Methodology
How we measure your business's visibility in AI
ELTRYS analyses the answers of ChatGPT, Claude, Gemini and Perplexity to understand how visible your business is, how it is positioned against competitors and what information the models give about it. The score is calculated over four components with defined weights and comes with the evidence that supports every finding.
The formula
Four components. One score out of 100.
Our methodology combines four measures to build an overall AI visibility score. Each one assesses a different aspect and contributes according to its weight.
Select a component in the table or in the chart to see how it affects the final score.
| Component | Weight | Example value | Contribution |
|---|---|---|---|
| 45% | 90 | 40.5 | |
| 20% | 70 | 14.0 | |
| 20% | 52 | 10.4 | |
| 15% | 88 | 13.2 | |
| Score | 100% | 78out of 100 |
Illustrative exampleIn this example, the four components add up to a score of 78.1 points. In the whole-number display, the score appears as 78 out of 100.
Components
What each component measures
To understand a business's visibility it is not enough to know whether it is mentioned. We also need to see its position, its presence against competitors, the sources that are cited and the accuracy of the information. That is why the score is built from four components that complement one another.
Presence & Ranking
What we measure
How often your business is mentioned in AI answers and in what position it appears when companies, products or services are recommended.
How we calculate it
We combine how often the business is mentioned with its position in the answers where it appears. Being absent from an answer is not penalised twice.
What you should know
A mention is not always a recommendation. That is why we also examine how the business is described.
- 1BNA
- 2Meridian Business Circle
- 3Summit Partners Network
A ranking like this shows the business's position in one particular answer. The overall result, however, rests on all the answers included in the assessment.
What happens when a component cannot be assessedIf the evidence is not enough to calculate a component reliably, it is marked "Not assessed", together with the reason. This component is treated neither as zero nor as a maximum score. The overall score is calculated over the assessed components, and it is clearly reported which ones are included. Two scores are fully comparable only when they rest on the same components.
Reading it
How to read your score
The score helps you understand the level of your business's visibility in the questions analysed. The score is a starting point for understanding the findings, not a guarantee of how the models will answer in the future.
Uncertainty
What you should know about the results
There is not yet a shared industry standard
Measuring visibility in AI is a developing field. The ELTRYS methodology is documented and transparent, but it is not presented as a universal standard.
The same question can produce different answers
AI models do not always answer the same way. The AI Audit repeats every question three times, while Monitoring uses repetition to follow changes.
We have no access to the models' internal processes
We analyse the answers that the models give. We cannot determine with certainty all the internal reasons why one business is recommended and another is not.
Within Information Accuracy, tone is assessed differently from facts
Factual claims can be checked against sources. Tone also requires interpretation, which is why the report explains the basis of the assessment.
Not every model cites sources
The sources component is calculated over the answers that provide identifiable sources, clearly showing the limits of the assessment.
The methodology may be updated
Significant changes are documented and communicated. We do not change the weights or the way a score is calculated after it has been produced.
Next step
Find out what AI answers say about your business
The AI Analysis uses this methodology to give you an initial assessment of your business's presence, with findings and recommendations.