Understanding Signals

Anacoic uses signals as our billing metric. Unlike traditional “events,” signals represent all data processing that happens on our edge infrastructure.

What is a Signal?

A signal is any data packet that our edge workers process. This includes:

1. Delivery Signals

Successful and attempted dispatches to ad platforms:

• Meta Conversions API
• TikTok Events API
• Google Ads Enhanced Conversions
• LinkedIn, Snapchat, Pinterest APIs
• Custom webhooks

2. Agent Signals (MCP)

AI Agent interactions via Model Context Protocol:

• Resource requests (pricing, docs)
• Tool invocations (book_demo, etc.)
• SSE connection maintenance

3. Security Signals

Traffic processed by our security layers:

• Domain lock validations
• Bot shield blocks
• Rate limit enforcement
• Unauthorized origin rejections

4. Retry Signals

Recovery and retry attempts:

• Failed delivery retries
• QStash queue processing
• Dead letter recovery

Why Signals Instead of Events?

Traditional analytics platforms only count “events”—user actions like page views or purchases. But Anacoic is infrastructure, not just analytics.

Metric	What It Counts	Best For
Events	User actions only	Product analytics (Mixpanel, Amplitude)
Requests	HTTP requests	API gateways (Cloudflare, AWS)
Signals	All infrastructure processing	Edge data delivery (Anacoic)

We chose signals because:

1. Fair billing: You pay for the infrastructure you actually use
2. Transparent: No hidden costs for retries, security, or AI features
3. Accurate: Reflects the true cost of processing your data

Signal Types Breakdown

┌─────────────────────────────────────────────────────────┐
│  YOUR WEBSITE                                           │
│  └─> Sends "Purchase" event                            │
│       ↓                                                 │
│  ANACOIC EDGE WORKER                                    │
│  ├─> Receive & validate (1 signal)                     │
│  ├─> Security check (1 signal)                         │
│  ├─> Dispatch to Meta (1 signal)                       │
│  ├─> Dispatch to TikTok (1 signal)                     │
│  └─> Dispatch to Google (1 signal)                     │
│       ↓                                                 │
│  Total: 5 signals for 1 event to 3 destinations        │
└─────────────────────────────────────────────────────────┘

How to Optimize Signal Usage

1. Use Client-Side Filtering

Don’t send unnecessary data:

// ❌ Sends test events
anacoic.track('Test Event');

// ✅ Only production events
if (!window.location.hostname.includes('localhost')) {
  anacoic.track('Purchase', { value: 100 });
}

2. Batch High-Frequency Events

Instead of sending every scroll event:

// ❌ Too many signals
window.addEventListener('scroll', () => {
  anacoic.track('Scroll');
});

// ✅ Batch engagement
let scrollDepth = 0;
window.addEventListener('scroll', () => {
  const newDepth = Math.floor(window.scrollY / window.innerHeight);
  if (newDepth > scrollDepth) {
    scrollDepth = newDepth;
    if (scrollDepth % 5 === 0) { // Every 5 viewports
      anacoic.track('Scroll Depth', { depth: scrollDepth });
    }
  }
});

3. Use Destination Filtering

Only enable destinations you need:

// ❌ Dispatches to all configured destinations
anacoic.track('Internal Event');

// ✅ Only specific destinations
anacoic.track('Purchase', { value: 100 }, {
  destinations: ['meta', 'google'] // Skip TikTok
});

4. Monitor Your Breakdown

Check your signal breakdown in the dashboard:

1. Go to Dashboard → Usage
2. View “Signal Types” breakdown
3. Identify optimization opportunities

Signal Type	Typical %	Optimization
Delivery	70-80%	Essential
Security	10-20%	Reduce bot traffic
Agent	5-10%	Limit unnecessary MCP calls
Retry	5-10%	Improve data quality

Signal Counting Examples

Example 1: E-commerce Purchase

anacoic.track('Purchase', {
  value: 150,
  currency: 'USD'
});

Signal count: 4 signals

• 1 delivery to Meta
• 1 delivery to Google
• 1 delivery to TikTok
• 1 security validation

Example 2: AI Agent Query (MCP)

GET /v1/mcp/sse?gid=your-gateway

Signal count: 1 signal per request

• Resource read = 1 signal
• Tool invocation = 1 signal

Example 3: Blocked Bot Traffic

A bot hits your endpoint without proper authentication.

Signal count: 1 signal

• Security block = 1 signal

Signal Pricing by Tier

Tier	Price	Included	Best For
Trial	Free	10,000 signals	Testing
Core	$0.001/signal	Pay per use	Low volume
Pro	$49/month	100,000 signals	Growing businesses
Max	$199/month	500,000 signals	High volume

Cost Examples

Monthly Signals	Core Cost	Pro Cost	Max Cost
5,000	$5 (minimum)	$49	$199
50,000	$50	$49	$199
100,000	$100	$49	$199
500,000	$500	$449	$199
1,000,000	$1,000	$949	$599

Frequently Asked Questions

Why am I being charged for blocked traffic?

Blocked traffic still requires our edge workers to:

1. Receive the request
2. Validate the origin
3. Check domain locks
4. Apply bot detection

This processing has real infrastructure costs.

Do retries count as extra signals?

Yes. Each retry attempt is a separate signal because it requires additional processing. To minimize retries:

• Ensure your pixel IDs and tokens are correct
• Send complete event data (improves match quality)
• Monitor destination health in your dashboard

How do I reduce my signal count?

1. Filter test traffic: Don’t send events from localhost/staging
2. Batch events: Group related actions instead of sending individually
3. Optimize destinations: Only enable platforms you actively use
4. Improve data quality: Better data = fewer retries

What’s the difference between a signal and an event?

Aspect	Event	Signal
Definition	User action	Any processed data
Examples	Purchase, PageView	Delivery, Security, Agent
Counting	1 per action	1+ per infrastructure touch
Billing	Per event	Per signal

Can I see my signal breakdown?

Yes! Navigate to Dashboard → Usage to see:

• Total signals by type
• Daily/weekly trends
• Cost breakdown
• Optimization recommendations

Need Help?

If you have questions about signal billing:

• Dashboard: Check your usage breakdown
• Docs: Read our optimization guides
• Support: Contact us at [email protected]