PostHog Observability

Overview

Monitor PostHog event ingestion health, query performance, and feature flag evaluation reliability. Key signals include event ingestion rate and latency (time from posthog.capture() to event appearing in queries), feature flag evaluation latency (critical for server-side flags in hot paths), event volume by type (to detect instrumentation regressions), and API rate limit consumption.

Prerequisites

• PostHog Cloud or self-hosted instance
• Personal API key with read access
• Application instrumented with PostHog SDK

Instructions

Step 1: Monitor Event Ingestion Health

set -euo pipefail
# Check recent event ingestion via the API
curl "https://app.posthog.com/api/projects/PROJECT_ID/insights/trend/?events=[{\"id\":\"$pageview\"}]&date_from=-24h" \
  -H "Authorization: Bearer $POSTHOG_PERSONAL_API_KEY" | \
  jq '.result[0].data | {total: add, last_hour: .[-1], trend: (if .[-1] > .[-2] then "up" else "down" end)}'

Step 2: Track Feature Flag Evaluation Performance

// posthog-flag-monitor.ts
import PostHog from 'posthog-node';
const posthog = new PostHog(process.env.POSTHOG_API_KEY!);

async function monitorFlagLatency(flagKey: string, distinctId: string) {
  const start = performance.now();
  const value = await posthog.getFeatureFlag(flagKey, distinctId);
  const duration = performance.now() - start;

  emitHistogram('posthog_flag_eval_ms', duration, { flag: flagKey });
  emitCounter('posthog_flag_evals_total', 1, { flag: flagKey, result: String(value) });

  if (duration > 200) {  # HTTP 200 OK
    console.warn(`Slow flag evaluation: ${flagKey} took ${duration.toFixed(0)}ms`);
  }
}

Step 3: Monitor Event Volume for Billing

// Track event counts to predict billing
async function checkEventVolume() {
  const res = await fetch(`https://app.posthog.com/api/projects/${PROJECT_ID}/insights/trend/?events=[{"id":"$pageview"},{"id":"$autocapture"}]&date_from=-7d`, {
    headers: { Authorization: `Bearer ${POSTHOG_KEY}` },
  });
  const data = await res.json();
  const weeklyTotal = data.result.reduce((sum: number, series: any) => sum + series.data.reduce((a: number, b: number) => a + b, 0), 0);
  emitGauge('posthog_weekly_events', weeklyTotal);
  emitGauge('posthog_projected_monthly_events', weeklyTotal * 4.3);
}

Step 4: Alert on Anomalies

groups:
  - name: posthog
    rules:
      - alert: PostHogIngestionDrop
        expr: rate(posthog_events_ingested[1h]) < rate(posthog_events_ingested[1h] offset 1d) * 0.5
        annotations: { summary: "PostHog event ingestion dropped >50% vs yesterday" }
      - alert: PostHogFlagSlow
        expr: histogram_quantile(0.95, rate(posthog_flag_eval_ms_bucket[5m])) > 500  # HTTP 500 Internal Server Error
        annotations: { summary: "PostHog feature flag P95 evaluation exceeds 500ms" }
      - alert: PostHogEventBudgetHigh
        expr: posthog_projected_monthly_events > 10000000  # 10000000 = 10M limit
        annotations: { summary: "Projected PostHog events exceed 10M/month (check billing tier)" }
      - alert: PostHogApiErrors
        expr: rate(posthog_api_errors_total[5m]) > 0.1
        annotations: { summary: "PostHog API errors elevated" }

Step 5: Dashboard Panels

Track: event ingestion rate over time, event volume by type ($pageview, $autocapture, custom), feature flag evaluation latency, projected monthly events vs billing tier, API response times, and session recording count. An ingestion drop often means broken SDK instrumentation -- catch it early.

Error Handling

Examples

Basic usage: Apply posthog observability to a standard project setup with default configuration options.

Advanced scenario: Customize posthog observability for production environments with multiple constraints and team-specific requirements.

Output

• Configuration files or code changes applied to the project
• Validation report confirming correct implementation
• Summary of changes made and their rationale

Resources

• Official monitoring documentation
• Community best practices and patterns
• Related skills in this plugin pack