Skip to content
SEO Madmanby Adam Hafez

Do backlinks still correlate with rankings in 2026?

Backlinko's 11.8M-result study and Ahrefs' 1M-keyword study both find a positive link between referring domains and rank, but the coefficient is weak.

Published: · Read time: 2 minutes

Written byAdam Hafez
Share
Plain Markdown
A brown metal chain against a white background

Key takeaways

  • Backlinko's 11.8-million-result study, updated April 2025, found position 1 pages average 3x more referring domains and 3.8x more backlinks than positions 2-10.
  • Ahrefs' own January 2025 study of 1 million US keywords found a Spearman correlation of only 0.255 between referring domains and ranking, a weak coefficient by normal convention.
  • Both firms explicitly warn that their own data cannot show causation, only association.
  • Google's Gary Illyes told Pubcon Pro in 2023 that links are not among the top three ranking factors and have not been for some time.
  • A third firm's claim that backlinks are worth exactly 13% of algorithm weight has no disclosed methodology behind it and should not be read as a measured fact.

Two firms with two different large samples both find the same shape of relationship: pages that rank higher tend to have more referring domains pointing at them. Where they disagree is on how much that relationship is worth leaning on, and both say so themselves.

The two studies, side by side

Backlinko’s ranking-factor study, last updated in April 2025, pulled backlink data from Ahrefs across 11.8 million Google results and found position 1 pages average 3x more referring domains and 3.8x more total backlinks than positions 2 through 10. That is a large, memorable gap, and it is the number most often cited to argue backlinks still decide rankings.

Ahrefs ran its own study in January 2025, across the top 1 million US keywords by search volume, and reported the actual Spearman correlation: 0.255 for referring domains, 0.250 for followed referring domains only. Both are weak by the normal reading of a correlation coefficient, where values below roughly 0.3 indicate a real but modest association, not values that would let anyone predict rank from link count alone. Ahrefs published this specific number knowing it undercuts the stronger-sounding multiple-based framing, and both posts add the standard warning: correlation is not causation, and neither study’s design can rule out reverse causation.

Google does not publish ranking-factor weights, so the closest thing to an official read comes from what its own staff say on the record. At Pubcon Pro in 2023, Google’s Gary Illyes told attendees links are not among the top three ranking factors and have not been “for some time,” adding that a page with zero external or internal links has ranked number one on the strength of content alone. That statement does not say links are worthless. It does say the industry’s mental model, where link count is close to determinative, is outdated by Google’s own account.

The 13% claim that should not be treated as fact

A frequently repeated figure holds that backlinks represent exactly 13% of Google’s algorithm weight, down from 15% in 2024. That number comes from First Page Sage’s own recurring internal estimate, presented without a disclosed methodology, survey instrument, or error margin. Google has never confirmed that any ranking factor carries a fixed percentage weight, and multiple current and former Google staff have said directly that no single factor gets a stable, query-independent share of the algorithm. Treat the 13% figure as one firm’s proprietary estimation exercise, not a measured fact, and do not cite it as if Google had disclosed it.

Why we care

The honest read is a modest, real, positive association: more referring domains track with higher rank, at a strength that leaves most of the variance in Google’s output unexplained. That is enough to justify earning links as part of a page’s overall authority, and not enough to justify treating link count as a lever that moves rank on its own. Budget link-building as one input among many, not the one your ranking strategy hinges on, and be skeptical of any number, including a precise algorithm-weight percentage, that a source presents as more certain than Google itself has ever been willing to state.

The evidence

Sample
Backlinko: 11.8M results; Ahrefs: 1M keywords, top-20 SERPs

Hypothesis: The number of unique referring domains linking to a page correlates with its Google ranking position, and this correlation persists in 2026 correlation studies despite years of predictions that AI and relevance signals would displace backlinks as a ranking input.

Method: Cross-referenced two independent, still-maintained correlation studies with disclosed methodology: Backlinko's ranking-factor study, run against Ahrefs' backlink index across 11.8 million search results, and Ahrefs' own Spearman-correlation study across the top 1 million US keywords by search volume. A third, widely repeated claim that backlinks represent a fixed percentage of Google's algorithm weight was checked against its source and against Google's own public statements on ranking-factor weighting, since Google has never published or confirmed such weights.

Findings

  • Backlinko (updated Apr 2025, with Ahrefs backlink data): position 1 results average 3x more referring domains and 3.8x more total backlinks than positions 2 through 10.
  • Ahrefs (Jan 2025, 1M US keywords, top-20 SERPs): Spearman correlation between referring domains and ranking position is 0.255, and 0.250 for followed referring domains specifically, both weak by conventional correlation-strength thresholds.
  • Both studies state directly that their results are correlational and cannot establish that backlinks cause the ranking difference observed.

Limitations: Correlation studies of this kind cannot separate 'backlinks cause rankings' from the more mundane explanation that older, more authoritative, better-resourced pages naturally accumulate both backlinks and rank over time, a shared-cause confound neither study can rule out. Google has never confirmed or published the weight any single ranking factor carries, so any study's headline number, however large the sample, describes association in Google's output, not the inner workings of the algorithm that produced it.

About the author

Adam Hafez
Adam Hafez

Founder

Founder, UpgradIQ, Inc.

Adam Hafez works on technical SEO and search measurement: how pages get crawled, indexed, ranked and now quoted by answer engines. He founded UpgradIQ, which reads Google Search Console and GA4 to tie ranking movement back to the changes that caused it. He publishes what the data supports and states the limits of it.

  • Technical SEO
  • Search Console and GA4 measurement
  • Answer engine optimization
  • Structured data

The briefing

One email when something in search actually changes. No digest padding.

Subscribe