What Is AI Share of Voice? How to Measure It Each Week

AI share of voice is the share of buyer-style prompts in which AI assistants like ChatGPT, Perplexity or Claude name your product. Ask ten buyer questions, get named in two, and your score is two in ten. No dashboard is required, just a fixed prompt list and a weekly check.

If you're a solo founder wondering whether ChatGPT knows you exist, this is the number to watch.

Key Takeaways

• The metric is a simple fraction: prompts that name you divided by prompts you tracked.
• Track each assistant separately, because a blended score hides where you're missing.
• Traditional share of voice counts ads, media and search, while the AI version counts named answers.
• A fixed prompt list, fresh sessions and a spreadsheet are enough to measure it weekly.
• Changing prompts mid-month or skipping claim checks will skew your results.

What Is AI Share of Voice?

AI share of voice is the share of tracked buyer prompts in which AI assistants such as ChatGPT, Perplexity or Google AI Overviews name your product. It tells you whether you're on the shortlist when a buyer asks an assistant for help. It's a fraction, not a ranking.

The formula fits on one line:

Prompts that name you ÷ prompts tracked

A citation is an assistant naming or linking your product in its answer. Every citation counts toward the top number. Every prompt you run counts toward the bottom.

Keep the prompt set fixed. Otherwise the fraction means nothing. The score is one measurable piece of brand visibility in AI search.

Which AI Assistants Should You Track?

Track each assistant separately. Answers differ by tool, and one blended score hides which assistant names you and which ignores you. Start with the five that most buyers reach for, check each by hand, and keep a separate column for each one in your log.

To check each one by hand:

• ChatGPT: open a new chat and paste the prompt as written.
• Perplexity: run the prompt and note the sources it lists.
• Google AI Overviews: search the prompt and read the summary at the top. See Google's guidance on AI features in Search for how these surface content.
• Gemini: start a fresh conversation for each prompt.
• Claude: same routine, new chat every time.

Answers change between sessions and locations. One check is a snapshot, not a verdict.

Picture a founder named in Perplexity but never in Claude. That split is useful. Perplexity reads the live web and found that founder's pages. Claude may lean on other sources. Now the founder knows where the gap is. If running this by hand every week gets old, Draftful's weekly AI citation tracking runs the same checks for you.

What Do Example AI Share of Voice Scores Look Like?

Scores for a small B2B invoicing tool usually fall into a few patterns. The three below show how the fraction reads in practice. Every number here is illustrative, not measured data. They exist to show the math and what each pattern tells you to do next.

1. Brand-new product. Prompt: "best invoicing tool for freelancers." Named in 0 of 10 prompts. Score: 0/10. You're invisible, which is normal on day one.
2. Named, but wrong. Prompt: "cheap invoicing software for agencies." Named in 4 of 10 prompts, but two answers quote an old price. Score: 4/10, with two answers needing fixes.
3. Strong in one place. Prompt: "invoicing app with recurring billing." Named in 7 of 10 prompts on Perplexity, 0 of 10 on Claude. Two very different scores.

How Is AI Share of Voice Different From Traditional Share of Voice?

Traditional share of voice measures your slice of ad spend, media mentions or search visibility against competitors. The AI version measures how often an assistant names you in answers to buyer prompts. One counts where you show up. The other counts whether a machine recommends you.

• What is counted: traditional counts ads, press or rankings. AI counts named answers.
• Where data comes from: traditional uses paid trackers. AI uses your own prompt runs.
• How often you check: traditional is often monthly. AI works best weekly.
• What it tells you: traditional shows exposure. AI shows whether you make the shortlist.

The honest contrast for a solo founder is cost. Paid share-of-voice trackers usually sit above a bootstrapped budget. The AI version needs a fixed prompt list and an hour of your time.

A founder can rank on page one of Google and still be missing from every AI answer a buyer reads.

How Do You Measure AI Share of Voice Each Week?

Each week, run the same fixed prompt list through each assistant. Record which products get named. Then divide the number of prompts that named you by the number of prompts you ran. That's your score for the week. No analytics tool is needed, just consistency and a spreadsheet.

1. Write your prompt list once, using buyer-style questions.
2. Open each prompt in a fresh session.
3. Note every product named and whether a link appears.
4. Calculate the fraction for each assistant.
5. Log it in a spreadsheet with the date.
6. Compare this week to last week.

Answers change, so a trend over several weeks matters more than any single week. When the hour stops being worth it, compare the Starter, Growth and Agency plan prices against your own time.

What Mistakes Skew Your AI Share of Voice Results?

Three errors do most of the damage. Changing the prompt list between weeks. Counting a mention without checking whether the claim inside it is true. Testing only your own brand name, which tells you nothing about whether buyers find you. Each one makes the number look better or worse than it is.

Here's how it happens. Say a founder rewrites three prompts mid-month. The score jumps, and it looks like progress. But the test changed, not the visibility.

A named product can still carry a wrong price or a feature you dropped last year. Check each claim before you act on it. The same habit applies to your own content, which is why we built a gate for checking draft claims before you publish. Nothing ships unsourced.

Start small. Run your free AI citation baseline on Draftful's homepage to see whether AI assistants already name your product before you spend a dollar on content.