Events

Close the loop between AI interactions and business outcomes.

Connect real-world user actions to your AI conversations. By tracking timestamped business events, you enable Greenflash to validate which interactions drive revenue, retention, and delight—and which ones lead to churn.

Events are the bridge between conversation and conversion. While tools like Amplitude or PostHog track what happened, Greenflash connects why it happened by linking outcomes directly to the AI interactions that preceded them.

Why Track Events?

Greenflash provides powerful analytics on user interactions, giving you visibility into conversation quality, sentiment, and engagement. Tracking events adds a critical layer of business context to these insights.

By correlating downstream user actions with upstream AI interactions, you can:

• Attribute Success: Link specific AI responses directly to business outcomes like upgrades or purchases.
• Validate Quality: Confirm heuristic signals (like sentiment) with tangible outcome data.
• Deepen Insights: Train Greenflash to identify the conversation patterns that drive real-world results.

Events work for any AI product that drives user actions, not just chatbots. Whether you’re building AI-powered sales tools, workflow automation, or content generation, events let you measure what matters: real business outcomes.

When to Send Events

Capture moments that matter. Send an event whenever a user reaches a milestone that represents value creation (or loss).

Positive Indicators

Signals that the user is succeeding.

• signup_completed
• trial_started
• task_success
• upgrade_purchased

Negative Indicators

Signals of friction or churn.

• cancellation_requested
• error_encountered
• workflow_abandoned

Contextual Indicators

Neutral signals that provide usage context.

• feature_viewed
• step_completed
• usage_logged

Integration Guide

Sending events is a simple API call. You can trigger this from your backend, frontend, or webhooks.

The Endpoint

POST https://www.greenflash.ai/api/v1/events

Basic Example: Customer Support Chatbot

Here’s how you might track a user upgrading to a “Pro” plan immediately after a helpful support chat:

curl --request POST \
  --url https://www.greenflash.ai/api/v1/events \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '{
    "eventType": "upgrade",
    "productId": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
    "conversationId": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
    "influence": "positive",
    "value": "149.00",
    "valueType": "currency",
    "properties": {
      "plan": "pro",
      "billing_cycle": "annual"
    },
    "insertId": "unique-id-123"
  }'

Beyond Chatbots: Other AI Product Examples

Events are valuable across many types of AI products. Here are some scenarios:

Sales Engagement: AI-Suggested Outreach

If your product helps sales teams with AI-generated comments or relationship-building suggestions, track when those interactions convert to real conversations:

# Track when a prospect replies to AI-suggested engagement
POST /api/v1/events
{
  "eventType": "prospect_replied",
  "conversationId": "abc-123",
  "influence": "positive",
  "value": "1",
  "valueType": "numeric",
  "properties": {
    "engagement_type": "linkedin_comment",
    "days_since_first_touch": 3
  }
}

# Track when the conversation advances to a meeting
POST /api/v1/events
{
  "eventType": "meeting_booked",
  "conversationId": "abc-123",
  "influence": "positive",
  "value": "50000",
  "valueType": "currency",
  "properties": {
    "pipeline_stage": "qualification",
    "engagement_touches": 5
  }
}

The insight: Which AI suggestions actually lead to pipeline? What patterns exist in successful engagement strategies vs. ignored ones?

Operations: AI Workflow Analysis

If your product uses AI to analyze and optimize business processes, track when recommendations get implemented:

# Track when an AI-identified improvement is acted on
POST /api/v1/events
{
  "eventType": "workflow_optimized",
  "conversationId": "workflow-session-456",
  "influence": "positive",
  "value": "40",
  "valueType": "numeric",
  "properties": {
    "process_area": "finance_operations",
    "improvement_type": "automation_identified",
    "estimated_hours_saved": 40
  }
}

# Track the business impact when automation goes live
POST /api/v1/events
{
  "eventType": "automation_deployed",
  "conversationId": "workflow-session-456",
  "influence": "positive",
  "value": "5000",
  "valueType": "currency",
  "properties": {
    "automation_type": "process_standardization",
    "departments_impacted": 3
  }
}

The insight: Are AI discovery sessions leading to real operational changes? Which types of insights drive the most value?

Lead Intelligence: AI Social Monitoring

If your product monitors conversations to identify high-intent prospects, track which signals lead to actual conversions:

# Track when an AI-identified lead responds to outreach
POST /api/v1/events
{
  "eventType": "lead_responded",
  "conversationId": "signal-789",
  "influence": "positive",
  "value": "1",
  "valueType": "numeric",
  "properties": {
    "intent_signal": "competitor_frustration",
    "platform": "reddit",
    "time_to_response_hours": 2
  }
}

# Track when that lead becomes a customer
POST /api/v1/events
{
  "eventType": "lead_converted",
  "conversationId": "signal-789",
  "influence": "positive",
  "value": "2500",
  "valueType": "currency",
  "properties": {
    "channel": "social_listening",
    "time_to_conversion_days": 14,
    "original_intent_type": "competitor_mention"
  }
}

The insight: Which intent signals convert vs. which are false positives? Optimize your monitoring criteria based on actual revenue data.

Key Fields

Field	Type	Required	Description
eventType	string	Yes	The name of the event (e.g., upgrade).
productId	uuid	Yes	Your Greenflash Product ID.
value	string	Yes	The measurable value (e.g., "149.00").
valueType	enum	No	How to interpret value: currency, numeric, text, or boolean. Defaults to text.
influence	enum	No	positive, negative, or neutral. Defaults to neutral.
qualityImpactScore	number	No	Fine-grained quality impact from -1.0 to 1.0. Overrides influence if provided.
conversationId	uuid	No	Link this event to a specific Greenflash conversation.
externalConversationId	string	No	Link using your own conversation identifier.
externalUserId	string	No	Associate the event with a user in your system.
eventAt	ISO 8601	No	When the event occurred. Defaults to now. Useful for importing historical data.
insertId	string	No	Unique key for idempotency. Prevents duplicate events on retry.
properties	object	No	Additional context as key-value pairs for filtering and analysis.

For the complete API specification, check out the API Reference.

Sampling

For high-volume applications, you can control the percentage of events that are ingested using the sampleRate field (0-1). Use forceSample: true for critical events that should never be dropped.

For detailed information on sampling behavior, including the differences between message and event sampling, see the Sampling documentation.

The Impact on Quality Scores

Events directly feed into your Product Quality Index (PQI) and Conversation Quality Index (CQI).

1. Weighted Scoring: You can configure the impact of events in your quality settings. This allows you to weight business outcomes (like purchases) more heavily than other signals in your CQI and PQI scores.
2. Validation: Events serve as the “ground truth” for your analytics. They turn assumptions into facts.

Recommended Workflow

1. Identify Milestones: Pick events that define success for your product (e.g., “Onboarding Complete”, “Subscription Started”).
2. Instrument Your Code: Add the API call to your backend logic where these events occur.
3. Link Conversations: Whenever possible, pass the conversationId if the event happened during or shortly after an AI session.
4. Verify: Check the Greenflash dashboard to see your events populating the Business Outcomes view.

Next Steps

Product Analytics

See how events drive your product metrics.

Connect Conversations

Learn how to link events to specific chats.