Competitor share of voice in AI answers
When a buyer asks the questions in your category, who gets named? Share of voice counts every brand mention across every answer — you against each rival, per engine and over time — and prints the margin of error beside every number, because at real sample sizes the margin is the difference between news and noise.
How it works
Every answer, every brand
Each run is checked for your product and every tracked rival. Rates are computed over answer cells, and failed engine calls are excluded from denominators instead of quietly deflating everyone.
Split by engine
The engine that loves your rival is often not the one that ignores you. Per-engine breakdowns show where each brand's strength actually lives.
Trended with honest deltas
Week-over-week movement is only reported when it beats the printed margin. Silence is what honesty sounds like when nothing moved.
In depth
Question-level drivers
When your share moves, the overview names the specific questions that flipped — won or lost — instead of leaving you to reverse-engineer a chart.
Rivals tracked by plan
Starter tracks 6 rivals, Growth and Scale 10. Rivals are the brands AI actually names in your answers — you'll see them in the inspector before you track them.
Rank without theater
You get your rank among tracked rivals, and when two brands sit inside each other's margins the UI says tied — we don't manufacture a #2 vs #3 drama the data can't support.
What we won't print: a daily share-of-voice wiggle chart. At weekly sample sizes, daily deltas are noise sold as insight; we re-measure weekly and only speak when movement clears the margin.
Common questions
What is AI share of voice?+
The percentage of AI answers in your category's buyer questions that name each brand. CiteGraph computes it from the same stored answers you can read in the inspector — every number is auditable.
How is it different from search share of voice?+
Search SoV measures rankings on a results page; AI SoV measures presence inside a single synthesized answer, where being absent means being invisible. The winner-take-most dynamics make gaps matter more.
Why do you show error bars when competitors don't?+
Because rates from 60–150 sampled answers genuinely carry ±8–13 points of uncertainty. Hiding that makes charts prettier and decisions worse. It's also how you can tell we're not inventing movement to keep you engaged.
See it on your own domain — first dossier in about a minute.