The question behind the test.
As riders move from conventional search into conversational systems, brands want to understand whether their history, products and positioning are represented consistently. We used Norton as a multilingual subject to explore how answers changed across language and phrasing.
The purpose was not to produce a league table or declare a permanent result. It was to understand variation and build a better testing method.
What we compared.
The test used comparable prompts in five languages and recorded the visible answers at the time of testing. We considered brand recall, themes, specificity, confidence and the sources or assumptions behind the response.
- English
- French
- German
- Spanish
- Italian
What the exercise taught us.
AI answers are dynamic. Language, market context, prompt structure, model version and available web information can all change the result. A useful visibility programme therefore needs repeated testing, clear methodology and stronger source content, not one screenshot presented as a permanent truth.
Five language versions were recorded as part of the original test asset set.
This was an observational test, not a controlled scientific benchmark. Results can change and should not be interpreted as a current ranking or endorsement.