Otterly, a popular AI visibility tracking tool, has removed Brand Mention Count from its Brand Reports and replaced it with ratio-based metrics that show how often a brand appears, how early it is named and how it is described. The company announced the change on October 4, 2026 in an update on its blog. The reasoning behind it matters beyond one tool, because it exposes a measurement mistake many teams make when they start tracking AI visibility.
What did Otterly change?
Otterly’s Brand Reports track mentions across ChatGPT, Perplexity, Google AI Overviews, Google AI Mode, Gemini, Claude and Microsoft Copilot. The update made five changes to them.
- Sentiment and position now appear on the main chart next to coverage.
- Brand Mention Count has been removed from reports.
- Annotations let users mark actions on the timeline.
- The Brand Ranking table now shows position and sentiment next to coverage.
- Prompt Performance has a toggle for prompts that are trending up or trending down.
Why did Otterly remove Brand Mention Count?
According to Otterly, the count is an absolute number that moves with the size of a tracking setup, not with actual visibility. The company gives a worked example. A brand tracked across 50 prompts and mentioned 20 times has 40% coverage. After 50 more prompts are added and the brand is mentioned 30 times in total, the mention count has risen by 50%, but coverage has fallen from 40% to 30%.
The count went up because the number of prompts grew, not because the brand became more visible. A report that shows only the count would have read that period as a win when the brand was being named in a smaller share of answers. Changing the date range distorts the count in the same way.
How do the three remaining metrics work?
Each metric answers a different question, and none of them depends on how many prompts are tracked.
- Brand Coverage is the number of prompts that mention the brand divided by all prompts in the selected time window. Higher is better.
- Average Brand Position is the sum of the brand’s positions across prompts divided by the number of prompts where it appeared. Lower is better, and 1 means the brand was named first.
- Brand Sentiment uses a Net Sentiment Score: positive mentions minus negative mentions, divided by total mentions, multiplied by 100. The range runs from −100 to +100, and higher is better.
Together they separate three things that a single count blurred: whether the brand appears, how early it is named and how it is described.
What do annotations add to the reports?
Annotations place events on the timeline chart, so a change in coverage, position or sentiment can be read against what was done at the time. In Otterly, marking a recommendation as completed creates an annotation automatically. Users can also add their own for work done outside the tool, such as a new page, a PR push or a listing update.
This closes a common reporting gap. A chart can show that coverage moved, but without a record of actions it cannot show whether the work caused the move.
How stable are the numbers?
Otterly also published stability data on prompt set size. Moving from 10 prompts to 100 reduced variation in reported Brand Coverage by about 73%, according to the company. At 10 prompts, 90% of results fell between 39% and 71%. At 100 prompts, 90% of results fell between 51% and 60%.
The practical reading is that a coverage figure from a small prompt set can swing by more than 30 points without any change in visibility. Month-to-month comparisons only mean something when the prompt set is large and held constant.
What does this mean for brands measuring AI visibility?
The update points to three habits that apply to any tool. Report rates, not raw counts, because rates survive changes in setup size. Keep the prompt set stable between reporting periods, and treat any period in which it changed as a new baseline. Log every action that could affect visibility next to the data, so that results can be tied to causes.
Sentiment and position add a second layer. A brand can have healthy coverage while being named last or described inaccurately, and neither shows up in a coverage figure alone. Tracking all three is the minimum for a report that leadership can act on, which is the approach behind AI brand mention tracking across multiple engines.
Otterly’s move is a signal that the category is maturing from counting mentions to measuring them. The metrics that last will be the ones that stay comparable when the setup changes.
