Use your metrics

Note: Metrics is in open alpha. Any team can turn it on — open Metrics and select Enable metrics in the onboarding view. The views on this page reflect the current product and may change before general availability.

Once metrics are flowing, open Metrics in PostHog. The metrics scene has three tabs:

• Overview — is anything arriving, and from which services
• Viewer — chart one metric, break it down, and pivot into logs and traces
• SQL — query your metrics directly as the posthog.metrics table

Overview

The Overview tab is the first place to check whether ingestion is healthy. It shows an ingestion rollup over the last day:

• Services, Metric names, and Active series counts
• A per-service table with each service's metric count, active series, and last seen time. A service that has gone quiet is flagged so you can spot a dropped exporter before it becomes an outage.

Click a service to jump into the viewer scoped to that service. If nothing has arrived yet, the tab links you to the setup guide.

Viewer

The Viewer tab is where you explore a single metric without writing any queries.

• Pick a metric by name to chart it over time.
• Choose an aggregation — sum, average, series count, min, max, p95, rate, or increase. The viewer recommends one based on the metric's type: increase for counters, avg for gauges, p95 for histograms and summaries.
• Group by an attribute to break the metric into one line per value, for example request rate per route or queue depth per worker.
• Filter with key=value chips to focus on one service, environment, or status, and adjust the date range.

From there you can shape the chart itself:

• Display type — line, area, or bar.
• Chart settings — add goal lines, switch the Y axis between linear and logarithmic, toggle begin-at-zero, and pin a min or max bound.
• Auto-refresh — refresh the chart every 15 seconds while you watch a deploy or an incident.

Spot what changed

When the metric you're charting moves, the viewer does the first pass of the investigation for you. If the most recent slice of the selected window is statistically different from the rest, an anomaly badge appears on the chart showing the percent change against the baseline and when the shift started. Click it to see the biggest movers — the label values that drove the change — and pick one to narrow the chart to it.

This is investigation, not alerting. The badge is computed when you open a chart — it helps you spot and explain a change you're already looking at. It doesn't watch your metrics around the clock or notify you; for that, set up an alert.

Pivot into logs and traces

A metric tells you something changed; logs and traces tell you why. The viewer keeps those one click away, scoped to the same service and time window:

• Exemplar markers — metrics sent with trace context show clickable dots along the chart baseline, one per traced emission, that open the trace directly.
• Samples — a side panel lists the raw emissions behind the chart. The Samples view links each emission to the trace it was recorded in; the Aggregates view shows per-series latest and total.
• Related — jump to the Logs, Error logs, or Traces for the chart's service and time window. Filter or group by a service first so the jump lands on the right slice.

Log-based metrics

A metric generated from your logs by a log-based metric rule carries a From logs tag. Click it to view the source logs the rule matches, or to open the rule in settings.

Query with SQL

The SQL tab gives you direct access to your metrics with SQL:

SELECT * FROM posthog.metrics LIMIT 10

Use it for anything the viewer doesn't cover: joining metrics against each other, custom math, or ad-hoc exploration of raw data points.

Save and share

When a chart is worth keeping, turn it into something durable:

• Save as insight saves the current chart as a reusable insight.
• Add to dashboard drops it onto an existing or new dashboard.
• New service dashboard builds a starter dashboard in one step — pick the metrics (optionally scoped to a service) and PostHog creates one insight per metric, each charted with its recommended aggregation.

Metrics table on a dashboard

A Metrics table widget puts one row per value of a label on a dashboard, with each column reading a different metric. Use it for the Grafana-style endpoint view: request rate, errors, and p95 latency side by side for every route, without building a separate insight for each one. New label values get a row automatically.

Note: The Metrics table widget is only available to teams enrolled in the Metrics alpha (the metrics feature flag). If you don't see it in the widget picker, you're not enrolled yet.

To add one:

1. Open a dashboard and click Add widget.
2. Choose Metrics table from the Metrics group.
3. Set the label to group rows by, for example http.route for one row per endpoint.
4. Add up to six columns. For each one, search for a metric and choose how to read it: a counter rate or increase, a histogram percentile such as p95, or an average, sum, min, or max.
5. Optionally add a label filter to a column. For an errors column, filter http.response.status_code with =~ 5...
6. Set the date range, sort column, and row limit (up to 100), then save.

Each cell shows the average over the date range with a sparkline. Sparklines in a column share a scale, so you can compare rows at a glance. Click a row value to open full charts for that row, or tick several rows and choose Open in full charts to compare them on one chart.

A few limits to know about:

• Shared and public dashboards show a placeholder instead of the table.
• The table keeps only as many rows as the row limit allows, and the sort column and direction decide which ones. With Highest first, the rows with the smallest values are dropped. With Lowest first, the rows with the largest values are dropped.
• The widget reads ingested metrics. A project that only sends traces or logs has nothing for it to show.

Alerting

Charts and dashboards are for when you're watching; an alert is for when you're not. To get notified when a metric crosses a threshold, save the chart as an insight first — metric alerts are built on insights, so they live alongside your other alerts on the insights and dashboards you already use.

1. In the viewer, click Save as insight (or Add to dashboard).
2. Open the saved insight, click Actions in the sidebar, and choose Alerts.
3. Click New alert, name it, and set the threshold you want to be notified about.

Alerts are threshold-based: you set a bound the metric must stay above or below, and PostHog notifies you (Slack, Microsoft Teams, or webhook) when it crosses. Because a metric alert is just an alert on a metric insight, it uses the same alerting platform and notification channels as every other insight.

See Alerts for the full setup, including thresholds and notification channels.

Next steps

• Get started with metrics
• Generate a metric from your logs
• How metrics works under the hood