Clairon

Domain Authority vs AI Citation Authority: Why DA Predicts Less Than 4% of AI Citations

Hugo Debrabandere

Hugo Debrabandere

Co-founder · Clairon

Apr 29, 2026

The first reflex when an SEO team starts thinking about GEO: “we just need higher Domain Authority.” The data says the opposite. Domain Authority correlates with AI citation probability at r=0.18. Squared, that’s r²=0.032, meaning DA explains roughly 3% of the variation in whether AI engines cite your brand. For comparison, E-E-A-T signals correlate at r=0.81 in the same study. The chain is broken. Domain Authority does not buy you AI citations the way it bought you Google rankings.

Below: what actually drives AI citation authority in 2026, the 5 signals that replace DA, and a worked example of a DR-30 brand outranking a DR-85 site on Claude.

The DA-AI citation correlation: r=0.18 explained

The 2026 Wellows analysis of citation signals across major AI platforms found:

  • Domain Authority (Moz DA / Ahrefs DR): r=0.18 with AI citation probability
  • Backlink count: r=0.218
  • Brand mentions: r=0.664
  • E-E-A-T signals: r=0.81

Squared (the share of variance explained):

  • DA: 3.2%
  • Backlinks: 4.8%
  • Brand mentions: 44%
  • E-E-A-T: 65.6%

The translation: AI engines are using fundamentally different inputs from Google’s ranking algorithm. A 5-year-old DR-85 site without named authors, original data or current refreshes will get out-cited by a 6-month-old DR-30 site that has all three.

r=0.18
Domain Authority correlation with AI citation probability
r=0.81
E-E-A-T signals correlation
40%
AI citation lift from adding visible author credentials

What actually drives AI citation authority

Three signal families do the work.

  • Brand mention density. How often your brand name appears on third-party sites in topical contexts. Not links. Mentions. r=0.664.
  • E-E-A-T signals at the page level. Visible author with credentials, named sources within the passage, current refresh date. r=0.81. Single highest-leverage edit: visible author byline lifts AI citations by ~40%.
  • Original data and proprietary insights. Pages with at least 3 unique data points are more likely to be cited in AI Overviews.

The 5 signals that replace DA

Mapping old signals (SEO) to new signals (GEO)
Old signal (SEO)New signal (GEO)How to measure
Domain Authority / DRBrand mention frequencyManual count on Reddit / Wikipedia / G2 / newsletters
BacklinksSource diversity in citationsCitation analysis across 3+ engines
On-page SEO scoreAnswer capsule + FAQ densityThe 5-min citation test
Internal linksExternal links to authoritative sourcesOne per H2 minimum
Page agedateModified freshnessUpdated within 30 days for ChatGPT, 90 for Perplexity

The reframe: DA was a single composite score. AI citation authority is a 5-axis profile. You need to score on each axis individually; a high score on one axis cannot compensate for a 0 on another.

Worked example: DR-30 outranks DR-85 in Claude

A real B2B SaaS rewrite, anonymized. Two competing pages on the prompt “best customer feedback tool for product teams”.

Page A (DR-85, 5 years old, never refreshed)

  • 2,400 words, no answer capsule, story-first opening
  • Author byline missing
  • 0 named external sources
  • dateModified: 2023-11-04
  • 4 backlinks per month average

Page B (DR-30, 6 months old, refreshed monthly)

  • 1,800 words, 40-word answer capsule, question-shaped H2
  • Visible author byline with credentials
  • 4 named external sources (Stripe State of SaaS, Forrester, ConvertMate, Princeton GEO paper)
  • dateModified: 2026-04-12 (refreshed every 4 weeks)
  • 0 new backlinks in 90 days

Result on Claude: Page B cited at rank #2. Page A cited at rank #11 (off the front of most answers). Citation gap: 8 positions, with no shift in DR or backlink profile.

How to build AI citation authority

Sequenced, in order. Skipping a step adds 3 to 4 weeks before the metrics move.

Add a visible author byline with credentials

Single highest-ROI move. +40% citation lift. Author bio: name, role, 1-line credential. Schema this with Author / Person markup.

Add one named external source per 150 words

Real companies, real studies, real authors. Link to originals. The brand-mention correlation kicks in here.

Add an answer capsule under every H2

40-to-60-word block answering the H2’s implicit question. +40 to +70% citation lift.

Refresh monthly

Bump dateModified, update one statistic or example, ping IndexNow. Pages updated within 30 days earn 3.2× more ChatGPT citations.

Build third-party brand mentions

5 to 10 substantive Reddit replies per month, refresh G2 / Capterra quarterly, pursue Wikipedia entry if notable.

What’s next

For tactic-level moves engineering each signal, read How to Get Cited by AI Search Engines.

For entity-level work strengthening brand mention frequency, read Entity Optimization Guide for GEO.

For the 12-week sprint integrating all 5 signals, read How to Do GEO in 2026.

Domain Authority told you who deserved a Google ranking. AI citation authority tells you who deserves to be quoted. Two different questions, two different answers.

Frequently asked questions

Does this mean Domain Authority is irrelevant?
No. DA still matters for traditional SEO ranking. But the marginal value of DA for AI citations is small (r=0.18). If you have 8 hours to spend, spend them on author credentials, named sources and refresh cadence, not on link-building campaigns.
What about backlinks specifically?
Backlinks correlate r=0.218 with AI citations (slightly higher than DA but still weak). Brand mentions correlate r=0.664. A Reddit thread that names your brand without linking is worth more for AI citations than a backlink from a low-authority blog.
How long does it take to build AI citation authority?
The page-level edits lift citations within 2 to 4 weeks. The third-party brand-mention work compounds over 8 to 12 weeks for the first measurable shift, then continues for 12 to 24 months.
Can I buy AI citation authority?
Sort of, but not the way you bought backlinks. You can pay for newsletter sponsorships, podcast appearances, industry research participation. You cannot buy a citation directly.
Why is E-E-A-T's correlation so much higher than DA?
Because AI engines explicitly trained on E-E-A-T signals as part of their alignment with Google's quality framework. The training data shaped the inference behavior.
How do I know if my page passes the E-E-A-T bar for AI?
Check 4 things: visible author byline with credentials, at least 3 named external sources, current dateModified (under 12 months), and a clear topic match. If 3 of 4 pass, you're competitive. If 4 of 4, you're in the top citation tier.
Summarize with Claude
Summarize with Perplexity
Summarize with Google
Summarize with Grok
Summarize with ChatGPT