Published on 2025-05-18 5:04 by Kevin Coyle

Analytics? Why Not

The project evolved from a deploy-ready full-stack app to an instrumented platform with backend analytics. Here’s a breakdown of the key improvements and why they matter.

Highlights

• Snowplow analytics added to the backend
  • New tracking module centralizes analytics configuration:

import os
from snowplow_tracker import Tracker, Emitter

emitter = Emitter(os.environ["SNOWPLOW_COLLECTOR"])
tracker = Tracker(emitter, namespace="backend", app_id="dude-scouts")

I first came across Snowplow when I was working as the Data Engineer/Data Scientist/MLE/Salesforce Dev at Alo Yoga. I was pushed to learn to use this when I mentioned that I wanted to build a data warehouse that contained click level data from the ecom site. I was told they didn’t have budget, which is all gucci. Restriction is the mother of all invention or whatever. Anywho, now I know how snowplow works and as someone who always loves the idea of a cool dashboard, I figured we could collect user analytics with Snowplow. Also, I could keep my chops a little sharper.

• First event wired into the API root to validate pipeline:

from app.core.tracking import tracker

@app.get("/")
async def root():
    tracker.track_struct_event("api", "root")
    return {"message": "Dude Scouts API", "environment": settings.environment}

• Safer, environment-aware configuration

  • My config.py got formatting cleanups and clarified environment handling for CORS and LocalStack endpoints.
  • Keeps dev/prod behaviors cleanly separated (LocalStack in dev, real AWS in prod).

• Backend endpoints enhanced with analytics signals

  • Community API now emits a structured event when a post is created:

from app.core.tracking import tracker

# inside create_post(...)
tracker.track_struct_event("community", "create_post", label=str(row["id"]))

• Frontend community UX refresh

  • src/app/community/page.tsx and [questionId]/page.tsx were modernized:
    • Added Navigation, improved layout/typography, and adopted shared UI components (Card, Textarea, Input).
  • AnswerForm.tsx, AnswerList.tsx, QuestionForm.tsx switched to design-system inputs for consistency and better accessibility.

• Test and dependency alignment

  • Backend requirements and pyproject updated to include Snowplow and keep test fixtures aligned.
  • My conftest ensures a sensible default for the collector in test runs.

Why this matters

• Observability from day one: With Snowplow wired in at the API edge and critical write paths, we gain immediate visibility into user activity and API health without waiting for frontend-only data.
• Deployment confidence: The earlier commit ensured Docker-first deployments; I added some commits that adds the analytics layer that teams typically need once traffic starts.
• DX and UX upgrades: Cleaner configuration boundaries reduce surprise behaviors across environments, while the refreshed community UI makes the feature more discoverable and pleasant to use.

What to configure after pulling

Some stuff I need to do, now that I’ve added in Snowplow

• Verify a Snowplow collector endpoint is reachable from the backend.
• Expand event coverage (votes, answers, errors) using tracker.track_struct_event(domain, action, {label, property, value}).
• Add alerts/dashboards on Snowplow events to monitor traffic and key flows post-deploy.
• Confirm CORS origins in prod

Summary

• Added backend analytics via Snowplow with a dedicated module and initial events.
• Updated environment docs for dev/prod collector configuration.
• Improved backend config clarity and frontend community UI consistency.
• Positioned the app for data-informed iteration right after deployment.

Written by Kevin Coyle