Configure Prompt Monitoring

Every AI Visibility project often generates hundreds of prompts. Not all of them are equally important — and you don't want to send every single one to four LLM providers every day. With Prompt Monitoring you explicitly choose which prompts go into the next tracking run, and you immediately see what that costs you per month.

What is Prompt Monitoring?

Every prompt in the project has a monitor_status (active, paused, excluded). Only prompts with status active are included in the nightly tracking run — all others stay in the database but are not queried again against the LLM providers. Historical data is retained in every case.

In addition, Rankion provides the following for every prompt:

• monthly_search_volume — search volume via DataForSEO (daily refresh)
• priority_score — log(volume + 1) * intent_weight with intent_weight derived from commercial_intent (high=1.0, medium=0.7, low=0.4)
• funnel_stage — derived from prompt_category (Buy / Compare / Decision / Awareness)

Where do I find the selection?

In the project detail view (/ai-visibility/{id}), you'll find the "Monitoring" button in the top right. Direct link:

/ai-visibility/{trackingProject}/monitoring

Understanding the Cost Preview

At the top, sticky, it reads: "X of Y prompts selected — ≈ €Z/month". The estimate follows:

monthly_call_count = active_prompts × providers_per_run × runs_per_month
monthly_cost_eur   = monthly_call_count × avg_cost_per_call_cents / 100

Defaults from config/tracking.php: 4 providers, 30 runs/month, 0.5 ct per call. For 50 prompts that comes out to ≈ €30/month.

As soon as you change filters or pause/activate prompts, the preview recalculates live — without anything being saved.

Activate Top-N by Priority

Click "Activate Top 50 by Priority" in the bulk toolbar. This sets the 50 highest priority_score values to active and all other visible ones to paused. Don't forget to save — until then these are just local overrides.

Pausing Individual Prompts

Every table row is clickable — one click toggles between active and paused. The effective status is immediately reflected in the cost preview.

What Happens to Paused Prompts?

• They do not go into the next tracking run.
• Existing LlmResult data remains unchanged — trends and insights continue to show all historical values.
• You can reactivate them at any time via the same page.

Backwards Compatibility

As long as you haven't selected anything in the project (monitor_user_decided = false for all prompts), the tracking run behaves as before: all prompts with is_active=true are monitored. The subset logic only takes effect after the first save.

API Reference

The UI uses these 4 endpoints — they are also directly available:

• GET /api/v1/tracking-projects/{id}/prompts/cost-preview
• PATCH /api/v1/tracking-projects/{id}/prompts/{promptId}/monitor
• POST /api/v1/tracking-projects/{id}/prompts/bulk-monitor
• POST /api/v1/tracking-projects/{id}/prompts/refresh-volumes

In addition, the AI agent tool preview_monitoring_cost can be called in the agentic chat: "What would it cost to monitor all prompts with Persona X?"