authority. Because DP is proprietary, independent researchers cannot reproduce or validate those findings using the same metric.

I also wanted to see the results segmented by industry. Ecommerce stores, SaaS companies, publishers, affiliate websites, and local businesses compete under different search environments. Breaking the dataset into verticals would have shown whether the same backlink patterns remained consistent across different industries.

My Review of Search Atlas’s Page-Level Backlink Analysis

Category	Rating	My assessment
Sample size	9.5/10	350,159 backlink placements across 13,002 pages provide one of the largest page-level backlink datasets available.
Methodology	9.5/10	Before-and-after measurements, statistical testing, semantic analysis, regression models, and machine learning create a rigorous research framework.
Data quality	9/10	The study reports detailed metrics, effect sizes, and statistical significance instead of relying on simple observations.
Practical value	10/10	The findings translate directly into backlink acquisition strategies, anchor text optimization, and outreach campaigns.
Risk of bias	7.5/10	Search Atlas promotes its own proprietary authority metric throughout the paper.
Overall score	9.1/10	One of the most comprehensive backlink studies I reviewed. The methodology is excellent, the dataset is substantial, and the anchor text analysis fills a gap left by most backlink research.

Study 3: Pitchbox’s Backlinks and AI Search Study (2026)

Pitchbox published a study on Backlinks and AI Search in 2026 to determine whether backlinks still influence visibility in AI search.

The study evaluated 20+ B2B software companies across seven commercial-intent keywords and measured their rankings in ChatGPT, Claude, Gemini, and Grok. Pitchbox then compared each company’s AI rankings with traditional backlink metrics, including total backlinks, referring domains, Domain Rating (DR), and high-authority backlinks. Pearson correlation analysis was used to identify which backlink signals aligned most closely with AI visibility.

Key findings from Pitchbox’s Backlinks and AI Search Study

Pitchbox found that backlink signals continue to correlate with visibility across major AI search platforms.

Backlink Metric	Correlation with LLM Rankings
Referring domains	+0.62
Total backlinks	+0.55
DR 35+ backlinks	+0.52
DR 65+ backlinks	+0.43

The study reached several additional conclusions.

• Referring domain diversity produced the strongest relationship with AI rankings.
• Total backlink volume remained an important ranking signal, although it was less influential than referring domains.
• High-authority backlinks improved AI visibility, but increasing Domain Rating alone produced diminishing returns.
• AI platforms rewarded websites with diverse backlink profiles rather than backlinks concentrated from a small number of domains.
• ChatGPT and Claude showed a stronger relationship with backlink metrics than Gemini and Grok, suggesting each platform weighs authority signals differently.
• Backlinks explained a significant portion of AI visibility, but brand mentions, semantic relevance, and content freshness also contributed to rankings.
• The study concluded that existing SEO investments continue to benefit AI search visibility because backlink authority transfers into LLM recommendations.

What I Liked About Pitchbox’s AI Search Study

This is one of the first studies that attempts to quantify the relationship between backlinks and AI search instead of assuming traditional SEO rules automatically apply to LLMs.

I also liked that Pitchbox compared multiple AI platforms instead of treating AI search as a single product. ChatGPT, Claude, Gemini, and Grok clearly do not behave identically, and separating their results made the analysis more useful.

The referring domain finding was another highlight.

Most discussions around AI search focus on content, prompts, and citations. Pitchbox showed that backlink diversity still has the strongest relationship with AI visibility, reinforcing a principle that has existed since the early days of Google’s PageRank.

The correlation table also presents the results clearly. Referring domains, backlink volume, and authority metrics are ranked side by side, making it easy to understand which signals matter most.

Limitations of Pitchbox’s Backlinks Analysis

The dataset is considerably smaller than the Google ranking studies reviewed earlier.

Pitchbox analyzed seven keywords within the B2B software industry. That provides useful evidence, but it does not represent ecommerce, local SEO, healthcare, finance, publishers, or other competitive verticals.

The study also measures correlation, not causation. Strong backlink profiles correlate with better AI rankings, but the analysis does not isolate backlinks from other authority signals such as brand recognition, product popularity, or content quality.

