How is a brand's visibility in AI answers measured?
AI visibility is best measured as a sampled frequency — the share of times your brand appears across many repeated runs of the same prompts — reported with its variance, never as a single rank. AI answers are non-deterministic: ask the same question twice and the brand list often differs.
Why a single number lies
SparkToro's testing found that asking ChatGPT and Google the same prompt 100 times returned identical brand lists in well under 1% of runs, and identical ordering essentially never. A one-off "we rank #2" screenshot is noise. The honest unit is a distribution: run each prompt 60–100+ times per engine and report the percentage of runs your brand appears in, with a confidence band.
What to track
Three things, per engine: presence (the frequency % you appear), share of voice (your presence relative to competitors on the same prompt set), and the citation gap (the independent sources cited in answers where you're absent). Tracking the gap is what turns measurement into action — it tells you exactly which sources to earn.
Honesty as a feature
Because the signal is noisy, the credible approach reports variance openly and never claims causation it can't support. A platform that shows you a precise rank is overfitting to randomness. New Map measures presence as a frequency band that tightens as it samples more — and every recommendation that follows is checked by an independent verifier before it reaches you.
Frequently asked
Why does my AI ranking change every time I check?
Because AI answers are non-deterministic — the model samples differently each run. SparkToro found identical brand lists in under 1% of repeated runs. That's why visibility must be measured as a frequency over many samples, not a single check.
How many samples are needed for a reliable read?
Enough to estimate a stable frequency with a usable confidence band — typically 60–100+ runs per engine per prompt set. Fewer than that and you're reporting noise.
What is share of voice in AI answers?
The share of relevant AI answers that cite your brand versus your competitors across the same prompt set — a like-for-like comparison, not an absolute score.
Published by New Map — the AI visibility OS. Updated 2026-06-14.