Alert Rules
Configure the conditions that trigger alerts — visibility drops, sentiment shifts, new competitor mentions, and more.
Overview
An alert rule is a condition you want the platform to watch. When that condition becomes true after a snapshot run, the rule fires and an alert is added to the Alerts feed (and optionally sent via email, webhook, or Slack).
Rules are managed at Settings → Alert Settings. Each project has its own set of rules.
Rule anatomy
Every rule has four parts:
| Part | Purpose |
|---|---|
| Metric | What you're watching (visibility score, mention count, average rank, sentiment, competitor presence, etc.) |
| Operator | How you're comparing (greater than, less than, changed by more than, equals) |
| Threshold | The number that triggers the rule |
| Scope | Global for the project, or scoped to a specific LLM, prompt, or competitor |
For example: visibility_score < 40 on ChatGPT would fire any time your visibility on ChatGPT drops below 40%.
Supported metrics
| Metric | What it measures | Common thresholds |
|---|---|---|
| Visibility score | % of snapshots mentioning your brand | < 50 or dropped by 20% |
| Mention count | Raw mentions across snapshots | < 10 in 7 days |
| Average rank | Position in numbered lists (1 = first) | > 5 |
| Sentiment ratio | % positive vs negative | negative > 20% |
| Competitor mentions | How often a named competitor appears | > your mentions |
| New competitor | A brand not in your tracking list appeared | fires on first detection |
| SERP position | Google AI Overview or Bing Copilot rank | > 3 |
Severity levels
Pick the right severity so noisy rules don't drown out important ones:
- Info — FYI, no action required (e.g., "a new source cited your page")
- Warning — Worth investigating within a day or two (e.g., "visibility dropped 10%")
- Critical — Needs same-day attention (e.g., "visibility dropped 30%" or "new negative sentiment spike")
Creating a rule
- Go to Settings → Alert Settings
- Click New Rule
- Give it a descriptive name (e.g., "ChatGPT visibility floor")
- Pick the metric, operator, and threshold
- Optionally scope it to a specific LLM, prompt, or competitor
- Choose severity
- Pick which notification channels should fire (in-app is always on)
- Save
The rule evaluates automatically after every snapshot run. You don't need to trigger anything manually.
Editing and pausing rules
- Edit — Change any field. Future evaluations use the new values; historical alerts are not rewritten.
- Pause — Temporarily stop a rule from firing without deleting it. Useful during known-noisy periods (e.g., a launch week).
- Delete — Remove the rule permanently. Historical alert history is preserved.
Plan limits
Each plan has a limit on how many alert rules you can have per project:
| Plan | Max rules per project |
|---|---|
| Free | 3 |
| Starter | 5 |
| Professional | 10 |
| Enterprise | 50 |
See Configuration → Organization Settings for details on plan limits.
Best practices
- Think in terms of "what would I want to wake up to?" Alerts should be things you'd take action on, not general observability.
- Start with floor-style rules (
visibility < X) before adding change-based rules (dropped by Y%) — they're more intuitive to reason about. - Scope rules where it makes sense. A rule scoped to your #1 priority prompt is more actionable than a global rule.
- Review rules quarterly. As your brand grows, yesterday's "concerning" numbers become today's baseline. Raise thresholds when appropriate.