How to Measure Share of Voice Across ChatGPT, Gemini, and Perplexity
How to Measure Share of Voice Across ChatGPT, Gemini, and Perplexity
Share of voice in AI search is the percentage of answers in which your brand is mentioned, recommended, or cited when people ask questions related to your products or services across ChatGPT, Gemini, and Perplexity. Measuring it shows how visible your brand is in the answers customers read before they buy.
This guide explains how to build a repeatable measurement system, score the results fairly, and turn the numbers into actions.
Why Share of Voice Matters More Than Rankings Now
Classic SEO tracks where your pages rank in ten blue links. AI search returns a single synthesized answer that usually names only a handful of brands, so being absent from an AI answer means you were never considered at all.
Share of voice captures that reality. It tells you how often you are part of the shortlist, which competitors own the conversation, and which topics you are losing. For businesses investing in GEO and AI search optimization, share of voice is the metric that proves whether the work is paying off.
Step 1: Build a Prompt Set That Reflects Real Buyer Questions
A prompt set of 30 to 100 questions is enough for most businesses. Include four types of prompts:
Direct comparison prompts: “What are the best SEO agencies in Pakistan?” or “Compare the top website design companies for small businesses.”
Problem prompts: “How do I improve my Google Business Profile ranking?” These reveal whether AI tools recommend you as a solution.
Category prompts: “Who offers guest posting services?” These show your standing in the broader market.
Branded prompts: “Is MSJ Digital a good SEO agency?” These test how AI tools describe you to people who already know your name.
Write prompts the way customers actually type them, including local phrasing and industry terms. Avoid prompts that mention your brand unless you are testing branded visibility on purpose, because a prompt that names you will inflate your score. Keep the same prompt set every month so changes reflect real movement.
Step 2: Run the Same Prompts on All Three Platforms
Each platform answers differently, so use a consistent routine for each:
ChatGPT: Run prompts in a fresh session where possible, and note whether the answer used web search or came from training data. Record every brand named and its order.
Gemini: Gemini leans on Google’s ecosystem, including Business Profiles, reviews, and Maps data. Record whether your business appears in the answer text, in a linked source, or in a local listing card.
Perplexity: Every answer cites sources, so record both mentions in the answer text and citations to your website, since a citation without a mention still drives traffic.
Run the full set at least once a month, on the same day if possible. AI answers change with model updates and fresh web data, so a single day’s snapshot is more comparable than results gathered across a week.
Step 3: Score Mentions Fairly
A simple mention count is a start, but not all mentions are equal. Score them so your share of voice reflects quality as well as quantity:
Mentioned by name: 1 point. Your brand appears in the answer.
Recommended: 2 points. The answer presents your brand as a good choice, not just one option among many.
Top position: 3 points. Your brand is listed first or described as the best option.
Cited source: 1 extra point. The answer links to your website or cites your content.
Negative framing: subtract 1 point. A mention that warns against you or reports bad reviews is not a win.
Calculate two numbers per platform: raw share of voice (answers mentioning you divided by total prompts) and weighted share of voice (your points divided by all brands’ points). Track the weighted number month over month, since it rewards recommendations over passing mentions.
Step 4: Track the Metrics That Explain Movement
Share of voice alone does not tell you why it moved. Track these supporting metrics alongside it:
Topic coverage: Which prompt categories do you win, and which do you lose? A business can dominate “local SEO” prompts while being invisible for “website design” prompts.
Sentiment: Are your mentions positive, neutral, or negative? AI tools repeat what they find online, so a drop in sentiment usually traces back to reviews or outdated information.
Source diversity: Which websites do AI tools cite when they mention you? If all citations come from your own site, you need third-party mentions to strengthen your position.
Referral traffic: Check analytics for visits from chatgpt.com, perplexity.ai, and Gemini. Rising AI referral traffic usually follows rising share of voice.
When share of voice drops, these metrics point to the cause: a competitor’s strong comparison page means a content problem, while outdated business information means a listings problem.
Step 5: Turn the Data Into Action
Each month, pick the three biggest gaps and act on them:
If you are absent from comparison prompts: Publish comparison-style content and earn mentions on independent review and directory pages, which AI tools trust more than self-promotion.
If you are mentioned but never recommended: Strengthen proof with case studies, client results, reviews, and clear service pages. AI tools recommend brands they can describe confidently.
If one platform lags the others: Fix platform-specific gaps: weak Gemini visibility often means incomplete Business Profile data, weak Perplexity visibility means thin third-party coverage, and weak ChatGPT visibility means content that does not answer the exact questions people ask.
Small businesses often see the fastest gains from fixing basics: consistent business details everywhere, a complete Google Business Profile, and a handful of genuinely useful pages that answer real customer questions. These foundations also support local SEO and traditional search rankings, so the work pays twice.
Common Mistakes When Measuring AI Share of Voice
Avoid these errors, which make the numbers misleading:
Testing only branded prompts: This inflates your score and hides the fact that new customers never see you.
Changing the prompt set every month: You cannot compare scores if the questions keep changing. Freeze the core set and add new prompts separately.
Ignoring negative mentions: A brand mentioned as a warning still counts in a raw count. Score sentiment, or your share of voice will flatter you.
Measuring once and stopping: AI answers shift with every model update. Monthly tracking is the minimum for a reliable trend.
How Often Should You Measure?
Monthly measurement suits most businesses. Weekly tracking is useful during a campaign, such as after publishing a major content series, because it shows whether the work moved the needle quickly. Daily tracking is usually noise: AI answers vary from session to session. Whatever cadence you choose, keep the method identical so the trend is trustworthy.
Start Measuring This Week
Begin with 30 prompts, run them on all three platforms, and record every mention in a spreadsheet. That first snapshot is your baseline. Next month, run the same prompts again and compare — the gaps will show exactly where to focus.
If you want help building your prompt set or turning AI visibility into customers, explore our GEO and AI search services and get a visibility audit today.
Leave A Reply Now