A single metric reads poorly

An isolated citation rate ranks real situations badly.

Take two brands tracked on the same prompt plan, the first of which is cited in 80 per cent of answers, but always in last place, never advised and in lukewarm terms, whereas the second appears in only 50 per cent of answers while consistently coming first, actively recommended and described in flattering language.

A plain citation rate would put the first ahead, whereas commercially it is the other way round. The Radar Index fits those six dimensions into one manageable figure without hiding the detail, since every axis stays viewable, comparable over time and exportable.

No black box. The six weightings are fixed and published below, so from the metrics shown in your workspace you can redo the calculation on the back of an envelope.

The detail

The axes and their weight, in detail

Each axis is first brought onto a 0 to 1 scale, then multiplied by its weight. The sum of the six gives the final score out of 100.

Axis	Weight	Source metric	How it is computed
Presence Do you appear in the answer at all?	25	`mention_rate`	Share of repetitions in which the brand is cited, across all engines.
Recommendation Are you advised, or merely listed?	25	`reco_rate`	Share of repetitions in which the assistant actively puts the brand forward rather than listing it.
Share of voice What room do you hold against other brands?	20	`sov`	Brand mentions as a proportion of all mentions of every brand detected.
Primacy How early in the answer do you arrive?	15	`avg_rank`	Average rank of appearance, brought onto a 0 to 1 scale with a horizon of 8 positions.
Stability Is your presence consistent?	10	`stability`	Consistency of presence across repetitions of the same prompt, steady presence counting for more than erratic presence.
Tone In what terms are you described?	5	`sentiment`	Balance of positive and negative mentions, brought onto a 0 to 1 scale.

The calculation, on the example above

Axis	Recorded value	Normalised	Weight	Points earned
Presence	62% of repetitions	`0.62`	25	15.5
Recommendation	34% of repetitions	`0.34`	25	8.5
Share of voice	28% of mentions	`0.28`	20	5.6
Primacy	average rank 2.4	`0.83`	15	12.4
Stability	0.81	`0.81`	10	8.1
Tone	balance +0.45	`0.73`	5	3.6
Radar Index				54 / 100

Reading the score

The reading bands

The bands qualify a situation at a glance, while the trajectory matters more: a score of 38 gaining 4 points a month is worth more than a score of 55 standing still.

0 - 19

Absent

20 - 39

Marginal

40 - 59

Established

60 - 79

Strong

80 - 100

Dominant

The reading frame

The exact scope of the score

Your prompt plan

The score covers the prompts you track, so it compares against itself over time and between brands on a shared measurement plan.

Multiple repetitions

Assistants never answer the same way twice, which is why each prompt is run several times per engine before anything is counted.

A score per engine

The consolidated Radar Index covers real gaps between assistants, and the engine-by-engine detail stays on screen since that is where decisions are taken.

A tracked trajectory

The score follows the movement recorded since your baseline, in a market that is moving too, which makes it a progress indicator rather than an absolute mark.

Live grounding and model weights. Assistants that browse the live web, such as Perplexity, AI Overviews and ChatGPT in search mode, respond to published content as crawls come through. Answers produced without browsing follow the models' training cycles. Georadar tracks the two situations separately, so you know which lever you are pulling and on what horizon.

What is your brand's Radar Index?

Create your account, approve your prompt plan and launch the first reading. The score appears as soon as it finishes, with the six axes broken out.

Calculate my Radar Index Frequently asked questions

The Radar Index, end to end