Meet an Expert

Why AI Share of Voice Differs Across Visibility Trackers

Three AI visibility trackers calculate share of voice three ways. Here is what each formula counts and why one brand's numbers differ across tools. Here is what each formula counts and why the numbers

Gilad Bechar Gilad Bechar
Nov 6, 2026 7 min read Tools

A marketing lead pulls AI share of voice for the same brand, in the same month, from three AI visibility trackers: Brand Radar, a mention-share tracker and a presence tracker. She gets three different numbers. None of the tools is broken. Each one divides a different count by a different total, runs a different set of prompts and weights those prompts differently. “Share of voice” is one name for several different metrics. If you compare them without reading the formulas, you get an argument that nobody can settle.

Key Takeaways

  • The mention-share tracker divides the responses that mention your brand by all brand mentions. The presence tracker’s Share of Voice splits a single pie across your chosen competitors. Brand Radar weights each mention by demand estimated from Google search volume.
  • The prompt set changes the result as much as the formula does. The mention-share and presence trackers run prompts you choose, while Brand Radar’s index runs prompts built from search data.
  • Share of voice inside AI answers counts mentions. Being cited as a source is tracked separately, and a brand can lead on one while trailing on the other.

What the Mention-Share Tracker Puts on Top of the Fraction and What It Puts Underneath

The mention-share tracker reports two numbers that people often mix up. Visibility Score is the percentage of tracked responses that mention your brand. It is an absolute rate, worked out for each brand on its own.

Share of Voice works differently. Its definition divides the number of responses that mention your brand by the total number of brand mentions across all responses. The total counts every brand that shows up in an answer. In a category where answers list long rows of vendors, every brand’s share drops, including the leader’s. In a category where answers name two or three brands, the same level of presence gives a much higher share.

Two settings change the result before any maths happens. The tracker runs prompts daily against the consumer versions of each engine, so the answers it records are the ones users see. Its Name Matching setting groups different versions of a brand name under one asset. If you leave a common variant out, the share is too low before any calculation starts.

The Presence Tracker Has Two Numbers That Answer to the Name

The presence tracker describes AI share of voice as competitive presence: the percentage of AI responses that mention your brand compared with competitors across tracked prompts. In its metrics reference, that rate is now called Mentions Rate (formerly Brand Presence): responses naming the brand, divided by total responses. Each brand’s rate is calculated on its own, so the rates for a brand and its competitors can add up to more than the whole.

The same tool also has a separate metric named Share of Voice. It splits one pie among the active competitive set, so the slices for all tracked brands add up to the whole. Add a competitor and every other brand’s slice shrinks, even though nothing changed in the AI answers. Remove one and the slices grow.

~49%
mention rate for Moburst’s own brand in AI answers, alongside 42,435 total AI citations and number 1 in category.See the case study.

This tracker counts presence as yes or no. A response counts once however many times it names the brand, and it only counts explicit mentions in the answer text. Citations go into their own metric, Citations Share of Voice: responses that cite a source you own, divided by responses that cite any source at all. It also lets you filter by persona, funnel stage, region and branded versus non-branded prompts. Each of those filters gives you a different share, all correct.

Brand Radar Weights Every Prompt by Search Demand

Brand Radar starts from impressions. Its help documentation defines AI Share of Voice as a brand’s percentage share of impressions compared with other tracked brands. An impression is a mention or citation weighted by an estimate the tool calls AI-adjusted volume.

That estimate takes the Google search volume of a prompt’s parent keyword and multiplies it by how much each AI platform is used. The usage estimates come from the vendor’s traffic data on AI platforms compared with Google organic traffic. When you look at several platforms together, the tool takes a weighted average, so platforms with more impressions count for more.

The prompts also come from a different place. The Brand Radar index builds its questions from Google’s People Also Ask data and a large keyword database, refreshes them monthly and reports over a rolling window. The vendor states plainly that the result models potential visibility, and that it does not claim a validated link between Google search volume and how often a question is asked inside an AI tool. Its help centre also warns that the metric can change a lot depending on how you set up brand and competitor entities.

A share of voice figure is a fraction, and nobody can read a fraction without knowing who chose its denominator.

Why Does the Same Brand Score Differently in Each Tool?

There are six places where the numbers come apart:

  • The unit being counted: responses (the presence tracker’s Mentions Rate), brand mentions (the mention-share tracker), or demand-weighted impressions (Brand Radar).
  • The total it is divided by: all responses, all brand mentions, or only the competitors you configured.
  • Weighting. The mention-share and presence trackers count every prompt equally. Brand Radar gives prompts with high search volume much more weight.
  • Who wrote the prompts.
  • Platform mix and how often data is refreshed: daily runs in one tool against a monthly index and rolling window in another.
  • Entity setup, meaning name variants, sub-brands and which competitors are switched on.

Take a brand that shows up reliably on the detailed, long-tail prompts its team wrote, but is missing from the one high-volume “best tools for X” question. In the two equal-weight trackers it looks healthy, because every prompt counts the same. In Brand Radar it looks weak, because one head query outweighs the rest. All three tools report that correctly.

Then there is the gap these tools only partly cover. A mention-based share of voice tells you how often a model names you. It does not tell you how often the model uses your pages as a source, which is what we call Share of Citation. Models tend to cite brands they have cited before, so a brand that is mentioned often but rarely cited is in a weaker position than its share of voice suggests.

What to Do Before Your Next Share of Voice Report

Start by writing down the formula next to every number you report. “Brand Radar AI SOV, impression-weighted, five competitors” means something. A bare percentage labelled “share of voice” does not.

  1. Use the same competitor list and the same brand name variants in every tool, then rerun. Many of the gaps between tools close at this step.
  2. Choose one tool as the record for trends and treat the others as cross-checks. Compare which way each number is moving, and stop comparing the levels.
  3. For the biggest gap, open the raw responses on five or six shared prompts and count the mentions by hand. This shows you which part of the formula is causing the difference.
  4. Check your custom prompt set against the high-volume questions Brand Radar shows for your category. If your prompts leave out the head queries, your equal-weight numbers are flattering you.
  5. Report mention share and citation share as two separate lines, and never average them together.

FAQs

Which Tool Gives the Most Accurate AI Share of Voice?

None of them is the most accurate, because they measure different things. The mention-share and presence trackers measure presence across prompts you choose, while Brand Radar models potential visibility weighted by search demand. Pick the one whose formula matches the question you are trying to answer.

Why Did My Competitive Share of Voice Change When AI Answers Did Not?

A normalised Share of Voice metric splits one pie among the active competitive set. Adding or removing a competitor changes every brand’s slice, even when the AI responses stay the same.

Is AI Share of Voice the Same as Share of Citation?

No. Share of voice in these tools counts how often a brand is named in answers. Share of Citation counts how often a model uses the brand as a source, and the two can move in opposite directions.

Can I Average Share of Voice Across Different AI Visibility Tools?

You should not. Each figure has its own count, its own total and its own prompt set, so an average of the three has no clear meaning. Track the trend in one tool and use the others to check it.

Gilad Bechar

About the author

Gilad Bechar

Gilad Bechar is the Founder & CEO of Moburst. Gilad serves as a mentor to rising startups at Microsoft Accelerator, The Technion, Tel-Aviv University, Unit 8200 and for strategic Moburst clients, and is the Academic Director of the Mobile Marketing and New-Media course at Tel-Aviv University.

Meet the whole team

Ready to make your brand the answer?

Our growth strategists take brands from overlooked to recommended. In one call you will see how the AI engines read your brand today, and leave with your own roadmap to dominate the new era of search.