TL;DR: AI share of voice (SoV) is the proportion of AI answers about your category in which your brand appears, measured across a prompt set and usually against competitors. The base formula is your appearances divided by total measured answers; the competitor-normalized version divides by all tracked brands' appearances so shares total 100%. There's no universal "good" number — because AI answers name only a few brands, the honest benchmark is relative: your share versus competitors and your own trend over time.
What AI share of voice means
AI share of voice measures how often your brand shows up when AI engines answer the questions your buyers ask — relative to how often anyone shows up, and usually relative to your competitors. It's the direct descendant of share of voice in traditional advertising and PR, moved to a new surface: instead of your slice of category ad spend or media mentions, it's your slice of the AI answers that now shape purchase decisions.
Concretely: if you track 40 buyer questions in your category and run them through ChatGPT, AI share of voice asks in how many of those 40 answers does your brand appear, and how does that compare to your competitors' appearances? It turns a vague sense of "are we visible in AI" into a number you can track, compare, and move.
For ecommerce specifically, it's become a metric worth owning because the surface it measures is now a real channel — Adobe measured AI-referred visitors converting 54% better than non-AI traffic in its May 2026 retail data. Share of voice is how you measure your position in a channel that's actively converting.
The formula
There are two forms, and they answer different questions.
Base (absolute) share of voice, per engine over a prompt set:
AI share of voice = answers featuring your brand ÷ total measured answers
This tells you your absolute presence — "we appear in 30% of category answers on ChatGPT."
Competitor-normalized share of voice, which most competitive dashboards show:
SoV (normalized) = your appearances ÷ Σ appearances of all tracked brands
This makes the tracked competitive set total 100%, so you can read position directly — "we hold 22% of the AI mentions in our category; the leader holds 34%."
A worked example makes the difference clear. Across 40 tracked prompts on one engine, suppose your brand appears in 12 answers, Competitor A in 22, Competitor B in 8:
- Absolute SoV: 12 ÷ 40 = 30%.
- Normalized SoV: 12 ÷ (12 + 22 + 8) = 29%, versus 52% for Competitor A.
Same brand, two true numbers. The absolute figure says you're present in a healthy third of answers; the normalized figure says you're a clear second to a competitor who dominates. Both matter, and reporting only one hides half the story.
The choices that change the number
Before you compare any two share-of-voice figures — across tools, across time, or across a competitor's claim — check three definitional choices, because each moves the number:
- What counts as an appearance. Some tools count only an explicit name mention; others count your brand as present if it's named or cited as a source. Counting either is more forgiving and catches answers that link your page without naming your brand — which is common, and which a name-only count misses entirely.
- The sample floor. Share of voice over six answers is noise. A trustworthy implementation won't report SoV below a minimum sample (a common floor is roughly ten measured answers plus at least one tracked competitor), because tiny samples produce wild, meaningless percentages.
- Prompt-set stability. SoV is only comparable over time if the prompt set stays fixed. If prompts change between measurements, a shift in your SoV might just be a shift in what you measured.
These aren't pedantic. Two tools can report very different SoV for the same brand purely from these choices — so the number is only meaningful alongside its definition.
Why there's no universal benchmark (and what to use instead)
The question everyone asks — "what's a good AI share of voice?" — doesn't have an honest universal answer, and it's worth being clear about why rather than inventing one.
AI answers structurally name only a handful of brands. That means "100% share of voice" is impossible and even high absolute numbers are rare and category-dependent. Rates vary enormously by category (a concentrated niche behaves nothing like a crowded one), by engine (engines mention brands at very different base rates), and by the definitional choices above. A benchmark that ignores all that — "aim for X%" — would be misleading precision.
So the honest benchmarks are relative, not absolute:
- Versus your tracked competitors. "Second of four brands in our category, 22% vs the leader's 34%" is interpretable and fair. It uses the only denominator that reflects the slots actually available — the brands AI actually names.
- Versus your own trend. "Up from 18% to 24% over the last quarter" tells you whether your work is landing, on a stable prompt set.
- Per engine, not blended. Because engines have different base rates, judge each engine's SoV against that engine, not against a flat cross-engine average.
This is also why credible tools don't ship invented cold-start benchmarks: a peer-percentile benchmark requires real aggregate data across many brands, and fabricating one would be worse than offering none. If you see a tool quoting a confident universal "good SoV" number, ask where the data came from.
Using share of voice well in ecommerce
A few practices turn SoV from a vanity metric into a decision tool:
- Segment by query type. Your SoV on "best [category]" discovery queries and your SoV on "[you] vs [competitor]" comparison queries are different stories. Losing the comparison queries points at missing comparison content; losing discovery queries points at a broader authority gap.
- Pair it with trend and attribution. Absolute gains can hide relative losses if the whole category is rising. Read SoV alongside your own trend and, where you can, alongside AI-referred traffic. The monitoring tools guide covers the measurement discipline.
- Treat a big single-scan swing as suspect. AI answers are volatile; smooth across measurements before declaring a win or a crisis.
Some GEO tools compute AI share of voice automatically, defining an appearance as named-or-cited and enforcing a sample floor so the number stays honest. Stride, for example, reports SoV for Shopify and DTC brands using a named-or-cited definition with a minimum-sample gate, alongside the per-competitor breakdown — but the method matters more than the vendor: whatever you use, know its definition and read the number relatively.
Frequently asked questions
What is AI share of voice?
AI share of voice is the proportion of AI answers about your category in which your brand appears, measured against a defined prompt set and usually against competitors. It answers "when AI engines respond to buyer questions in our space, how often are we the brand that shows up, relative to everyone else." It's the AI-search equivalent of share of voice in traditional media.
How do you calculate AI share of voice?
The base formula is your brand's appearances divided by total measured answers, per engine, across a prompt set. The competitor-normalized version divides your appearances by the sum of all tracked brands' appearances so the shares total 100%. Both are useful — the first shows absolute presence, the second shows your position relative to competitors.
What's a good AI share of voice for an ecommerce brand?
There's no universal benchmark, because AI answers name only a handful of brands and rates vary enormously by category, engine, and how the metric is defined. The honest benchmark is relative — your share versus your tracked competitors, and your trend over time. Being second of four tracked brands in your category is a meaningful, interpretable number in a way an absolute percentage isn't.
Does AI share of voice count mentions or citations?
It depends on the tool, and the choice changes the number. Some count only explicit name mentions; others count your brand as present if it's either named or cited as a source. Counting either is more forgiving and captures answers that link you without naming you. Always check which definition a tool uses before comparing numbers across tools.
How is AI share of voice different from traditional share of voice?
Traditional share of voice measures your presence in advertising or media against competitors. AI share of voice measures your presence inside AI-generated answers to buyer questions. The concept is the same — relative visibility — but the surface is synthesized answers that name only a few brands, so the dynamics and the realistic numbers differ from paid media.
Stride measures AI share of voice for Shopify and DTC brands across multiple engines. The free audit includes a share-of-voice read against the competitors AI names in your category — no account required.