Understanding your numbers
What each metric means
Ocita reports four metrics for each brand. Each one has a value and a 90% range, and is labeled Stable or Directional. A brand’s Overview shows all four, for the engines and week you’re looking at.
Visibility
The share of sampled answers that mention your brand. If 40 of 100 answers name you, visibility is 40%.
A mention counts once per answer, however many times the name appears. Any of the brand’s aliases counts.
Share of voice
Your mentions as a share of all mentions of you and your tracked competitors. If answers name you 30 times and your competitors 70 times between them, your share of voice is 30%.
It depends on which competitors you track. Adding or removing one changes what you’re measured against. The SoV rank in Portfolio and the ranking in Competitors come from it.
Citation rate
The share of answers that link to at least one page on your domain or a subdomain. It’s worked out over answers that carry citations. Answers with no sources at all are left out and counted separately as No sources shown in Sources.
Average position
Where your brand lands, on average, when an answer lists brands in a bulleted or numbered list of at least two items. Lower is better: 1.0 means first every time. It only counts answers that put the brand in a list, and the Overview shows how many lists that was. Its change is shown in places rather than points.
How each answer is scored
- Mention: the name or an alias appears as a whole word, in any capitalization.
- Position: the brand’s place in a top-level list item. No list, no position.
- Citation: the links the engine returned, plus any web addresses in the answer’s text.
How Ocita measures has the full rules, and the glossary defines the terms.
Where the answers come from
Ocita collects what each engine’s own app shows, or for AI Overviews, Google’s results page, through a data provider, from the country you set. A signed-in person can still see something different, since the apps add memory and personalization. Repeating each prompt shows the overall pattern rather than one person’s screen.
Which answers count
The screens show the brand’s latest finished weekly run, or its first check until one exists. A prompt needs at least 3 completed answers on an engine to count that week; one with fewer is left out, and the run log says so. On-demand checks don’t change the metrics.