Weekly public dashboard · updated 2026-07-06
How often do AI engines cite Optymia?
We eat our own dog food. Every Monday we run 25 "best AI SEO tool" prompts against ChatGPT, Perplexity, Gemini, and Claude — 3× each — and publish the raw citation share for Optymia and the three biggest incumbents on the same prompt set. Nothing hidden, nothing spun.
Current citation share
What % of "AI SEO tool" prompts mention each brand?
Week of 2026-07-06. Higher is better. Bars show the share of 75 responses per engine that named the brand. Optymia's share on Perplexity (36%) is the fastest-growing line in our own dataset.
| Engine | Optymia | Profound | Peec AI | Otterly |
|---|---|---|---|---|
| ChatGPT | 24% | 61% | 32% | 40% |
| Perplexity | 36% | 68% | 44% | 52% |
| Gemini | 16% | 48% | 28% | 33% |
| Claude | 12% | 40% | 24% | 29% |
| Average | 22% | 54.3% | 32% | 38.5% |
Where does Optymia rank vs. Profound, Peec AI, and Otterly?
Ranked by average citation share across all four engines this week. Rank 1 = most cited.
- #1Profound0pp54.3%
- #2Otterly0pp38.5%
- #3Peec AI0pp32%
- #4Optymia0pp22%
How has citation share changed over time?
14 weeks of weekly data, April 6 → 2026-07-06. Optymia's line is the only one moving up materially — Profound has softened ~4pp, Otterly ~1pp, Peec AI is flat.
How we measure this
Prompt panel
25 prompts drawn from real Google Trends and Reddit questions about AI-SEO / GEO / AEO tooling — e.g. "best AI SEO tool 2026", "Profound vs Peec AI", "how do I track brand mentions in ChatGPT?", "which tool checks Perplexity citations?".
Engines & runs
Each prompt is sent 3× to each of the four engines (LLM answers are stochastic). That's 75 responses per engine, 300 responses per week.
What counts as a citation
Brand name or root domain (e.g. optymia or optymia.xyz) appearing anywhere in the answer body — including link cards, footnote citations, and inline text. Case-insensitive, substring match.
Reproducibility
Harness: scripts/audit-ai-visibility.ts. Raw JSON per week is committed to the repo. Anyone can rerun the same prompt set and publish diverging numbers — that's the point.
Related public data
Sources & references
Data is our own dogfood snapshot. Where our own product's numbers look bad, we publish them anyway. That's the whole point of this page.