Average rank

What it measures#

Average rank answers when AI engines name us, how early do they do it?

Order matters in an AI answer. The first brand in a “best tools for X” answer is the de facto recommendation; the eighth is a footnote. Visibility and share of voice treat both the same. Average rank is the metric that separates them.

In the app it is labelled Avg citation rank on the dashboard and Avg rank on the Prompts page, and it is shown with a hash: #2.3. Despite “citation” in the label, it is the position of your brand name in the answer text, not the position of your page in the list of sources.

How it is calculated#

• For every answer, the brands it names are numbered by order of first appearance: the first brand named is 1, the next new brand is 2, and so on. Each brand keeps only its earliest position.
• Your position in each answer that named you goes into the average. Answers that did not name you are left out, so the metric describes quality of presence, not coverage.
• The average covers every engine, country and tag in the current filters and date range, and is shown to one decimal place.
• The change indicator is the difference from the previous period of the same length. A negative change is good: it means you moved up.

Worked example#

In the last 30 days your brand was named in four answers:

(1 + 2 + 4 + 2) ÷ 4 = 2.25, displayed as #2.3.

Prompts where you were not named at all do not appear in this number. They are counted by not ranked instead.

Where rank appears#

The same position data is summarized four ways, depending on where you look:

The Prompts header figure is usually better than the dashboard one, because it takes each prompt’s best answer rather than every answer. Compare like with like.

Average rank vs. neighbouring metrics#

A good rank with low visibility means you lead where you appear but are missing from most prompts: fix coverage first. Poor rank with high visibility means you are included but not recommended: work on positioning.

What moves it#

• Answers that are lists. “Best X” and “X vs. Y” prompts produce ranked lists, where order carries most weight. They move this metric most.
• Where you are cited from. An answer built on one listicle can mirror that listicle’s order. If the main source for a prompt ranks you seventh, a better position on that page may help more than anything on your own site. Check Sources for the pages behind each prompt.
• Clarity about who you are best for. Prompts often describe a situation (a team size, a budget, a use case). Pages that state specifically who you are best for give engines a reason to put you first for that situation.
• Newly won prompts. Winning a new prompt at position 6 can worsen the average while visibility improves. Read rank next to visibility, not alone.

Related#