skills.rest
PostHog
PostHogOFFICIAL
Explore all skills by this author →
Website𝕏@PostHogGitHub

triaging-web-analytics-support

Triage web analytics support tickets into diagnoses, reply drafts, and fix PRs.

713|118|Updated Aug 11, 2020
Search Tags:#sql#playwright#web analytics#Debugging#PostHog#Zendesk#support triage
One-click install
npx skills add https://github.com/PostHog/posthog-foss --skill triaging-web-analytics-support
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: triaging-web-analytics-support
Source: https://github.com/PostHog/posthog-foss/tree/main/.agents/skills/triaging-web-analytics-support
Command: npx skills add https://github.com/PostHog/posthog-foss --skill triaging-web-analytics-support

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

Support queues for web analytics products mix explainable metric semantics with real bugs, and manually investigating each ticket across ticket systems, error tracking, and production data is slow and inconsistent. This Skill turns a pile of open tickets into grounded reply drafts and draft fix PRs.

Core Features & Use Cases

  • Queue Enumeration: Query open tickets from the in-app conversations product and Zendesk mirrors via SQL, including full comment history extraction and requester-to-org resolution across US and EU regions.
  • Shape-Based Diagnosis: Classify each ticket into one of six diagnostic shapes (frontend crash, metric mismatch, count drop, tracker loading, ad-platform integration error, channel misclassification) and run the matching playbook with worked evidence chains.
  • Artifact Production: Generate reply drafts grounded in file:line references or query results, plus draft fix PRs that follow public-repo safety rules (no customer names or traffic volumes).
  • Use Case: A support engineer asks to triage the web analytics queue; the Skill enumerates open tickets, diagnoses a reported pageview drop as bot traffic blocked at the customer's edge, and drafts a reply explaining the finding.

Quick Start

Triage the open web analytics support tickets and draft replies for each one.

Frequently Asked Questions about triaging-web-analytics-support

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I triage web analytics support tickets?▼

Enumerate open tickets from the conversations product via SQL against system.support_tickets, classify each into a diagnostic shape like frontend crash or count drop, then run the matching playbook. Produce reply drafts grounded in code references and draft PRs for confirmed bugs.

Why do two PostHog web analytics numbers not match?▼

Mismatches are usually scoping semantics: overview tiles count sessions with any matching event, while breakdown rows scope by session entry values like $entry_utm_campaign. Align the filter property with the breakdown definition before treating it as a bug.

How do I investigate a pageview count drop in PostHog?▼

Pull the raw daily event counts first to determine which layer dropped, since query-time exclusions never alter stored counts. A $pageview drop with flat $pageleave concentrated in a few user agents indicates bot traffic stopped executing the SDK, often due to edge or WAF changes.

Can I audit why PostHog undercounts compared to another analytics tool?▼

Yes, run a runtime loading audit with Playwright against the live site, checking load method, first-request timing, and ad-blocker survival. GTM-delivered scripts die when googletagmanager.com is blocked even with a proxied api_host, and consent latency loses quick bounces.

Why does a channel show as Direct instead of the correct source?▼

Channel classification is query-time via channel_definitions.json, and unknown utm_source values with stripped referrers fall through to Direct. Fix by adding definition rows and rebuilding the ClickHouse dictionary, not by changing the pinned fallback behavior.

What are the limitations of automated support ticket triage?▼

The keyword filter on last_message_text misses differently phrased tickets, so the Slack mirror channel must also be read. EU customer data is not queryable from the US MCP project, and headless-browser audits cannot observe real event capture since SDKs suppress automation.

Related Skills

Explore More Software Engineering Skills→
djinilabsdjinilabs
8

Zendesk Customer Context

Retrieve and summarize customer ticket history and context from Zendesk.

Official
Intermediate
ReviewStageReviewStage
259

fixing-pr-comments

Triage unresolved GitHub PR review threads and apply fixes via GraphQL.

Official
Advanced
langfuselangfuse
34.0k

analyze-cloud-costs

Analyze Langfuse Cloud infrastructure costs using Metabase cost marts and dashboards.

Official
Intermediate
martinjlowmmartinjlowm

zendesk-ticket

Retrieve Zendesk ticket metadata and full comment threads with attachments.

Community
Advanced