---
title: "The investigation workflow | Grafana Labs"
description: "Search, filter, visualize, investigate, act on it"
---

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

## Metrics investigation workflow

This pattern guides you from a broad view to specific findings. A typical investigation moves through five steps, though you can jump between them or skip the ones you don’t need.

| Step            | What you do                                                                                              | What you get                                           |
|-----------------|----------------------------------------------------------------------------------------------------------|--------------------------------------------------------|
| **Search**      | Type part of a metric name in the search bar                                                             | List of matching metrics                               |
| **Filter**      | Use the filters to narrow scope by label values                                                          | Reduced set of time series                             |
| **Visualize**   | View the automatic graph of the selected metric                                                          | Time series visualization                              |
| **Investigate** | Use the Related metrics tab (metrics with similar names) and the Related logs tab (matching log streams) | Related metrics, and a pivot from a metric to its logs |
| **Act on it**   | Add to a dashboard, create an alert, save the view, copy the URL to share, or open in Explore            | Each action carries your query and time range forward  |

You don’t need to know exactly what you’re looking for when you start.

You also don’t need to aggregate or apply rate functions. Filter to what you care about and read the graph, which reveals trends and anomalies that raw numbers hide. You can compare multiple time series side by side, for example CPU across servers, to spot correlations and narrow down a root cause.