AI Share of Voice: Calculation, Benchmark, Lift

By Matt Griffin, founder of Formative Digital. Brantford, Ontario. Published 2026-04-26. 2,300 words.

What AI Share of Voice actually measures

AI Share of Voice quantifies your brand's competitive position within AI engine answers. It is the most actionable single metric in an AI visibility program because it captures both presence (your Mention Rate) and competitive context (whether your competitors are taking the share you are not getting).

The metric matters because a brand that holds steady at 25% SOV while a competitor moves from 15% to 35% is losing the race even if the absolute SOV number is unchanged. Movement against named competitors is the actionable signal; raw SOV without competitive context misses the point.

How to calculate it (worked example)

Step-by-step example for a Brantford-area accountant calculating SOV against ChatGPT for accounting-related queries.

Step 1: Build the prompt battery. 30 prompts covering branded ("what do you know about [your firm]"), category ("best accountant Brantford"), comparison ("[your firm] vs [competitor]"), problem-intent ("I need help with bookkeeping in Brantford, who do I call").

Step 2: Run the prompts. Private browsing window, ChatGPT, capture the response for each prompt. Save responses with date, version, and any personalization context.

Step 3: Count entity mentions. For each response, identify every named brand or business entity mentioned. Tally per brand. Aggregate across all 30 responses.

Step 4: Apply the formula.

• Total entity mentions across 30 responses: 240
• Your brand appears: 18 times
• SOV = 18 / 240 × 100 = 7.5%

Repeat for each engine (Perplexity, Gemini, AI Overviews) and aggregate by engine importance to your audience.

Benchmarks by company stage

Numbers vary by category but the pattern we see across client audits is consistent.

• New entrant (0 to 12 months in market): 0 to 5% SOV. Brand has minimal trained-knowledge representation and no third-party citation footprint.
• Established small business (1 to 3 years, regional): 5 to 15% SOV. Brand mention on long-tail queries; rarely the primary recommendation on category-defining prompts.
• Mid-market established (3+ years, multi-region): 15 to 30% SOV. Consistent mention across category prompts; cited on long-tail and most mid-tail queries.
• Category leader: 35 to 60% SOV. Named in nearly every category-relevant prompt; cited as primary source on commercial-intent queries.
• Monopoly position (rare): 60%+ SOV. Brand is so dominant in its category that AI engines treat it as the default answer.

The Mattress Miracle program documented at our case studies page moved from sub-5% SOV in mattress-retail prompts to estimated 30 to 40% within 12 months.

How AI SOV differs from classical Share of Voice

Classical SOV (the term used in PR, advertising, and pre-AI marketing) measures brand mention frequency in earned media, advertising impressions, or organic search results. AI SOV measures brand mention frequency inside AI engine responses.

Three meaningful differences.

Different surface. Classical SOV captures earned and paid media. AI SOV captures the synthesized layer where buyers research before clicking through to anything.

Different competitive set. Brands that compete in classical SOV (regional advertisers, PR rivals) may not be the same brands that compete in AI SOV (which can include national chains AI engines suggest by default plus information sources like industry directories).

Different actionability. Classical SOV is moved by ad spend and PR investment. AI SOV is moved by content discipline, schema, third-party citations, and entity grounding. The levers do not overlap fully.

Per-engine vs aggregate SOV

SOV is engine-specific by default. Calculate it per engine; aggregate weighted by engine importance to your specific audience.

Typical weighting for a B2C local business in 2026:

• Google AI Overviews: 50% (largest distribution)
• ChatGPT: 25%
• Perplexity: 10%
• Gemini: 8%
• Microsoft Copilot: 5%
• Apple Intelligence: 2%

Typical weighting for a B2B SaaS company in 2026:

• ChatGPT: 35%
• Google AI Overviews: 25%
• Claude: 15% (B2B audience skews Claude-heavy)
• Perplexity: 15% (research-driven buyers)
• Gemini: 7%
• Copilot: 3%

Audit your audience's engine preferences (GA4 referrer data, customer interviews, industry surveys) and weight accordingly.

Weighting prompts by importance

Not every prompt in your battery has equal commercial value. A simple weighting scheme captures this.

Weight 3 (high commercial intent): "best [category] near [city]," "how much does [your service] cost," "[your service] for [specific use case]." These prompts directly drive purchase decisions.

Weight 2 (moderate intent): "what is [topic in your category]," "how does [your product type] work," "compare [your category options]." Educational but pre-purchase.

Weight 1 (low direct intent): "history of [topic]," "general [category] information." Brand-awareness layer, not direct conversion driver.

Weighted SOV = Sum (mentions × weight) / Sum (total entity mentions × weight) × 100. The weighted number better reflects revenue-impacting visibility than raw SOV.

The 4-step lift program

Same four moves apply across most categories. Detail at Brand Visibility in AI and ChatGPT.

1. Anchor brand in Wikidata. Single highest-leverage move. Propagates across ChatGPT, Perplexity, Gemini, Apple Intelligence, AI Overviews. One-time effort, propagation 2-8 weeks for Knowledge Graph and quarters for trained-knowledge corpora.

2. Publish honest comparison content. Take back the framing your competitors control. Schema each comparison page with FAQPage, lead with 40-60 word direct answer.

3. Earn third-party citations. 85% of brand mentions originate third-party. HARO/Connectively, podcast guest appearances, Reddit and YouTube participation in your industry.

4. Deploy connected JSON-LD schema. Article + Person + Organization + FAQPage + HowTo on every cornerstone in a connected @graph. ~40% citation lift versus pages without.

Realistic SOV movement: +3 to +8 points in first 90 days (Perplexity and AI Overviews), +10 to +25 points across full year (compound across all engines).

Tools that calculate it for you

Manual calculation is sustainable for one brand and one engine. Past three engines or five competitors, automated tools save material time.

Tools with native SOV calculation:

• Profound, AthenaHQ, BrandRank.AI: Multi-engine SOV with competitor breakdown.
• Otterly: SOV with engine-by-engine view.
• Peec.ai: SOV plus source-attribution detail.
• Ahrefs Brand Radar: SOV-equivalent metric across 11 indices.
• TrackAIMentions: Entry-tier SOV for ChatGPT, up to 3 competitors.
• HubSpot AEO Grader: Free starter SOV check.

Full tool comparison and pricing at Best ChatGPT SEO Tools 2026.

For the broader brand visibility framework, see Brand Visibility in AI and ChatGPT. For the methodology guide on AI visibility tracking, see AI Visibility Tracking. For the underlying GEO discipline, see The 12 Vectors. For our team to build the prompt battery and track SOV across engines, see Formative Digital services.