Written by: Mariana Fonseca, Editorial Team, AI Growth Agent
Key Takeaways
- AI share of voice numbers vary across vendors because each tool counts mentions, weights prominence, and measures visibility differently, so cross-vendor comparisons break down.
- Vendors like Profound, Otterly.ai, and HigherVisibility apply distinct formulas to the same brand data, so a 31% score in one dashboard can legitimately appear as 17% or 22% in another.
- The real question is which tool actually changes the number by producing, publishing, and self-healing content on an owned property.
- Monitoring-first tools leave the execution gap to the client. Closing that gap requires an autonomous engine that maps the full universe of buyer queries and acts on it without manual intervention.
- Traditional search tools show you where your brand stands. AI Growth Agent makes your brand the answer.
What Does AI Share Of Voice Mean?
AI share of voice is the proportion of AI-generated answers that mention or cite a brand. It is measured across surfaces like ChatGPT, Perplexity, Google's AI Mode, and Google AI Overviews, and expressed as a percentage relative to a defined competitor set.
That definition sounds clean. In practice, a tool is counting one or more of four distinct things:
- Mentions: whether the brand appears anywhere in an AI-generated answer, regardless of how many times or where.
- Weighted Prominence: where in the answer the brand appears and how substantively it is described, with first-paragraph recommendations scoring higher than trailing mentions.
- Answer-Level Visibility: the percentage of tracked prompts where the brand appears at all, treating each prompt-response pair as a binary yes or no.
- Impressions: how often the brand surfaces across AI responses, including Google's native Merchant Center definition, which calculates share of voice as a brand's AI impressions divided by total impressions across a defined competitor set.
Each of these is a legitimate measurement. None of them produces the same number.
Why AI Share Of Voice Is Not A Standardized Metric
Vendors count mentions, weight prominence, treat answer-level visibility, and pull impressions differently, so two dashboards can show different numbers for the same brand in the same week and both be internally consistent.
LLM Pulse's June 2026 guide states that no two tools agree on what "share of voice" means, and the same dataset produced three different numbers for the same brand: 20% mention-based, 16.8% position-weighted, and 31.4% citation-based. Each number is correct under its own definition. When a tool reports a single figure without disclosing which formula it used, the number cannot be reasoned about.
Three concrete examples from public vendor documentation show how differently the same label is applied:
- Profound calculates share of voice as the number of responses that mention your brand divided by the total number of all brand mentions across all responses. If a brand is mentioned 20 times out of 100 total brand mentions, its share of voice is 20%.
- Otterly.ai counts each day/prompt/AI engine execution as a binary 1 or 0 brand mention. Even if a brand is mentioned multiple times within a single response, it counts as one mention. The denominator is total brand mentions across the benchmark set, not total responses.
- HigherVisibility measures answer-level visibility: the number of tracked prompts where a brand appears divided by total prompts tested. This is a prompt-level binary, not a mention-share calculation.
These are three different metrics under the same label. Each vendor weights mentions, citations, and answer position differently, so the same brand can show a different score in two tools on the same prompts. Treat the number as vendor-specific and monitor its direction over time rather than comparing absolute scores across tools.
The non-comparability problem compounds across AI engines. A 2026 Writesonic study of 161,286 prompts found that ChatGPT, Gemini, Perplexity, and Google AI Overviews cited the same domains in only 3.8% of cases. A brand that wins ChatGPT can be invisible on Perplexity or Gemini, and an aggregated share of voice figure hides that story entirely.
How The 30% Rule Relates To AI Share Of Voice
Thresholds are another place where SOV numbers get misread. Vendors sometimes imply a number should clear 30%, but the 30% rule is not an official Google or AI guideline. The phrase is used informally to suggest AI should handle repetitive drafting or research tasks while humans remain responsible for reviewing, editing, fact-checking, and adding expertise. There is no universally accepted 30% rule for artificial intelligence.
A separate empirical finding uses similar language. LLMs disproportionately quote passages from the opening third of a page when answering prompts, meaning answer placement on the page is a controllable citation lever. That is a content-structure observation, not a governance rule.
