SKYA is an AI Visibility Analyst platform that tracks how ChatGPT, Gemini, Perplexity, Claude, and Google AI Overview mention your brand. Run live AI visibility audits, SEO insights, and competitor intelligence, no setup required.

Which AI engines does SKYA monitor?

SKYA monitors ChatGPT, Google Gemini, Perplexity, Claude, and Google AI Overview, the five AI engines that account for the majority of AI-generated search traffic.

Is there a free trial?

Yes. SKYA offers a 7-day free trial with no credit card required.

What is AEO and why does it matter?

Answer Engine Optimisation (AEO) is the practice of optimising your content so AI systems cite and recommend your brand in their answers. As users shift to AI assistants for search, AEO determines whether your brand appears in AI-generated responses.

How is SKYA different from traditional SEO tools?

Traditional SEO tools track Google rankings. SKYA tracks how AI engines like ChatGPT, Gemini, and Perplexity actually mention, or overlook, your brand in their answers. It is purpose-built for AI search visibility, not blue-link rankings.

Do I need technical knowledge to use SKYA?

No. SKYA is fully web-based. You can run AI visibility audits and access insights without any coding or technical setup.

How SKYA Measures AI Search Volume (Methodology)

2026-07-17 · AI Visibility · By SKYA

AI search volume is the missing metric of the AEO era. This is the exact 3-layer methodology SKYA uses to measure prompt volume across ChatGPT, Gemini, Claude, Perplexity and Google AI Overview, and how we stay honest about what is observed vs modeled.

How SKYA Measures AI Search Volume AI search volume is the missing metric of the AEO era. Google Search Console tells you what people searched on Google. It does not tell you what they asked ChatGPT, Gemini, Claude, Perplexity, or Google's AI Overview. This is how SKYA closes that gap, and how we stay honest about which numbers are observed and which are modeled. The 3 layers of our volume model Every AI Volume number you see inside the Prompt & Query Research panel is tagged with one of three confidence levels: Measured - the number comes directly from a data provider that indexes AI prompt data (for ChatGPT baseline) or from a live SERP call (for Google AI Overview). Probed - the number is refined from SKYA's own first-party dataset built by weekly per-platform probes. Modeled - the number is derived from the ChatGPT baseline multiplied by that platform's estimated share of the AI answer market. We surface those tags on every platform tile so nothing is ever presented as gospel when it is actually an estimate. The ChatGPT baseline For ChatGPT specifically we lean on a paid AI keyword-volume dataset that tracks monthly prompt volume for millions of queries. This gives us a real 12 month series per keyword. It is not perfect (no dataset is), but it is a real, indexed signal, not a guess. That series drives the big number and the sparkline you see at the top of every research result. The platform breakdown Once we have the ChatGPT baseline, we translate it into a full 5-platform picture using published market-share coefficients: ChatGPT (measured baseline) Google AI Overview Gemini Perplexity Claude These coefficients are stored in a table (pqrplatformcoefficients) so we can update them the moment a credible new market-share study drops. Every non-ChatGPT number starts life as "modeled" and gets upgraded to "probed" once we have enough first-party probe data for that (keyword, platform) pair. Weekly probes: how "modeled" becomes "probed" Every Monday 03:00 UTC a scheduled job (pqr-weekly-probe) sends recent prompts to each of the four non-baseline platforms: Perplexity, queried directly Gemini, queried directly Claude, queried directly Google AI Overview, read from a live AI Mode search result For each probe we record: was the platform presented, did it cite the brand, in what position, and how long did the call take. Once we have 3 or more probes for a (keyword, platform) pair inside a 30 day window, the platform's confidence flips from modeled to probed and the observed presence rate is blended into the volume estimate. AIO probes flip to measured immediately, because a live SERP call is ground truth. The 7-day cache We cache volume snapshots per (keyword, country) in pqrvolumesnapshots for 7 days. If you research the same keyword twice inside a week, the second call reuses the snapshot instead of hitting the paid data provider again. You can see the "cached" pill under the total when this happens. Trend direction The little "▲ +18%" or "▼ -12%" pill…

← Back to the SKYA Journal