Why the numbers can be trusted
AI answers are not deterministic. Bad Popcorn treats every score as a sample: which prompts, which markets, which models, how often, and how stable the result is.
01
Prompt selection
Prompts are commercial questions shoppers ask before they buy, clustered by market and intent.
- Default markets are the Netherlands, Belgium, Germany, and France. Dutch prompts lead; each market is asked in its own language.
- Clusters separate branded, non-branded, purchase, comparison, and research intent.
- Demand is directional: we rank opportunity from visibility gaps, not from a fabricated search-volume number.
02
AI measurement
Each run asks the same prompt on the platforms in your plan and stores the answer.
- Platforms include ChatGPT, Claude, Gemini, Perplexity, and the others enabled on the account.
- Runs go through provider APIs. That is not the same as a logged-in consumer app with personalization.
- Location and language follow the prompt country. Repeats show volatility instead of hiding it.
03
Visibility scoring
A mention is not a recommendation. The score separates how the brand shows up.
- Mention: the brand or product is named.
- Citation: your site, or a source you control, is used as evidence.
- Position and recommendation strength: whether you are the suggested choice or a footnote.
- Context: sentiment and whether the facts match the catalogue.
Confidence
Every reading should be read with its sample size, last check, and how much it moved. A single answer is an anecdote. A repeated run on the same Netherlands or Europe prompt is a measurement.
Analyze my brand