When a vendor shows you a number and asks whether it clears a threshold, focus on what the number measures and whether the prompt set is representative of your actual buyer universe.
What 100% Share Of Voice Really Indicates
100% AI share of voice would mean every tracked AI response mentioned your brand and none mentioned a competitor. This is rare outside genuinely uncontested categories. Before treating it as real category-wide dominance, check whether the prompt set is too narrow, such as branded-only queries.
Google's own Merchant Center documentation states that if a Merchant Center account has no defined competitors available, the share of voice metric will display as 100%. That result is a data artifact and not a competitive signal.
AI share of voice only becomes fully meaningful once competitors are tracked. Without competitors as a baseline, share of voice is always 100% by definition. When a vendor shows you 100%, ask how the competitor set was defined and whether the prompt set includes the full universe of buyer queries, including non-branded and contested ones.
That same scrutiny applies when comparing tools. The next section breaks down how each major platform defines SOV and whether it stops at reporting.
AI Share Of Voice Tools Compared
The table below shows how seven major tools define share of voice and which engines they cover. Most of them focus on monitoring and reporting instead of publishing content on properties you own.
| Tool | Engines Tracked | SOV Methodology |
|---|---|---|
| Profound | ChatGPT, Claude, Gemini, Perplexity, Google AI Overviews (coverage varies by plan) | Frequency of brand mentions in AI-generated answers relative to competitors; share of voice equals brand mentions divided by total brand mentions across all responses |
| Peec AI | ChatGPT, Google AI Overviews, Google AI Mode, Perplexity, Gemini, Microsoft Copilot by default; Claude and others as upgrade models | Prompt-level tracking across major language models; multilingual competitor sets across markets |
| Otterly.AI | ChatGPT, Perplexity, AI Overviews, Gemini, Copilot | Binary mention count per day, prompt, and engine execution; multiple mentions in one response count as one |
| Semrush (AI Visibility Toolkit) | ChatGPT, Gemini, Perplexity, SearchGPT, Google AI Mode, Google AI Overviews; Claude only in the Enterprise AIO product | Mentions plus position within AI responses; Enterprise version also factors in topic search volume for ChatGPT |
| Ahrefs (Brand Radar) | Google AI Overviews, ChatGPT | Dual-index methodology tracking 466 million monthly organic prompts |
| AthenaHQ | 8+ major LLMs (self-serve tier lists nine engines; free tier limited to five) | Share of voice within GEO workflows; ACE scores content before publication |
| Scrunch AI (Now Sitecore) | Nine major AI platforms, including ChatGPT, Perplexity, Google AI Overviews, Gemini, Claude, Microsoft Copilot, Grok, Meta AI, and Google AI Mode | Tracking with prescriptive recommendations; Core starts at $250 per month for 125 prompts, five seats, and four LLMs |
| HubSpot (AI Search Grader) | ChatGPT, Perplexity, Gemini | Free entry-point benchmarking with brand visibility scores and share of voice |
The right tool depends on where you sit and how you plan to act on the data.
If You Already Run An SEO Suite: Semrush and Ahrefs add AI visibility as a feature inside the dashboard you already pay for. Semrush's AI Visibility Toolkit calculates AI share of voice based on both how many times a brand is mentioned and how high it appears in AI answers. Ahrefs Brand Radar tracks 466 million monthly organic prompts across Google AI Overviews and ChatGPT. Both wait for a human to drive every step.
If You Are Enterprise: Profound is the most visible name in the category, with $155M in funding and pricing from around $399 per month. It tracks brand visibility and citations across ChatGPT, Gemini, Perplexity, and AI Overviews, though engine coverage depends on the plan: Starter tracks only ChatGPT, Growth adds Perplexity and Google AI Overviews, and Claude and Gemini require the Enterprise tier. In 2026 it added an execution layer: a node-based agent builder with prebuilt templates and a background agent that proposes prioritized work. That layer is supervised workflow automation attached to an analytics dashboard, where the automation runs on workflows the client selects or assembles, and a human approves every publish.
