How Conviora measures AI visibility
In Conviora, AI visibility is how often — and how strongly — AI engines mention and recommend a brand when they answer unbranded buyer questions for one specific market and language. This page describes exactly how that is measured and what the measurement cannot tell you.
What “AI visibility” means here
For each AI answer, Conviora classifies whether the brand was present, how strongly it was recommended, how prominently it appeared, and whether the brand’s own domain was cited as a source. Answers are collected from real AI engines, not simulated.
Scope: Brand + Market + Language
Every measurement belongs to one exact combination of brand, market and language. Questions are written for that market and asked in that language, and the result describes only that context. Conviora does not produce a single global brand score.
The Full Baseline: 15 prompts × 5 AI engines
The Full Baseline asks 15 prompts in each of 5 AI engines, for 75 AI answers in total.
| Prompt type | Prompts | AI engines | Answers | Used for |
|---|---|---|---|---|
| Unbranded buyer questions | 12 | 5 | 60 | The main visibility score |
| Branded diagnostic prompts | 3 | 5 | 15 | Diagnostics only — not scored |
| Full Baseline total | 15 | 5 | 75 |
Why branded prompts do not affect the score
A branded prompt names the brand, so an AI engine will almost always mention it. Counting those answers would inflate the score and hide real gaps. The 3 branded prompts are therefore diagnostics: they show how AI describes the brand when asked directly. Only the 60 answers to the 12 unbranded buyer questions are scored.
The five AI engines
- ChatGPT
- Gemini
- Claude
- Perplexity
- Google AI Mode
Results are reported per engine as well as combined, because the same brand can be visible in one engine and absent from another.
How the visibility score is calculated
| Signal | Weight |
|---|---|
| Presence — the brand is included in the answer | 40% |
| Recommendation strength | 30% |
| Prominence — position within the answer | 20% |
| Citation of the brand’s own domain | 10% |
The weighted result is normalised to 0–100. Only successfully collected and classified answers to scoring questions count. Answers that fail to collect are reported separately instead of being counted as absences.
Evidence for every number
Each observation records the prompt, engine, model, date and source evidence, so every score can be traced back to the answers behind it.
Recurring, comparable monitoring
Monitoring repeats the measurement for the same brand, market and language on a recurring schedule, so results can be compared with earlier measurements of the same scope rather than with a different set of questions or markets.
Before/after verification
When an improvement has been implemented, Conviora re-measures the buyer questions that improvement targeted and compares the result with the earlier measurement of the same scope. Verification reports what changed; it does not claim the improvement caused the change.
Instant Check versus Full Baseline
The free Instant Check asks a small set of buyer questions in one AI engine. It is an early signal for one context and is not comparable with the 60-answer Full Baseline score.
Limitations
- AI answers vary. The same question can produce a different answer minutes later, so every result is a sample, not a fixed ranking.
- Every measurement is a snapshot of one point in time. Engines update their models, sources and behaviour without notice.
- Results apply only to the brand, market and language that were measured. They do not describe visibility anywhere else.
- The buyer questions are a representative sample of buyer intent, not every question a buyer could ask.
- Movement observed after an improvement is implemented is an observation, not causal proof. Other changes in the engines or the market can produce the same movement.
- Conviora does not guarantee that visibility or recommendations will improve.
See the methodology applied to your brand
Run the free Instant Check for your website.
Run the free Instant Check