VISIBILITY · SHARE OF VOICE
How much of your category are AI agents handing to competitors?
Share of Voice quantifies your slice of every AI answer in your category - across surfaces, queries, and competitors. House of Zelena moved 22% → 56% on ChatGPT in 6 months. ITC MasterChef moved +89% AI Share of Voice in 30 days.
What is Share of Voice in AI search?
Share of Voice in AI search is the proportion of answers across a category in which your brand is named, measured against the competitors named alongside you. Where Citation Rank looks at one answer, Share of Voice looks at a whole topic - and tells you whether you are gaining or ceding ground in the questions that decide purchases.
Share of Voice is the percentage of category-relevant AI answers your brand appears inside - broken down by platform, query type, and competitor.
- →ITC MasterChef: +89% AI Share of Voice in 30 days
- →House of Zelena: #6 → #1 across 3 LLMs in 6 months
- →Continuous tracking against 8 named competitors per category
- →Citation source mapping for the answers you're winning and losing
How it works
Four capabilities, one signal
Category-level coverage
Pick your category (D2C beauty, CPG frozen, athleisure, etc.) - see the full set of queries an AI agent might be asked, and your share of the resulting answers.
Per-platform splits
Your SoV on ChatGPT versus Perplexity versus Gemini isn't the same. We break it out so you can target the surface with the highest gap.
Time-series tracking
Weekly trend lines per platform. See whether content, PR, or schema work is moving the number - or whether you're losing ground to a specific competitor.
Source attribution
Which sources are LLMs citing for the answer? Your blog? A review aggregator? A press feature? We surface the citation graph so you can replicate what's working.
The category, not the query
Buyers do not ask one question. They ask a dozen variations, on different engines, over several weeks, and the brands that keep appearing across that spread are the ones that end up on the shortlist. Share of Voice measures the spread rather than the instance.
That is why the composition of the prompt set matters more than the number itself. A Share of Voice figure is only as meaningful as the questions inside it - which is why every cluster in Tru Commerce is auditable: you can open it, see exactly which questions are counted, and change them.
Reading it against the right competitors
The competitors an engine names alongside you are frequently not the ones on your internal battlecard. In agentic commerce specifically, when engines answer category questions they overwhelmingly name platforms and payment networks - Shopify, Stripe, Mastercard, PayPal, Visa - rather than the venture-funded startup cohort most positioning decks are built against.
Share of Voice reports who actually appears, not who you expected. If the gap between those two lists is large, that is usually the most valuable thing the measurement will tell you, and it often redirects a roadmap.
What a healthy trend looks like
Share of Voice moves slowly and in steps rather than smoothly. A typical pattern: nothing for several weeks, then a cluster gain after a third-party source starts corroborating you, then a plateau. Week-to-week noise on a single cluster is normal; a sustained direction across several clusters is signal.
The most common misreading is treating a fall as a penalty. Usually it is dilution - a competitor entered the answer set and everyone's share compressed. The per-answer view tells you which of the two happened.
Case study
+89% AI Share of Voice in 30 days
Side by side
Citation Rank, Share of Voice and Visibility Score - which one answers your question
These three measure adjacent things and buyers routinely conflate them. The short version: Citation Rank is per-question, Share of Voice is per-category, Visibility Score is the roll-up. Most teams need the first two and report the third.
| Citation Rank | Share of Voice · | Visibility Score | |
|---|---|---|---|
| Question it answers | For this specific prompt, where did we appear? | Across the category, how much of the answer space is ours? | Overall, are we getting more or less visible? |
| Unit of measurement | One prompt, one engine, one region | A cluster of prompts, aggregated | All tracked prompts, rolled into one figure |
| Changes when | A single answer reorders | Competitors gain or lose ground across a topic | The aggregate moves - usually slowly |
| Best for | Diagnosing why a specific answer went wrong | Deciding which topics to invest in next | Reporting the trend to people who will not read prompt-level data |
| Who uses it | SEO and content leads | Category and growth leads | Executives and board reporting |
| Fails when | Used alone - one prompt is noisy | Read without knowing which prompts are in the cluster | Treated as a diagnosis rather than a summary |
| Mentions vs recommendations | Separated per answer | Separated per cluster | Blended - by design |
A single composite score cannot tell you what to fix. If a number moves and you cannot name the answer that moved it, you are looking at the wrong one of these three.
Go deeper
Related products and proof
Citation Rank
Drop from the category view into the individual answer.
Visibility Score
The single roll-up figure for reporting.
Why a single AI visibility score is a vanity metric
What a blended number hides.
Agentic commerce statistics 2026
Who AI engines cite and who they name, measured across 13,246 citations.
FAQ
Common questions
It is the proportion of tracked answers in a category where your brand is named, measured against every other brand named across the same answer set. Each cluster is auditable - you can see exactly which questions are counted, on which engines and in which regions.
Usually dilution rather than a penalty. When a new competitor starts appearing in the answer set, every existing brand's share compresses even if none of them lost ground in absolute terms. The per-answer view shows which of the two happened.
Enough to cover the real variation in how buyers ask, which in practice means a cluster rather than a handful. Too few and normal answer variance dominates; the figure moves and tells you nothing.
Yes. Engines diverge sharply in which sources they trust - far more than regions do - so a blended cross-engine figure can stay flat while your position on one engine collapses.
Citation Rank is your single position score across platforms. Share of Voice is the percentage of category answers you appear in. Citation Rank is 'where you stand'; SoV is 'how much of the conversation you own.'
We map 30-200 representative buyer queries to a category, validated against real search and LLM-conversation logs. You can refine the query set during onboarding.
Yes - name up to 8 competitors per category in Growth, unlimited in Enterprise.