AI Share of Voice: How to Measure It, Plus the Best Tools (2026)

AI share of voice is your brand's mentions divided by all brand mentions in AI answers to a fixed set of category prompts, calculated per engine. To measure it, build an unbranded prompt set, run it on a schedule across ChatGPT, Gemini, Perplexity and Google AI Mode, and count mentions and citations. Nudge AI Visibility does this at brand and product level.

This guide is for the ecommerce or growth lead who has been asked, "How much of the AI conversation in our category do we own?" It covers the formula, how to build the prompt set behind it, how to run a citation gap analysis, and seven tools compared. If you want the full list of AI search metrics first, start with our guide to AI search visibility metrics for ecommerce teams.

How do I measure share of voice in AI search for my product category?

You measure it with a fixed prompt set, repeated runs, and a simple ratio. There are two numbers worth tracking, and tools use the term for both, so be clear which one you are reading.

Share of voice (mention share) = your brand mentions ÷ total mentions of all tracked brands, across the same prompts, on one engine.

Visibility rate (response share) = AI answers that mention your brand ÷ total AI answers in the prompt set, on one engine.

Mention share tells you how much of the conversation you own against competitors. Visibility rate tells you how often you show up at all. Ahrefs, for example, defines share of voice as the percentage of AI responses in your topic set that mention or cite your brand versus competitors, which sits closer to the second number.

Here is a worked example with illustrative numbers. You run 50 unbranded prompts on ChatGPT. Across the 50 answers, your five tracked brands are mentioned 120 times in total, and you are mentioned 18 times in 15 different answers.

Three rules keep the number honest:

1. Calculate it per engine. ChatGPT, Gemini, Perplexity and Google AI Mode cite different sources and name different brands. A blended number hides where you are weak.

2. Exclude branded prompts. A prompt that names your brand will almost always mention you. Share of voice belongs on unbranded category prompts, where shoppers have not picked a brand yet.

3. Read the trend, not the day. AI answers vary between runs, so judge movement over 30 days or more.

Nudge AI Visibility runs a tracked set of topics and prompts across ChatGPT, Gemini, Perplexity and Google AI Mode. For each answer it records whether your brand was mentioned, which competitors were, and which sources were cited. Every prompt carries a branded or unbranded tag, so the category share is never inflated by prompts that name you.

How do I find out which prompts shoppers use in my category?

No AI platform publishes its prompt logs, so every prompt set is an informed sample. Build it from the language your shoppers already use, then check it against prompt research data.

1. Start from search queries. Pull question-style queries from Google Search Console ("best", "for", "vs", "under $") for your category pages. These are the closest public proxy for how shoppers phrase a need.

2. Mine your own customer language. Support tickets, product reviews and on-site search show the problems shoppers describe before they name a product ("running shoes for flat feet", "sunscreen that doesn't pill under makeup").

3. Add prompt research data. Semrush's AI Visibility Toolkit includes Prompt Research for topic volume and intent, Profound offers Prompt Volumes ("See what users are prompting AI Answer Engines"), Otterly.AI lists a Keywords to Prompts feature, and Ahrefs Brand Radar runs a database it describes as 551 million+ search-backed prompts.

4. Group prompts into topics. Six to ten topics of five to ten prompts each is a workable start. Mix intents: "best X for Y", "X vs Y", "is X worth it", and "where to buy X".

5. Tag every prompt branded or unbranded, and keep the branded ones out of your share of voice calculation.

6. Prune quarterly. Drop prompts no shopper would ask, and add prompts for new products or seasons.

Before you write prompts, it helps to know whether AI can read your products at all. Our free catalog scan checks how readable your product data is, which is the base every recommendation builds on.

In Nudge, the prompt corpus is built for your category and executed on a schedule, and the platform shows which prompts and categories matter and where your gaps are.

Which websites does ChatGPT cite for my category?

To see this, log the sources attached to every answer in your prompt set, then count them by domain and by page. A tool does this automatically. By hand, open each answer, copy the cited links into a sheet, and tally them after a week of runs.

You will usually find a few kinds of source, and Peec AI's own source categories (competitor, editorial, reference, UGC) are a useful starting set:

• Competitor sites, often their product and comparison pages.

• Editorial and review sites, such as roundups and buying guides.

• Reference sources, such as encyclopedic or spec pages.

• User-generated content, such as forums and community threads.

• Retailers and marketplaces that stock your category.

• Your own domain, if you are lucky.

Do this per engine. A domain that dominates ChatGPT answers in your category may barely appear in Perplexity. Our guide to AI platform citation patterns covers why the platforms differ and what citations mean for the pages shoppers land on.

Nudge reports citations at domain and page level for every tracked prompt. It also separates total citations of your own domain from Content Citations, which count only citations of pages Nudge created, so you can see what new work earned rather than what your existing product pages already had.

How do I run an AI citation gap analysis for an ecommerce brand?

A citation gap analysis shows the prompts where competitors are named or cited and you are not, and the sources that put them there. It turns a share of voice number into a work list.

1. Filter to lost prompts. From your unbranded set, keep the prompts where at least one tracked competitor is mentioned and you are not.

2. List the sources on those answers. Record every cited domain and page for each lost prompt, per engine.

3. Classify each source. Is it a competitor page, an editorial roundup, a forum thread, a retailer, or a reference page?

4. Check whether you could appear there. Is your product stocked by that retailer? Are you in that roundup? Does a thread discuss your product?

5. Pick a fix per gap. Off-site gaps call for PR, review or retailer work. On-site gaps call for a page that answers the prompt better than the one being cited, or for fixing product data the engine cannot read.

