Skip to content

Rankry Metrics Reference

Rankry reports a small set of metrics per model and overall. Each is defined below with its range and how it is computed. Read the metrics together: a high Visibility with a poor Position or Sentiment still signals a problem.

  • Each metric and its range
  • How the overall Score is computed
  • How citation metrics work
MetricRangeDefinition
Visibility0 to 100%Share of tracked prompts where your brand is mentioned (mention rate)
Position0 to 100%How highly you rank inside an answer, normalized from rank
Sentiment0 to 100% per classShare of positive, neutral, and negative claims about your brand
Diversity0 to 100%Share of models where your brand appears
Score0 to 100Overall standing, a weighted composite (see below)
TrenddeltaChange from the previous report
ZoneRange
TopAbove 70%
Needs work30% to 70%
Low visibleBelow 30% with some mentions
Invisible0%

Position normalizes rank so that being named first counts most:

position = (10 - rank) / 9 × 100

Rank 1 maps to 100, and rank 10 maps to 0. Rankry also tracks first-place rate, the share of prompts where you are named first.

The Score combines quality and presence. Quality weights Position, Sentiment, and Diversity. Presence acts as a gate, so low Visibility dampens the result:

qualityScore = position × 0.40 + sentiment × 0.35 + diversity × 0.25
Score = round(qualityScore × sqrt(visibility / 100))

Position is weighted 40%, Sentiment 35%, and Diversity 25%. The square-root gate means a brand that is rarely mentioned cannot score highly on quality alone.

Diversity also reports how aligned the models are, from unified to fragmented:

LevelMeaning
UnifiedThe models agree closely
AlignedMost models are similar
MixedModerate variation
FragmentedHigh disagreement
MetricDefinition
Citation shareShare of cited sources that point to your domain
Source gapCited competitor domains you are absent from

Sources are classified by type (review, community, publisher, wiki, competitor, own) and prioritized by how often they are cited and how actionable they are.

The Score also weighs Position, Sentiment, and Diversity. Being mentioned poorly or inconsistently lowers it.

Yes. Position is normalized so that 100 is the top of the answer and 0 is the bottom.

There is no universal target. Track the trend and compare against competitors.