AI citation share by source category | Trakkr Research
Current category mix in the live Trakkr citation index.
Methodology: Built from the live citation index backing Trakkr Research, using 8,777,687 citations across 208,567 unique domains observed between October 3, 2025 and April 17, 2026.
Summary
The benchmark pattern reveals a highly fragmented citation landscape where a massive 84.51 percent long tail absorbs the vast majority of citation volume. Among named categories, structured factual layers like Reference lead with 6.75 percent, followed by Social at 3.82 percent, indicating that AI models rely heavily on a broad mix of domains rather than a few concentrated sources.
Benchmark rows
| Metric | Value | Context |
|---|---|---|
| Reference category share | 6.75% | Reference sites are the largest named category in the current index. |
| Social category share | 3.82% | Social sources matter materially, led by YouTube, Reddit, and LinkedIn. |
| Review category share | 0.89% | Reviews are smaller in raw share but disproportionately commercial. |
| Docs category share | 0.20% | Docs are cited less often overall, but they show up in technical query classes. |
| Long-tail share | 84.51% | Most citations are spread across a very large long tail of domains. |
Ranked view
| Item | Value | Detail |
|---|---|---|
| Long-tail mixed domains | 84.51% | Most citations land outside the named source buckets. |
| Reference | 6.75% | The cleanest structured factual layer in the index. |
| Social | 3.82% | Community and creator content remains part of the answer surface. |
| Reviews | 0.89% | Small in share, strong in buying-intent use cases. |
| Docs | 0.20% | Niche overall, important for implementation queries. |
Related pages
Continue through the same study cluster.
- which sources dominate ai citations - Related answer page
- how concentrated is the ai citation landscape - Related answer page
- wikipedia leads the current citation index - Related fact page
Data & Sources
- Where AI Gets Its Answers - Flagship study behind this page
- Page JSON - Machine-readable companion file