> ## Documentation Index
> Fetch the complete documentation index at: https://rankahead.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Prompts: Track Your Brand Across AI Models
> Create, schedule, and run queries that Rankahead sends to multiple LLMs simultaneously to measure where your brand appears in AI-generated answers.
Prompts are the foundation of Rankahead's visibility tracking. Each prompt is a query — written the way a real user might ask an AI assistant — that Rankahead sends to every connected LLM provider on a schedule you control. After each run, Rankahead analyzes the responses to determine whether your brand was mentioned, how prominently, and what topics are being discussed around it.
## Create your first prompt
Navigate to **Dashboard → Prompts** in the sidebar. If you have no prompts yet, you will see an empty state with a **Create prompt** button. Click it to open the prompt form.
Enter a short, descriptive label in the **Name** field. This name appears in the run history and dashboard charts — it is for your reference only and is not sent to any LLM. Example: `Best project management tools`.
In the **Query** field, enter the question exactly as a user might type it into ChatGPT or Perplexity. Write the query in natural language; do not reference your brand name in the query itself, since the goal is to see whether AI models surface your brand unprompted.
Good examples:
* `What are the best tools for tracking brand visibility in AI search?`
* `Which platforms help with generative engine optimization?`
* `What should I use to monitor how my company appears in ChatGPT?`
Enter your **Brand name** — the exact name Rankahead should look for in AI responses. Optionally enter your **Brand domain** (e.g., `rankahead.ai`) to enable domain-level citation tracking alongside keyword mentions.
Select the **Locale** that matches your target market. Locale affects how queries are interpreted by LLMs that support multilingual inputs. The default is `en-US`.
Enter a cron expression in the **Schedule** field to run this prompt automatically. Leave it blank to run manually only. See [cron schedule examples](#cron-schedule-examples) below for reference.
Toggle **Active** to enable the prompt, then click **Save**. Rankahead immediately queues the first run if you set a schedule, or you can trigger a manual run from the prompt detail page.
## Prompt fields reference
A human-readable label for the prompt. Used in the run list, dashboard charts, and gap analysis views. Keep it short and descriptive so you can identify it at a glance.
The exact question sent to each LLM provider. Write it in natural language as a user would, without mentioning your brand. The same query text is used for every provider, ensuring consistent comparisons across LLMs.
The name Rankahead searches for in each AI response. This should match how your brand is commonly referenced — for example, `Rankahead` rather than `Rankahead Inc.`. The match is case-insensitive.
Optional. When provided, Rankahead also checks whether AI responses include links or references to your domain. This feeds into the [Citations](/dashboard/citations) dashboard.
A standard five-field cron expression that controls when the prompt runs automatically. See [cron schedule examples](#cron-schedule-examples) below. Leave empty to disable automatic scheduling.
The language and region context for the prompt. Affects response language on providers that support locale-aware queries. Examples: `en-US`, `en-GB`, `fr-FR`, `de-DE`.
When toggled off, the prompt is paused. Scheduled runs are skipped and the prompt does not appear in visibility score calculations. You can reactivate it at any time without losing run history.
## Cron schedule examples
The schedule field accepts standard cron syntax: `minute hour day-of-month month day-of-week`.
| Expression | Schedule |
| ------------- | ---------------------------------------- |
| `0 9 * * 1` | Every Monday at 9:00 AM UTC |
| `0 9 * * 1,4` | Every Monday and Thursday at 9:00 AM UTC |
| `0 0 * * *` | Daily at midnight UTC |
| `0 9 1 * *` | First day of every month at 9:00 AM UTC |
| `0 */6 * * *` | Every 6 hours |
A weekly schedule (e.g., every Monday morning) is a good starting point for most brands. Daily runs give you more granular trend data but consume more run quota depending on your plan.
## Running a prompt manually
To trigger a run immediately without waiting for the next scheduled time, open the prompt's detail page and click **Run now**. The run is queued instantly and typically completes within 30–90 seconds depending on provider response times.
You can run a prompt manually even if it has no schedule set. This is useful for one-off checks after publishing new content or after a competitor announcement.
## Reading run results
Click any row in the run list to open the full result detail. Runs are listed most recent first and paginated in groups of 20. Each run shows:
* **Status**: `PENDING`, `RUNNING`, `COMPLETED`, or `FAILED`
* **Started at** and **completed at** timestamps
* **LLM results**: The raw response from each provider, with brand mentions highlighted
* **Brand analysis**: A summary of how the brand was referenced — mentioned, not mentioned, or cited as a source
* **Keyword suggestions**: Topics and keywords that appeared in AI responses that your brand could better target
Each provider's response is shown in full, with your brand name highlighted wherever it appears. If your brand is not mentioned, the response is still shown so you can see how AI models are answering the query and which competitors or sources they reference instead.
The brand analysis panel summarizes the run across all providers: how many mentioned your brand, what sentiment was expressed (positive, neutral, or negative), and where in the response the mention appeared (opening recommendation, supporting detail, or footnote). This data feeds directly into your dashboard KPIs.
After analyzing responses, Rankahead extracts the topics and terms that appear frequently in AI answers for this query. These are presented as keyword and topic suggestions — terms your content should cover to improve the likelihood of appearing in future AI responses to similar queries.
A run with status `FAILED` means at least one provider returned an error or timed out. Failed runs are excluded from visibility score calculations. Open the failed run to see which providers succeeded and which did not. You can re-trigger the run manually from the same detail page.
## Managing multiple prompts
As you scale your tracking, you will likely accumulate prompts across different topic areas, product lines, or markets. A few practices to keep things organized:
* Use consistent naming conventions — for example, prefix by category: `[Product] Best CRM tools`, `[Brand] What is Rankahead?`
* Disable prompts for topics you are no longer actively targeting rather than deleting them; this preserves historical run data
* Group similar queries together by locale to track regional differences in AI responses
* Review keyword suggestions across runs regularly — they are one of the fastest ways to discover new content opportunities
Each prompt runs against all LLM providers simultaneously. If you have four providers configured, one prompt run generates four LLM responses. Your plan's run quota is measured per LLM response, not per prompt.