---
title: "Safe experimentation workflow | Grafana Labs"
description: "How to safely roll out configuration changes"
---

> For a curated documentation index, see [llms.txt](/llms.txt). For the complete documentation index, see [llms-full.txt](/llms-full.txt).

## Safe experimentation workflow

1. Create pipeline
2. Set matcher for ONE test collector (e.g., collector.ID=“test-collector-01”)
3. Activate and verify (no errors, acceptable performance)
4. Expand matcher to small group (e.g., env=“staging”)
5. Verify at scale
6. Expand to full target (e.g., env=“production”)

## Why this is safe

> **Isolation + Fallback + Instant rollback + Staged rollout = Safe to experiment**

You’re not gambling with production. You’re making controlled changes with multiple safety nets. Start small, verify, expand. If something goes wrong at any step, deactivate and try again. Recovery is fast and doesn’t require deployments or restarts.

## Coming up next

You’ll categorize your collector fleet in the learning path. You’ll assign attributes like `team`, `service.name`, and `env`, the foundation that makes precise configuration and safe rollouts possible.