Unexpected Sources AI Engines Still Cite

The LLM Junkyard: Unlikely Sources Causing LLM Citations

Info

Source: NP Digital
Date: July 2025
Category: AI & GEO Optimization
Study Methodology: Review of 700 AI citations identifying source quality, authority, and relevance.

Not all AI citations come from great sources. This data reveals the types of low-quality or surprising domains that still surface in LLM responses.

Essential Statistics

Some low-authority sites appeared repeatedly in AI citations.
Outdated content was still referenced.
Forum-style content appeared sporadically.
Content relevance sometimes outweighed quality.
Citation quality varied by engine.

Key Takeaways

AI citation quality is inconsistent.
Relevance can override authority.
Low-quality sources still leak through.
Not all citations imply endorsement.
Monitoring AI outputs is necessary.

Actionable Insights

Do not assume AI citations equal quality validation. Review sources critically.
Strengthen your own content relevance to crowd out weaker competitors.
Monitor AI responses for misinformation risks in your category.
Update evergreen content regularly to avoid being displaced by outdated sources.
Use citations as signals, not guarantees of trust.

AI citations are signals, not judgments. — Neil Patel

In this webinar, Neil Patel and the NP Digital team will show why trying to outpublish bigger brands is rarely the best approach.

In this webinar, Neil Patel and the NP Digital team will show why trying to outpublish bigger brands is rarely the best approach.