Frequently asked

Do you measure voice assistant answers?

50 questions15 engines17 surfaces Get your free AI visibility check

The short answer

Much of what a voice assistant says out loud is produced by text answer engines we already measure. Samsung's Galaxy Bixby serves web-question answers as Perplexity-based search summaries with source cards (One UI 8.5), Apple has been reported to have agreed to bring Google's Gemini to Siri (CNBC and TechCrunch, January 2026), and ChatGPT voice conversations are answered by GPT itself. Perplexity and GPT are judged places CiteAngle measures with seven repeated runs per query — so a large share of "which brand gets cited in the spoken answer" reduces to the citation coordinates of those text engines. What stays voice-specific — engines still mid-rollout like the Siri rebuild, spoken-answer truncation, device-layer localization — we track as monitored items until they settle.

Why this matters

When a spoken answer gets you wrong there is no screen to point at. Nothing gets filed, nothing gets fixed, and the answer keeps being read out.

The text engines producing those spoken answers are read on the same grid, and the ones still rolling out are listed as watch items with the date they change.

Questions buyers read next

See every question

Where this answer comes from — Published pricesHow we measureTerms, defined

Ready to turn US visibility evidence into a market plan? Tell us the US buyer, target state and decision date.

Request a written proposal