How to measure share of voice in AI answers
In short
AI share of voice is your brand's portion of all brand mentions across a fixed set of questions and engines, measured over a time window rather than from single answers.
How do you choose the competitor set?
Start with the brands assistants already name in your category, not the ones on your internal battlecard. Run your questions, collect every brand mentioned, and keep the recurring names. Then add any strategic competitors you want tracked deliberately.
Discovered
Brands the assistants name repeatedly. These define the real competitive set.
Declared
Brands your business considers competitors, tracked even if assistants rarely name them.
Excluded
Retailers and marketplaces, unless you are one, since they distort the denominator.
Why segment by engine and intent?
Because the aggregate hides the work. A brand can hold 30 percent share on one engine and 5 percent on another, or be strong in comparison questions but absent from best-of lists. The segment level is where a specific action becomes obvious.
Intent segmentation is often more useful than engine segmentation for content teams. Losing the "alternatives to" questions calls for different work than losing the "best for beginners" questions.
How often should it be reported?
Monthly for stakeholders, weekly for the team doing the work. Run checks more often than you report, so each reported number is an average across many answers rather than a single run that happened to land on reporting day.
Key takeaways
- Define the competitor set before you measure, or the denominator moves under you.
- Report share of voice next to mention rate: one is relative, the other absolute.
- Segment by engine and by question intent, because averages hide the actionable gaps.
- Keep the raw answers so any disputed number can be checked in seconds.
Frequently asked questions
Is share of voice better than mention rate?
They answer different questions. Mention rate tells you how visible you are in absolute terms; share of voice tells you how visible you are relative to competitors. Report both, because one can rise while the other falls.
How many questions do I need for a reliable number?
Enough to cover your buying intents in each market, typically dozens rather than a handful, run repeatedly. Reliability comes from repetition across a fixed set, not from a large one-off sample.
Should retailers count as competitors?
Usually not. If an assistant names a marketplace as a place to buy, that is a different signal from naming a competing brand. Track them separately or exclude them so the denominator stays meaningful.
See what AI says about your brand
Add one domain and get your first visibility report across five assistants. Free to start, no card required.