Metrics explained

Every metric in Rankry comes from the same raw material: answers. One answer is one prompt sent to one engine in one scan. A project with 25 prompts on 3 engines produces 75 answers per scan.

All metrics follow the period and the engine filter in the project header.

Visibility

The share of answers that name you.

Visibility = answers that name you ÷ answers received.

If ChatGPT names you in 15 of 25 answers and Gemini in 10 of 25, your Visibility is 25 ÷ 50 = 50%.

Every mention counts the same, whether you’re first on the list or eighth. How high you appear is what Position measures.

Label	Visibility
Top Performance	70% or more
Needs Work	30–69%
Low Visibility	above 0, under 30%
Invisible Zone	0%

Position

Your average place in the answers that name you. Lower is better: 1.0 means you’re always first.

Position is averaged over the answers that name you, per engine, then across the engines that rank you. Answers that don’t name you are left out, so they don’t drag Position down; they lower Visibility instead. If no answer names you, Position shows N/A.

Example: named first in 6 answers and third in 4 gives (6 × 1 + 4 × 3) ÷ 10 = 1.8.

Label	Position
Excellent	2.0 or better, or first in every answer
Good	2.1–3.5
Average	3.6–5.0
Below Average	worse than 5.0

Sentiment

How positively the answers describe you, out of 100.

Rankry reads what each answer says about you and labels the answer Positive, Neutral or Negative, by the majority of its statements. A tie goes to Negative first, then Positive. For each engine, Rankry takes the share of positive answers; the score is the average across the engines that said something about you.

The Dashboard tile shows the positive share. The label also looks at the mix:

Label	When
Mostly Positive	60% or more positive
Mixed Sentiment	40–59% positive
Needs Attention	under 40% positive and more negative than positive
Neutral	under 40% positive and mostly neutral

Each prompt also shows a sentiment chip in the Prompts table, so you can find the answers behind a low score.

Diversity

How many of your engines name you at all.

Diversity = engines that named you at least once ÷ engines that answered. If 4 of your 5 engines name you, Diversity is 80%.

Label	When
Full Coverage	every engine names you
Partial Coverage	more than 40% of engines, but not all
Low Coverage	40% of engines or fewer
Not Detected	none

Low Diversity means you depend on one or two engines. The Model comparison card shows which ones are missing you.

Your slice of all mentions in your competitive set.

Share of voice = your mentions ÷ mentions of you plus your tracked competitors. The shares add up to 100%.

Visibility and Share of voice answer different questions. You can be named in 80% of answers and still hold 25% share of voice if three rivals are named just as often. See Share of voice.

Rankry Score

One number from 0 to 100 that sums up your standing.

The score blends how good your mentions are with how often they happen:

Quality = 40% Position + 35% Sentiment + 25% Diversity. Position is turned into a 0–100 scale where first place is 100 and tenth is 0.
Rankry Score = Quality × √Visibility.

The square root means visibility matters most when it’s low: going from 9% to 25% visibility lifts the score far more than going from 64% to 81%.

Example: Position 2.0 (89 on the 0–100 scale), Sentiment 70, Diversity 80% gives Quality 80. With 49% Visibility, the score is 80 × 0.7 = 56.

Label	Rankry Score
Strong	75–100
Good	55–74
Average	35–54
Weak	15–34
Critical	1–14

The Competitive leaderboard applies the same formula to every brand, so scores compare directly.

Changes over time

Dashboard tiles compare the period’s latest figures with its first scan. A longer period shows a longer trend; a 7-day period with one scan shows no change.