Which sources AI answers cite.
Illustrative examples of how cited sources can be compared across engines and categories.
Key findings
Shown from an illustrative example set until enough live workspace scans accumulate for this quarter.
- ›Reddit is the single most-cited source in buyer AI answers this quarter, at ~31% of cited venues.
- ›Only ~11% of cited domains overlap between ChatGPT and Perplexity. Citation patterns differ between engines.
- ›Gemini cites Reddit at ~0.1% while Google AI Mode cites it at ~44%. Citation share varies by engine.
- ›Recorded citations identify the sources used in AI answers.
Most-cited venues, Q2 2026
| Category | Presence leader | Share | Biggest open gap | Movement |
|---|---|---|---|---|
| Recovery devices for athletes | PulseCore | 31% | noise-level questions | ▼ slipping |
| Home fitness equipment | MotionLab | 24% | small-space setup | ▲ rising |
| AI sales assistants | (open) | 38% | pricing transparency | ▲ rising |
| Community tooling | (open) | 12% | intent vs. keyword | ▬ flat |
| Auth / identity | (open) | 29% | self-host story | ▲ rising |
Illustrative example, with fictional company names, while live category data accumulates across workspaces.
Where each engine reads
Citation patterns vary by AI engine.
Wikipedia + Reddit >25%; editorial press heavy; Reddit share volatile
→ Reddit + Q&A + editorial; track continuously
Reddit dominant (~24%); skews LinkedIn, NIH, G2
→ Reddit + G2 / review + LinkedIn for B2B
Reddit ~44% of social cites; classic SEO still converts
→ Reddit + owned-surface evidence + indexed forums
Reddit ~0.1%; rewards traditional SEO + reference authority
→ Owned surfaces, docs, high-authority Q&A
Citation share over time
Citation share by source over the last five quarters.
Methodology
We run each category’s buyer query set across ChatGPT, Perplexity, Google AI and Gemini, multiple passes per prompt (answers are probabilistic; one pass is noise), and resolve every cited URL to its canonical venue. Weights are the share of resolved citations per venue, per engine, computed from live workspace scans once enough have run; until then, the sections above show a clearly labeled illustrative example instead of measured figures. Cite this page: grounds · The Grounds Index · Q2 2026.
Want your category’s full table, with who owns each room and where your gap is?
Run a free scan