If You Are Mid-Market: Peec AI targets mid-market brands and agencies with multilingual, multi-country tracking across 14+ languages. It builds competitor sets across markets rather than a single language, and its Starter plan starts at about €89 ($95) per month with a 7-day free trial. Otterly.AI is the cheapest entry at from $29 per month with a free trial, and covers ChatGPT, Perplexity, AI Overviews, Gemini, and Copilot. AthenaHQ tracks brand mentions across 8+ major LLMs, with its self-serve tier listing nine engines (including ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Gemini, Claude, Microsoft Copilot, Grok, and DeepSeek) and its free tier limited to five, wrapping share of voice inside GEO workflows built for teams, from around $295 per month.
If You Are An Agency: Scrunch AI (now Sitecore) tracks nine major AI platforms, including ChatGPT, Perplexity, Google AI Overviews, and Gemini, among others such as Claude, Microsoft Copilot, Grok, Meta AI, and Google AI Mode, and combines tracking with prescriptive recommendations, with Core starting at $250 per month for 125 prompts, five seats, and four LLMs. HubSpot offers a free AI Search Grader for entry-point benchmarking across primary answer engines.
Traditional search tools show you where your brand stands. AI-focused monitoring tools extend that view into AI answers, but they still rely on your team to execute.
How To Evaluate An AI Share Of Voice Tool Before You Buy
Every vendor will show you an impressive number. These five questions reveal whether that number can be reasoned about and whether the tool will actually move it.
- How Is SOV Computed? Ask whether the tool counts mentions, weights prominence, measures answer-level visibility, or pulls impressions. If a tool reports a single AI SOV figure without telling you which formula it used, you cannot reason about it. When the vendor cannot explain the formula in one sentence, the number cannot be trusted.
- Which Engines Are Covered? Ask which AI surfaces the tool tracks: ChatGPT, Perplexity, Google AI Mode, Google AI Overviews, Gemini, Claude, Copilot. Ask whether the tool scrapes the real interface or relies on API responses. API-mode answers had at least one meaningful drift versus the real chatgpt.com interface on 96% of prompts tested across 1,000 queries.
- Is The Number Comparable Week Over Week? Ask whether the prompt set is locked. A panel that changes week to week produces a trend line that measures your prompt-writing, not your visibility.
- Is Prompt Count Metered Or Unlimited? Ask how many prompts are included and what happens when you exceed the limit. Monitoring tools often cap clients at a small set of tracked prompts, which means they only ever see the slice of their market they already thought to ask about.
- Does The Tool Only Report Or Also Execute? Ask whether the tool produces content, publishes on a site you own, and self-heals it over time. When the answer is that it hands you a list of recommendations, you are buying a rearview mirror.
The Execution Gap: Measurement Without Action
AI share of voice functions as a diagnostic. Monitoring-first tools tell you where you stand and hand the work back. Most teams over-buy on dashboards and under-buy on execution. The 2026 action layers monitoring tools added still leave the client to review, publish, and maintain everything. A rearview mirror with a to-do list taped to it is still a rearview mirror.
Closing the loop requires four steps: map the full universe of seed terms and long-tail queries, produce authoritative content, publish on an owned property, and self-heal it over time. Execution engines that ship fixes typically take 2 to 8 weeks per fix to register measurable movement. Because that lag is built in, the sooner the engine is running, the sooner the number changes.
Stop letting AI define your brand at random. Control the narrative across online search with an engine that can act on what the dashboards reveal.
AI Growth Agent: The Engine That Changes The Number
AI Growth Agent is the autonomous engine that maps a brand's universe across online search and wins it on autopilot. It is built for mid-market and enterprise companies that already have an identity and now need to control the narrative around it.
The architecture follows a different model from the monitoring-first tools in the comparison table above. Content creation sits at the core of the business as the primary product. The engine maps, writes, publishes, and self-heals on a site the client owns. The four pillars that feed the system are:
- Search Intelligence: a complete portrait of the traditional search landscape, covering positioning, competition, and search volume, taken from raw situation to an actionable diagnosis.
- AI Analytics: brand value and consumer behavior across the whole journey, from external touchpoints like Google and AI-tool queries through content consumption, demographics, and sentiment.