My Review of Pitchbox’s Backlinks and AI Search Study

Category	Rating	My assessment
Sample size	7.5/10	Limited to seven commercial keywords and B2B software companies.
Methodology	8.5/10	Pearson correlation analysis across four major LLMs provides a solid analytical framework.
Data quality	8.5/10	Clear reporting of correlation coefficients and platform-specific observations.
Practical value	9.5/10	One of the few studies directly connecting backlink profiles with AI search visibility.
Risk of bias	8/10	Pitchbox benefits from proving backlinks remain valuable, although the study openly discusses its limitations.
Overall score	8.6/10	An important early contribution to AI SEO research. The dataset is relatively small, but the findings provide compelling evidence that backlink diversity continues to influence visibility across major LLM platforms.

Study 4: Semrush’s AI Search Backlink Study (2025)

Semrush published Do Backlinks Still Matter in AI Search? in October 2025 in collaboration with Kevin Indig from Growth Memo. The study analyzed 1,000 randomly selected domains from the Semrush AI Visibility Toolkit to determine whether backlinks still influence visibility across ChatGPT, ChatGPT Search, Gemini, Google AI Overviews, and Perplexity.

The researchers compared AI mentions and AI Share of Voice against backlink metrics, including total backlinks, referring domains, follow versus nofollow links, text versus image links, and Semrush’s Authority Score. Both Pearson and Spearman correlation analysis were used to identify linear and threshold relationships.

Key findings from Semrush’s AI Search Backlink Study

Finding	Result
Authority Score vs AI Share of Voice	Pearson 0.23, Spearman 0.36
Authority Score vs AI mentions	Pearson 0.65, Spearman 0.57
Follow links vs AI mentions	Pearson 0.334, Spearman 0.504
Nofollow links vs AI mentions	Pearson 0.340, Spearman 0.509
Image links vs AI mentions	Pearson 0.415, Spearman 0.538
Text links vs AI mentions	Pearson 0.334, Spearman 0.472

Semrush also reported several practical observations.

• Higher Authority Scores increased AI visibility only after websites crossed higher authority thresholds.

• Backlink quality produced a stronger relationship with AI visibility than backlink quantity.
• Nofollow backlinks performed almost identically to follow backlinks across every AI platform tested.

• Image backlinks produced stronger correlations with AI visibility than traditional text links.

• ChatGPT relied most heavily on backlink authority.
• ChatGPT Search and Perplexity placed less emphasis on backlink authority than the other AI platforms.
• Gemini and ChatGPT responded more positively to authoritative nofollow links.
• Google AI Overviews and Perplexity showed a stronger preference for follow links.
• Perplexity and ChatGPT Search demonstrated the strongest relationship between image backlinks and AI visibility.

What I Liked About Semrush’s AI Search Study

This study goes far beyond asking whether backlinks matter.

Semrush examined backlink quality, authority thresholds, follow and nofollow links, image backlinks, text backlinks, AI mentions, and AI Share of Voice within the same dataset. That produces a much clearer picture of how modern AI systems evaluate authority.

The findings around nofollow backlinks were particularly interesting. The difference between follow and nofollow links was almost nonexistent, suggesting AI platforms value trusted references regardless of the HTML attribute attached to the link.

I also liked the image backlink analysis.

Original charts, infographics, product screenshots, and data visualizations receive far less attention than written content, yet Semrush found that image backlinks produced stronger relationships with AI visibility than traditional text links. That creates a compelling case for investing more heavily in visual assets.

Kevin Indig’s explanation of authority thresholds was another highlight. The data shows that authority compounds over time. Small increases in backlink authority produce limited gains, while larger improvements appear after websites reach stronger authority levels.

Limitations of Semrush’s AI Search Study

The research focuses entirely on AI search visibility rather than traditional Google rankings.

Semrush also relies on its proprietary Authority Score, which cannot be directly compared with Google’s internal ranking systems.

Finally, the study measures correlation, not causation. It identifies strong relationships between backlink signals and AI visibility without proving that backlinks alone caused those results.

My Review of Semrush’s AI Search Study

Category	Rating	My assessment
Sample size	8.5/10	1,000 domains across five AI platforms provide a solid foundation for AI search research.
Methodology	9.5/10	Pearson and Spearman analysis reveal both linear and threshold relationships.
Data quality	9.5/10	Multiple backlink signals were evaluated instead of relying on a single authority metric.
Practical value	10/10	The findings directly influence backlink strategy, digital PR, and AI SEO.
Risk of bias	8/10	Authority Score is proprietary, although the methodology is transparent throughout the study.
Overall score	9.2/10	One of the most comprehensive studies on backlinks and AI visibility. The findings challenge long-held assumptions about nofollow links while introducing new evidence for image backlinks and authority thresholds.

