Share of Voice in AI
Why it matters
A search ranking shows where a page sits on a list. An AI answer often gives only a few names, so a business is either in the answer or invisible. A buyer asking "which tool should I choose" may never see page two.
Without a measure, nobody notices which questions a competitor wins. A number turns a vague feeling of falling behind into a target and shows which prompts to work on first.
It also shows whether effort pays off. Content, reviews and links that build topical authority should, over months, raise the share. If the share stays flat after real work, the cause is somewhere else, and the prompt-by-prompt data tells you where to look.
How to apply it
- Write 30 to 100 prompts that mirror real buying questions at each stage, from "what is" to "which tool should I choose".
- Run the same prompts on a schedule across the engines that matter. Answers vary from run to run, so repeat each prompt and look at the trend, not a single result.
- Record every mention, including ones without a link. A brand named in the text is still present.
- Run the same prompts for named competitors, so the number has a benchmark.
- List every prompt where a competitor appears and the business does not, and turn each into a content or authority task.
What it is
Buyers now ask ChatGPT, Perplexity, Gemini or Google's AI Overviews for recommendations. Share of voice in AI measures how often a brand appears in those answers compared with its competitors. The basic sum is simple: run a fixed list of prompts, count the answers that name the brand, and divide by the total. A stricter version counts only answers that cite the brand's own pages. A broader version compares mentions of the brand with mentions of all named competitors.
Common mistakes
- Changing the prompt list between runs, which makes trends meaningless.
- Reading one run as the truth.
- Counting only links and missing plain mentions.