- Bot Tracking: every bot interaction, traditional crawlers and AI training agents alike, including every crawl, citation, and training sweep.
- AI Ranking: where the brand appears in AI answers and how that position evolves week over week, which becomes the new leaderboard in a world without static ranked lists.
The full universe is refreshed weekly. Prompt count is never a billed metric. Clients see their entire universe instead of a capped handful of tracked terms. The engine operates at Level 4 autonomy: it creates plans, executes them, handles its own errors, and alerts a human only when it hits a roadblock it cannot resolve. The human manages by exception while the engine drives the work.
Across the first twelve weeks, clients average more than 12,000 additional AI citations and mentions, over 100,000 additional bot visits, and a 20%+ lift in impressions. Content indexes in as little as ten days, and the first article goes live within a week. Reporting isolates exactly what AI Growth Agent generated, separate from the visibility the brand already had, so the number is defensible every week.
What To Do With An AI Share Of Voice Number Once You Have It
The number functions as a diagnostic. Use it to find where you are not winning, then orient an execution engine toward that white space. A brand might dominate "best [category]" comparison queries but underperform on "how to" educational prompts, and that insight directly guides content strategy.
Report AI share of voice per engine rather than aggregated, because an aggregated figure hides the real story. As noted earlier, engine-level visibility diverges sharply, so report per engine. The action is content that maps the universe, publishes on an owned property, and self-heals over time.
Frequently Asked Questions
How Do I Measure AI Share Of Voice?
Define a prompt set that covers the buyer journey, run each prompt across the AI engines your buyers use, and count how often your brand appears relative to competitors. The formula depends on what you are counting: mentions, weighted prominence, answer-level visibility, or impressions. Lock the prompt set before you start, or month-over-month comparisons are meaningless. Most categories need a minimum of 50 prompts for a defensible measurement, with 100 to 200 as the practical sweet spot. Run each prompt multiple times per engine, because AI outputs are non-deterministic and a single run reflects sampling noise rather than a stable signal.
Why Do Vendor Numbers Differ?
Vendors count mentions, weight prominence, treat answer-level visibility, and pull impressions differently. As explained above, vendors define SOV differently, so a 30-point gap between tools can be legitimate. The five most common sources of discrepancy are a different prompt panel, a different model tier, a different sample count, a different mention definition, and unweighted versus weighted scoring.
What Engines Should Be Tracked?
Track the engines your buyers actually use: ChatGPT, Perplexity, Google AI Mode, Google AI Overviews, Gemini, Claude, and Copilot. Report AI share of voice per engine rather than aggregated, because an aggregated figure hides the real story. Engine behavior differs significantly. Perplexity and Google AI Mode are retrieval-heavy and cite many sources per answer, while Gemini's standalone app synthesizes answers with few or no visible source links, making brand mention and sentiment metrics more relevant than citation counts for that surface.
Is A Monitoring Tool Enough?
A monitoring tool tells you where you stand. It does not change where you stand. When the tool hands you a list of recommendations and leaves you to produce, publish, and maintain the content, you are buying a rearview mirror. The 2026 action layers monitoring-first tools added still require a human to review, approve, and maintain everything. Closing the loop requires an execution engine that maps the universe, produces authoritative content, publishes on a site you own, and self-heals it over time.
How Fast Can An Execution Engine Move?
AI Growth Agent goes from kickoff to the first published article in about one week, with content indexing in as little as ten days. As noted in the product section, clients see measurable movement within the first twelve weeks. The standard engagement is a three-month pilot, because indexing takes time and varies by industry, but clients see movement early. The brands cited in AI search this year are training the next generation of models with their own story.
Conclusion
Every AI share of voice tool measures differently, so the numbers are not comparable across vendors. The real question is which tool actually changes the number. Monitoring-first tools tell you where you stand and hand the work back. The 2026 action layers they added still leave the client to review, publish, and maintain everything. Closing the loop requires an engine that maps your universe, produces authoritative content, publishes on a site you own, and self-heals it over time.
Traditional search tools show you where your brand stands. AI Growth Agent makes your brand the answer. Find out if AI Growth Agent is a fit for your team and see how fast your first article can go live.