Study 5: SEO Gets Toxic Backlinks Case Study (2024)

Matthew Mellinger, co-founder of SEO Gets and owner of Local SEO Partners, published Do Toxic Backlinks Impact Google Rankings? A Local SEO Case Study in 2024 to answer a question that has divided the SEO industry for years.

Can toxic backlinks hurt a website’s Google rankings?

Google has repeatedly stated that its algorithms ignore most spam backlinks instead of penalizing websites for them. Matthew tested that claim after taking over SEO for a client’s pool cleaning business, which inherited an aged domain containing hundreds of suspicious backlinks from the previous owner.

Key findings from SEO Gets Toxic Backlinks Case Study

Matthew documented the entire recovery process and recorded several notable observations.

• The client’s pool cleaning website showed almost no organic growth after months of local SEO, citation building, backlink acquisition, and Google Business Profile optimization.
• The client’s website could not rank #1 for its own brand name, while the company’s Facebook page ranked above the official website.
• Ahrefs uncovered hundreds of spam backlinks pointing to a deleted page from the previous version of the client’s website.
• Most toxic backlinks came from user-generated content (UGC) pages linking to a 404 page that no longer existed.

• Matthew submitted a Google Disavow file covering more than 200 referring domains.

• Google Search Console recorded a 4x increase in impressions within 24 hours of submitting the disavow file.

• Organic visibility improved, but ranking growth eventually stalled.
• Matthew migrated the client’s website to a new domain using a similar brand name.
• The client’s new website generated almost four times more impressions than the old domain’s highest point and 16 times more impressions than before the disavow file was submitted.

• After rebuilding authority through local citations, niche edit backlinks, and a press release, the client’s new website ranked in Google’s top 10 for multiple local service keywords.

What I Liked About SEO Gets Toxic Backlinks Case Study

Most backlink studies focus on acquiring links.

Matthew explored the opposite side of the discussion by investigating whether poor-quality backlinks can suppress organic performance. That makes this case study different from every other study I reviewed.

I also liked that Matthew documented the campaign from beginning to end. He explained why he suspected the backlink profile, what happened after submitting the disavow file, and why he ultimately decided to abandon the old domain entirely.

The biggest takeaway for me is that backlink quality deserves as much attention as backlink acquisition. Every backlink audit should include a review of suspicious referring domains instead of focusing only on building new links.

Limitations of SEO Gets Toxic Backlinks Case Study

This is a single client case study, not a controlled experiment.

The client’s SEO campaign included a domain migration, new citations, additional backlinks, digital PR, and ongoing optimization. Those changes make it impossible to isolate the impact of the disavow file from the rest of the SEO campaign.

My Review of SEO Gets Toxic Backlinks Case Study

Category	Rating	My assessment
Relevance to backlink SEO	10/10	One of the few studies examining whether toxic backlinks can limit organic growth.
Methodology	7.5/10	The campaign timeline is well documented, although several SEO changes occurred during the recovery.
Data quality	8/10	Google Search Console and Ahrefs data support the observations with real client performance metrics.
Practical value	9.5/10	Reminds SEOs that backlink audits should focus on backlink quality as well as backlink acquisition.
Confidence level	7/10	Based on one client campaign, so the findings should be interpreted as observational evidence rather than definitive proof.
Overall score	8.4/10	A valuable case study that broadens the backlink discussion beyond link building. It highlights the importance of maintaining a clean backlink profile and investigating toxic links when rankings fail to improve despite ongoing SEO work.

Study 7: Miles Burke’s Cheap Backlinks SEO Experiment (2025)

Miles Burke published Are Cheap Backlinks Worth the Money? My 3 Month SEO Experiment to answer a question every SEO has considered at some point.

Can cheap backlinks improve Google rankings?

Rather than relying on opinions, Miles created a brand-new test website using an aged domain with zero backlinks and a Domain Rating of 0. He then purchased an $11.52 backlink package on Fiverr, tracked the results for three months, and monitored every change using Ahrefs and Google Analytics.

Key findings from Miles Burke’s Cheap Backlinks SEO Experiment

Miles documented the entire experiment, from purchasing the backlinks to measuring their long-term SEO impact.

• The Fiverr seller promised 50 white-hat backlinks from websites with a Domain Authority above 65.

