No black box. Here's exactly how AnswerRank turns "what does the AI say about us?" into a number you can move.
We generate the buying-intent questions your customers actually ask — "best <category>", "<category> for beginners", "best premium/budget <category>", "what do experts recommend", and more. Growth tracks 25; Pro tracks 100. You can add your own.
Every prompt runs against ChatGPT, Google AI Overviews, Gemini and Perplexity — weighted by real-world usage (40 / 30 / 18 / 12) so your score reflects where buyers actually are.
Each engine is scored on five signals, then blended by usage weight into your overall AI Visibility Score.
| Signal | Weight | What it captures |
|---|---|---|
| Presence | 40% | Share of prompts where your brand is mentioned at all — the foundation of everything. |
| Prominence | 20% | How early and prominently you appear — named first beats a footnote. |
| Sentiment | 15% | Positive, neutral or negative framing of the mention. |
| Share of voice | 15% | Your mentions vs. every competitor's, across all answers. |
| Citation | 10% | Whether your own domain is cited as a source the engine trusts. |
Every competitor is measured on the same battery, so share of voice is apples-to-apples.
We re-run automatically (weekly or daily) and chart the trend, so you can prove progress.
Gaps become a prioritised action plan — the content, schema and mentions that get you cited.
These engines don't invent recommendations — they synthesise what they were trained on and what they can retrieve. In practice, brands win AI answers by:
Getting named in the "best of" roundups, Reddit threads and YouTube reviews the models lean on.
Clear Q→A pages, comparison tables and FAQ/Product schema the models can lift verbatim.
Original data, specs and a Wikipedia-grade fact base engines can attribute to you.
Recent, updated content that overrides stale narratives already in the models.