Intro on Grafana Labs

Journey overview

Wed, 14 Oct 2026 15:16:42 +0200

Stop applying every recommendation manually

Adaptive Metrics looks at how you use your metrics and suggests changes to how much detail you keep for each one. Each change is a recommendation. Auto-apply applies new recommendations as they’re generated, so you spend less time on the list. The mode you choose decides whether it also applies recommendations that add series. This journey covers whether it fits, what to protect first, how it treats recommendations, and how to turn it on.

What you’ll learn

Here’s what this journey covers:

Whether it fits: Signals that show you’re ready to stop applying recommendations manually
What to protect first: Why to exempt the metrics and labels whose full detail you need before anything applies automatically
How it treats recommendations: The two ways it can handle recommendations that would increase what you store, and when each fits
Turning it on in the UI: Where you turn it on, starting from Overview
Other ways to turn it on: When to use the UI, the API, or Terraform
What you still control: Which settings and recommendations stay yours after it’s on

Journey duration

~20-30 minutes

Auto-apply overview

Wed, 14 Oct 2026 15:16:42 +0200

Auto-apply overview

Auto-apply is a setting that applies new recommendations as they’re generated, so you don’t have to open the list and choose which to apply. The mode you choose decides whether recommendations that add series are applied too, so some recommendations can still wait for you to apply them. If you want to keep certain metrics as they are, create an exemption. It keeps a metric, or a label, out of recommendations. A label is a name and value pair attached to a series, such as service="checkout".

You turn auto-apply on for a segment. A series is a single list of values recorded over time, identified by its labels. A metric is all the series that share a metric name. A segment groups series by a label you choose, such as one team’s series, so one metric can have series in more than one segment. Adaptive Metrics recommends changes for each metric in each segment, so the same metric can get different recommendations in different segments. The Default segment contains the series that aren’t assigned to a named segment.

For example, you turn on auto-apply for the Default segment. For the series in that segment, Adaptive Metrics then applies the new recommendations your mode allows as they’re generated. If you’ve exempted the latency metrics for your checkout service, auto-apply leaves those metrics as they are.

Why use it?

Less manual work: New recommendations are applied without you opening the list each time.
Keeps up with your usage: As Adaptive Metrics updates its recommendations, your segment stays tuned for cost and observability.
Control by segment: You turn it on for the segments you choose and leave it off for the rest.

Is auto-apply right for you?

Wed, 14 Oct 2026 15:16:42 +0200

Is auto-apply right for you?

Check where you are with Adaptive Metrics before you enable auto-apply.

A good fit

Signal	Why it matters
You’re comfortable with the recommendations you’ve applied manually	Auto-apply then applies the new recommendations your mode allows, without manual review.
Exemptions protect your critical metrics	Exemptions keep the metrics or labels you choose out of recommendations, so auto-apply leaves them out too.
Your usage keeps changing	Adaptive Metrics updates its recommendations as your usage changes, and auto-apply applies the new ones your mode allows so you don’t review each one yourself.
Some of your segments are ready and others aren’t	Auto-apply is set for each segment, including Default, so you can turn it on for the ones that are ready and leave the rest as they are.

Not yet

Signal	Where it’s covered
You haven’t used Adaptive Metrics yet	Introduction to Adaptive Metrics
Your critical metrics don’t have exemptions yet	Use exemptions
You’re still deciding which recommendations to apply	Review and apply recommendations