AI Share of Voice

AI Share of Voice is your brand's qualifying mentions divided by all qualifying brand mentions across a fixed panel of category queries, engines and time windows. The definition of record is the Machine Relations glossary entry; the formula, the denominator rules and how to measure it by question shape are on AI Share of Voice: Definition, Formula, and How to Measure It. This entry answers the question a marketing team asks next: how the number relates to brand visibility and authority, and which metric to read beside it.

What It Tells You About Brand Visibility

Traditional share of voice counted media mentions. AI share of voice counts mention breadth inside the answers buyers actually read, so it moves with how many independent sources corroborate your claims rather than with how much you publish. That makes it a presence measure: a rise says engines name you more often across the panel you defined, relative to every brand tracked on the same panel.

Read it as reach, in the same register a marketing team reads impressions. It answers one question well: in the conversations we chose to watch, how much of the naming is ours.

What It Does Not Tell You About Authority

Authority is a selection decision, not a mention count. An engine naming your brand in a sentence it assembled from someone else's page has not treated you as a source; it has treated you as a fact inside another publisher's answer. The metric that tracks that selection is share of citation, which counts how often engines cite your brand as the source of the claim.

So a brand can be mentioned widely and cited rarely. That is the most common failure pattern behind a healthy-looking dashboard, and it makes authority look better than it is. It also runs the other way: a narrow, heavily cited presence in two question shapes is a stronger commercial position than a wide, uncited presence across twenty.

Share of voice is also descriptive of the past, not predictive of pipeline. It says nothing about whether the mentions reached a buyer, survived to a shortlist, or produced revenue. Keep the claim at the level the measurement supports.

Why One Blended Number Misreads Your Standing

A single share-of-voice figure averaged across every prompt shape measures the composition of your prompt library, not your standing in the market. Change the mix of question shapes you ask and the number moves without anything changing in the market.

The Machine Relations Index shows how large that distortion is. In release 2026-09-21 (window 2026-05-10 to 2026-09-21, 16,146 observed answer runs across six engines), the Index measured 13 grade-A domains across four or more question shapes inside a single category, giving 84 category-domain pairs. The median pair had a 6.4x spread between the domain's strongest and weakest question shape in that same category, and 90% of pairs spread by 2x or more.

One pair makes it concrete: in Consumer Finance, forbes.com was cited in 39.4% of "best X" answer runs (52 of 132) and 2.2% of "X vs Y" runs (3 of 138), in the same category, window and engine set. A blended average of the two would describe neither.

The operational consequence is a fixed contract before you report anything: fix the panel, the engines, the aliases and the counting rules, then measure each question shape on its own denominator. The measurement protocol sets that contract. The AI Visibility Audit runs it across the major engines.

What to Read Beside It

Metric	Counts	Reads as
AI Share of Voice	Your qualifying mentions over all brand mentions on the panel	Breadth of presence
Share of Citation	How often engines cite your brand as the source	Depth, and the closest proxy for authority
AI Mention Rate	Share of panel queries where you appear at all	Baseline coverage
AI Visibility Score	Breadth and depth rolled into one composite	Direction, once the inputs are fixed

Breadth and depth move independently, so report them side by side and never let one stand in for the other. Use Citation Share for category-level competitive analysis, and see GEO for the optimization framework and Machine Relations for the discipline the metrics belong to.

How to Raise It

The primary lever is earned media in publications engines already cite, because breadth is distributed across independent publishers rather than concentrated in brand-owned pages. In the same Index release, editorial publications account for 13,560 cited-domain observations across 1,231 distinct domains, against 11,280 observations across 823 vendor-owned domains: the editorial pool is wider, so each additional independent source adds panel coverage a brand cannot add by publishing more of its own pages. AirOps puts the same relationship at brands being 6.5x more likely to be mentioned through third-party sources than their own domains in its October 2025 sample of top-of-funnel commercial queries — one vendor's data.

More high-authority mentions produce more AI citations, and citations are what raise share of voice and share of citation together.

Frequently Asked Questions

How does AI share of voice relate to brand visibility and authority?

It measures mention breadth: how often your brand appears across a fixed panel of category queries and engines, relative to every brand tracked on the same panel. That reads as visibility. It does not establish authority, because being named inside an answer is not the same as being cited as the source of it. For authority, read share of citation next to it.

Does a rising share of voice mean more pipeline?

No. It is a descriptive presence measure over the panel you defined. A rise says engines name you more often there; it carries no claim about reach to a buyer, shortlist survival or revenue. Attribute pipeline with referral and self-reported-source data, not with this number.

Should I report one blended share of voice number?

No. In the Machine Relations Index release 2026-09-21, the median grade-A domain measured across four or more question shapes in one category spread 6.4x between its strongest and weakest shape, and 90% of those pairs spread 2x or more. Report per question shape on its own denominator, and state the panel, engines and window beside the figure.

How is it different from share of citation?

Share of voice counts mentions; share of citation counts how often engines cite your brand as a source. Breadth and depth move independently, and the AI Visibility Score rolls both up.

Where is the formula?

On AI Share of Voice: Definition, Formula, and How to Measure It by Question Shape, with the denominator and counting rules.