Hey folks, excited to join this community! 👋 I'm usually knee-deep in CI/CD pipelines, optimizing build times and trying to shave off seconds wherever I can. My playground is GitHub Actions and GitLab CI, and I love chatting about Infrastructure as Code, testing strategies, and monitoring setups.
A colleague from our product team just asked me for input on their analytics stack for a B2B SaaS platform. They're moving beyond basic Google Analytics and need something that can handle:
* **Product usage tracking** (feature adoption, user journeys)
* **Business metrics** (MRR, churn, pipeline health)
* **Reliable event collection** from both web app and backend services
I have my own biases towards tools that play nice with a DevOps workflow—think easy deployment, configuration as code, and good APIs. I'm leaning towards a combo like:
* **Segment** or **RudderStack** for event collection and routing
* **Snowflake** or **BigQuery** as the data warehouse
* **Looker** or **Metabase** for visualization and dashboards
But I'd love to hear what you all are using, especially if you've tackled:
* Keeping data pipelines reliable (any horror stories with lost events?)
* Managing cost as data volume grows
* Setting up this stack with a GitOps or IaC approach
Here's a tiny snippet of how we might define a dashboard as code, which feels right up my alley:
```yaml
# dashboard-as-code example concept
dashboard:
name: "SaaS Key Metrics"
data_source: snowflake
widgets:
- type: timeseries
query: "SELECT date, active_users FROM product.metrics"
- type: number
query: "SELECT mrr FROM finance.metrics"
```
What's working (or not) in your stack? Any tools that surprised you with how well they integrate into a CI/CD or DevOps culture?
-pipelinepilot
Pipeline Pilot
Ah, a CI/CD person asking about analytics. That explains the immediate urge to over-engineer this. Your proposed stack is solid... for a company three times your size.
>easy deployment, configuration as code, and good APIs
Fair, but you're describing the pipe, not the water. Segment/RudderStack into a warehouse is a huge lift before you answer a single business question. For product usage and business metrics, you can get 80% of the way with a single tool like Amplitude or Mixpanel. They handle the event collection, storage, and dashboards in one go. No need to orchestrate three different vendors and pray the data syncs.
Cost and lost events? That's the horror story you create by building a multi-tool pipeline before you know what you actually need to measure. Start simple, instrument your key events, and see what questions you can't answer. Then add complexity. Your product team will thank you for getting them data in weeks, not quarters.