• Miles received 66 backlinks within 48 hours instead of the advertised 50.
• The backlinks were placed inside 500+ word articles containing one contextual link to his website.
• Most articles were generated using article spinning software and contained poor-quality, unreadable content.
• According to Ahrefs, 37 linking websites had a Domain Rating above 60, 21 websites had a Domain Rating between 10 and 60, and 8 websites had a Domain Rating of 0.
• Most backlinks came from free web hosting platforms, subdomains, and automatically generated websites.
• After 13 weeks, Ahrefs detected only 2 backlinks from the entire campaign.
• 34% of the backlinks had already disappeared within three months.

• Google Analytics reported no increase in organic traffic.
• The experiment produced no measurable improvement in rankings or SEO performance.

What I Liked About Miles Burke’s Backlink Experiment

Unlike most backlink discussions, Miles spent his own money and tested the claims himself.

The experiment also reflects a situation many beginners face. Cheap backlink packages promise hundreds of high-authority links for a few dollars, making them attractive to businesses with limited budgets. Instead of repeating industry advice, Miles verified whether those promises translated into better rankings.

Another point I appreciated was that he looked beyond the backlink count.

He inspected the linking pages, reviewed the content quality, checked how many backlinks remained live after three months, and monitored ranking changes through Ahrefs and Google Analytics. That produced a much more complete evaluation than simply counting how many backlinks were delivered.

The most important takeaway for me is that backlink quality always outweighs backlink quantity. Receiving dozens of backlinks made no measurable difference because the links came from low-value websites with little editorial oversight.

Limitations of Miles Burke’s Cheap Backlinks SEO Experiment

The experiment tested one Fiverr seller and one backlink package, so the results cannot represent every low-cost backlink service.

The campaign also used a brand-new website with no existing authority. Established websites with stronger backlink profiles could respond differently to similar backlinks.

My Review of Miles Burke’s Cheap Backlinks SEO Experiment

Category	Rating	My assessment
Relevance to backlink SEO	10/10	Directly answers whether cheap backlinks improve rankings.
Methodology	8.5/10	Real-world experiment with a controlled test website and a three-month observation period.
Data quality	8/10	Ahrefs and Google Analytics provided measurable SEO data throughout the experiment.
Practical value	10/10	Saves readers from wasting money on low-quality backlink packages.
Confidence level	8/10	Limited to one provider, but the methodology is transparent and easy to reproduce.
Overall score	8.9/10	One of the most practical backlink experiments in this review. The results reinforce a simple lesson: buying cheap backlinks increases backlink counts, but it does not improve rankings, traffic, or long-term SEO performance.

My Final Verdict: Are Backlinks Still Important for SEO in 2026?

After reviewing seven backlink studies, covering more than 11.8 million Google search results, 350,000 backlink placements, 1,000 domains, multiple AI search platforms, and real-world SEO experiments, my conclusion is straightforward.

Backlinks still matter.

The evidence is remarkably consistent. Every large-scale study found a positive relationship between high-quality backlinks and better visibility in Google Search or AI search engines. The debate is no longer about whether backlinks matter. It is about which backlinks matter.

The studies repeatedly showed that editorial backlinks, topical relevance, referring domain diversity, and authority produce the greatest SEO value. Simply increasing backlink numbers is not an effective strategy. A single link from a trusted, relevant website can outperform dozens of low-quality links.

I also found that backlink quality extends beyond traditional Google rankings. Both the Pitchbox and Semrush studies showed that authoritative backlink profiles continue to influence visibility in ChatGPT, Google AI Overviews, Gemini, Claude, Grok, and Perplexity. As AI search becomes more common, backlinks remain an important trust signal for both search engines and large language models.

At the same time, the studies challenged several outdated assumptions. Cheap backlink packages failed to improve rankings. Toxic backlink profiles created real SEO challenges. Nofollow links and image backlinks showed more value than many SEO professionals expected. These findings reinforce that modern link building is about earning trust, not manipulating algorithms.

If I had to focus on only four backlink metrics, they would be:

• Referring domains
• Topical relevance
• Editorial quality
• Authority of the linking website

Everything I reviewed points back to those four signals.

My biggest takeaway is simple.

Stop chasing backlink quantity.

Create content that deserves citations, build relationships within your industry, earn links from relevant publications, and monitor the quality of your backlink profile over time.

That approach has survived every Google update covered in these studies, and I believe it will remain the most reliable backlink strategy for years to come.