Concept
Prompts
See how ChatGPT, Gemini, Claude, and Google AI Mode answer questions about your market and how often they mention your brand.
The Prompts section tracks what AI assistants say about your brand. You store a set of questions — prompts — and GripIQ asks them repeatedly, then analyses every answer for the brands it mentions, the tone it uses, and the web pages it cites.
Where to find it
Section titled “Where to find it”Open Prompts in the navigation rail. The section has three tabs:
- Overview — the metrics for the whole project.
- Prompts — the list of prompts you track, with a count badge.
- Answers — every captured answer, grouped per prompt.
How answers are collected
Section titled “How answers are collected”Each active prompt is sent to three AI models: ChatGPT (OpenAI), Gemini (Google), and Claude (Anthropic); on every scan it also gets a Google AI Mode answer (see below). GripIQ uses one model per assistant, chosen to match what that assistant serves its own users by default, so the results reflect what a real user of that assistant is told. GripIQ keeps this in step with the assistants themselves: when an assistant changes the model it serves its users by default, GripIQ moves to that model with it. Answers are grouped per assistant, so a model update never splits your filters or metrics — you may only notice a shift in an assistant’s answers or trend line around the date it changed.
On every scan, each active prompt also gets a Google AI Mode answer: the answer Google shows in AI Mode for that question, collected for the prompt’s region in your project’s language. Google AI Mode appears as its own assistant, separate from Gemini, so its answers never mix with Gemini’s numbers. When Google shows no AI Mode answer for a question, no answer is stored for that scan.
Prompts are asked on your project’s scan schedule: daily, weekly, monthly, or disabled. There is no button to run a prompt immediately — answers appear after the next scan.
When it answers, the assistant may search the web. Every prompt has a region: a new prompt gets your project market unless it is given a region of its own. Each model is asked to answer as a person in that region who speaks your project’s language, and its web search is biased to that region. Every answer records the region it was generated for. The pages the model actually consulted are stored with the answer as Sources. For Google AI Mode, the sources are the pages Google cites in its AI Mode answer.
Every answer is then analysed automatically. GripIQ records which brands are mentioned, in which order, the sentiment of the answer, and its tone. Brand detection comes from that analysis rather than from plain text matching, and each answer counts at most once per brand.
GripIQ also classifies each prompt as Branded when its text names a brand — one of your brands or any widely known brand — or Non-branded when it does not, so you can narrow the Overview tab to either group with the Brand filter.
Key metrics
Section titled “Key metrics”- Visibility — the percentage of answers in the selected date range that mention any of your own brands.
- AI share of voice — your brand’s slice of all brand mentions across every monitored brand.
- Rankings — per brand: visibility, average sentiment, and average position.
- Position — where your brand appears in an answer relative to other brands. Lower is earlier.
- Average sentiment — a score from −1 to +1 across the answers in the range.
- Top domains — the websites the models cite most often in their answers.