AI visibility
AI visibility Nine AI engines, tracked Prompt research Find the questions worth tracking Competitor analysis Who wins on Google and in AI answersSEO
SEO tools Rank tracking, backlinks, and more Keyword research Volume, difficulty, and intent Site audits A health score and ranked fixes Local SEO Map pack grids and reviewsWorkflow
Closed-loop marketing From finding to task to report Ask Doily Answers and tasks from your data Reporting Branded reports that send themselves
The Doily team Team Reports for AI visibility keep quoting a "share of voice" figure. What does it measure, and how do you check that a number you are shown is calculated in a way that means something?
The Doily team Team Share of voice in AI answers measures how often your brand appears compared with the other brands tracked for the same prompts. There are 2 common ways to calculate it, and tools do not always say which one they use.
Answer coverage. Answers that mention you, divided by all answers collected. If you ran 100 answers and were named in 20, coverage is 20%.
Mention share. Your mentions divided by the mentions of every tracked brand. If you were named 20 times, competitor A 50 times, and competitor B 30 times, mention share is 20 out of 100, or 20%.
The 2 figures are equal in that example by coincidence. They drift apart when answers name several brands at once, and mention share is the one that depends on who you chose to track.
Whatever the formula, the number is only comparable under the same conditions:
Read it as a trend per engine and per prompt group. A move from 20% to 28% on category prompts on one engine tells you something. A single blended figure across 9 engines mostly hides what is happening.
Pair it with citations. Share of voice says how often you appear. Cited sources say why, and they show you which pages to improve.
The Doily team Team A worked example, with made up numbers.
Say you collect 20 answers. Your brand is named in 8 of them, so your mention rate is 40%. A competitor is named in 12, a mention rate of 60%. Mention rates across competitors can add up to more than 100%, because one answer can name several brands.
Now count every brand mention across all 20 answers. Say there are 50 in total and 8 are yours. Your share of mentions is 16%, and the shares for all brands add up to 100%. The numbers are invented to show the arithmetic, not to suggest a benchmark.
The Doily team Team Report each engine separately.
A blended average across engines can hide a big gap. Google's AI Overviews and AI Mode are a good example. Ahrefs compared 540,000 pairs of responses to the same queries and found the two features cited the same URLs only 13.7% of the time, even though the answers were 86% similar in meaning.
If you track one and assume the other follows, you will misread your position.
Your pages sit on page one for the searches that matter. Then you ask ChatGPT for recommendations in your category and it names other businesses, never yours. It looks like a contradiction. What explains it, and what…
AI engines can only cite what they can fetch. If a firewall or a robots.txt rule is turning their crawlers away, you could be missing for a reason that has nothing to do with your content. How do you check?
llms.txt is a plain text file at the root of a site that lists key pages for AI systems. Plenty of advice says to add one. Does it change whether AI engines mention or cite you?