6. Re-measure. Re-run the same prompts after 30 days and compare share of voice and citation share for those prompts only.

Nudge runs this loop inside the platform. Gaps from prompt executions become reviewable opportunities (a page to build, a product to fix), and a person approves each one before anything ships. For product-data gaps, Nudge Catalog Enrichment scans every product for AI readability, verifies findings against the live product page, and generates a JSON-LD snippet for deployment.

Which are the best AI share of voice tools?

All seven tools below track brand mentions and cited sources across AI engines. They differ on engines, prompt limits, product-level detail, and whether they help you act on what they find.

Plans and prices were checked on each vendor's own site on November 11, 2026. They change often, so confirm before you buy.

1. Nudge AI Visibility

Nudge AI Visibility is built for consumer brands that sell products. It runs your tracked prompts across ChatGPT, Gemini, Perplexity and Google AI Mode and records brand mentions, competitor mentions and cited sources for each answer. Competitive benchmarking compares your mentions against a named competitor set across every prompt, which is the share of voice view.

The part generic trackers miss is product mentions: how often each of your products is named, per prompt and per platform. For a catalog brand, share of voice by SKU shows which products carry your category presence and which never come up. Agent Analytics adds the AI traffic hitting your store, classified by request type and broken down by page, which GA4 does not show.

Nudge also covers the step after the report. Opportunities turn gaps into reviewable work, Catalog Enrichment fixes the product data, and Nudge Funnels build pages for the shoppers who arrive from an AI answer, on your own domain.

Limit: Nudge is built for consumer brands selling online. If you run a B2B or non-commerce site, a general-purpose tracker will fit better.

2. Peec AI

Peec AI names share of voice directly: its plans track "visibility, position, sentiment, and share of voice" daily against competitors. Source analytics include citation share, and each source is classified as competitor, editorial, reference or UGC, which is handy for a citation gap analysis. Starter is $95 per month for 50 prompts, Pro $245 for 150, and Advanced $495 for 350, each with three active models; extra models are a paid add-on and Perplexity sits in the add-on list and Enterprise.

Limit for a DTC brand: It reports the gap well, but you take the fixes elsewhere. See Nudge vs Peec.

3. Otterly.AI

Otterly.AI lists Share of Voice KPIs and a Brand Visibility Index on every plan, plus citation source tracking and an AI prompt research tool. Pricing is now shown in euros: Lite is €29 per month for 16 prompts, Standard €189 for 120 prompts, and Premium €489 for 403 prompts. It covers seven engines on a credit system, where Claude queries cost three credits and others one.

Limit for a DTC brand: The plan cards say "4 of 7" engines while the comparison table implies all are trackable on credits, so confirm engine coverage with sales.

4. Ahrefs Brand Radar

Ahrefs Brand Radar defines share of voice as the percentage of AI responses in your topic set that mention or cite your brand versus competitors. It shows which pages and domains get cited within topics, across AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini and Copilot, with Claude available. Prompt tracking is free on paid Ahrefs Lite plans and above, and the AI Visibility Index starts at $199 per month.

Limit for a DTC brand: It is strongest for teams already in Ahrefs. The page does not describe product-level tracking.

5. Semrush AI Visibility Toolkit

The Semrush toolkit costs $99 per month with 25 tracked prompts, and 50 more cost $60 per month. Its Brand Performance report analyzes share of voice, sentiment and how AI positions competitors against you, and surfaces prompts where rivals are cited and you are not. Prompt Research adds topic volume, difficulty and intent. Semrush names Google AI Mode and ChatGPT as covered platforms, and there is no free trial.

Limit for a DTC brand: 25 prompts runs out fast across several product categories.

6. Profound

Profound tracks citations, sentiment, ranking and competitive presence across AI platforms. Its free trial runs 50 prompts daily for seven days on ChatGPT, Gemini and Google AI Overviews. Enterprise covers up to 10 engines, including Perplexity, Google AI Mode and Claude, with ChatGPT Shopping and full Prompt Volumes access. Enterprise pricing is custom.

Limit for a DTC brand: Beyond the trial, pricing is quote-only. See Nudge vs Profound.

7. Scrunch

Scrunch includes citation tracking on every plan, across ChatGPT, Claude, Gemini, Perplexity, Google AI Mode and AI Overviews, and Meta. Starter is $300 per month month to month ($250 billed annually) with 350 custom prompts and 1,000 industry prompts; Growth is $500 ($417 annually) with 700 custom prompts.

Limit for a DTC brand: Its pricing page does not mention share of voice by name, so check how competitor comparison is reported before you commit.

For a broader comparison of AI search tracking platforms, see our roundup of tools for ecommerce SEO teams.

How do you choose an AI share of voice tracker?

Match the tool to the decision you need to make.

• You need one benchmark number for leadership. Any tool here works. Check that it reports share of voice per engine and lets you exclude branded prompts.

• You need to know which products win or lose. Choose a tool with product-level mentions. This is where Nudge sits.

• You need a citation gap work list. Look for source classification and lost-prompt filtering (Peec AI, Semrush, Nudge).

• You need to act on the gaps. Choose a platform that connects findings to catalog fixes and new pages, or budget for the team that will do that work.

What do you do after you know your AI share of voice?

Share of voice is a state metric. It tells you where you stand today, not what caused a change, so do not present a rise as lift without a controlled test. The work that moves it is the citation gap list: better product data, better answers on your own domain, and presence on the sources engines already trust.

Then make sure the shoppers who do arrive from an AI answer land on a page that matches what they asked. Nudge covers both moments, the recommendation and the visit that follows. To see your category's share of voice on your own prompts, book a demo.