Datadog Pricing: Where the Bill Grows

Datadog does not have one price. It has a pricing model with a price list, and every product on it is billed on its own unit: hosts, containers, custom metrics, gigabytes ingested, log events indexed, spans indexed, test runs and sessions. The bill grows where one of those units tracks something your architecture multiplies, which is usually hosts that autoscale, tags that carry unbounded values and logs that nobody searches.

This post explains the billing units from Datadog's own documentation, read on 4 November 2026, so you can see where growth comes from. It prints no dollar amounts on purpose: Datadog's list-price page carries no date and rates differ by plan and commitment, so read them on datadoghq.com/pricing/list. If you are weighing a move rather than a cleanup, Datadog alternatives compared sets the options side by side, and AceMQ's Datadog consulting and support covers the platform you already run.

Per-host billing is predictable on a static estate and loose on an elastic one. For infrastructure hosts, Datadog's usage page reports the 99th percentile of distinct hosts across all hours in the month, which discounts a brief spike but not a sustained one. For containers it reports the high-water mark. Before negotiating anything, graph hourly host and container counts for three months. A node pool that scales up each night and is slow to scale down shows here first.

Datadog defines a custom metric as a unique combination of metric name and tag values, including the host tag, and counts the distinct ones for each hour before averaging across the month. A gauge with tags for 200 hosts, five endpoints and four status codes can reach 4,000 custom metrics, whether or not anyone queries them. Histograms count as five by default, and distributions as five, or ten once percentiles are added.

Pro plans include 100 indexed and 100 ingested custom metrics per host, and Enterprise 200 of each, pooled across hosts. Metrics without Limits separates what code sends (ingested) from what stays queryable after a tag allowlist (indexed), so dropping tags nobody filters on lowers the indexed count without changing the code. Pod names, request ids and customer ids are the usual offenders: they belong in logs or traces, not in metric tags.

Logs are billed twice. Ingestion is per gigabyte, and indexing is per million events, priced by how long you keep them searchable. Debug logs from a chatty framework are cheap to receive and expensive to index for 30 days. Exclusion filters, shorter retention for low-value indexes and Flex Logs for data you must keep but rarely search are the three levers inside Datadog.

APM follows the same pattern: a host fee for each APM host, a charge for gigabytes of spans ingested, and one for indexed spans by retention. Sampling keeps error and slow traces at full fidelity and thins routine successes, which is a design decision rather than a setting.

Datadog's list page shows three price columns: billed annually, month-to-month and on-demand. For most products the annual rate is the lowest, and month-to-month and on-demand are often the same. The page does not state per-host allotments for products such as container monitoring and custom metrics. It points existing customers to their own billing agreement, so the contract, not the list, decides what is included, and contracted rates can differ from the list. Datadog documents how custom metrics are counted in detail.

The order that avoids regret is to attribute volume by origin, remove what nobody queries, move unbounded identifiers out of tags, and only then ask whether the platform is still the right one. A typical engagement for that first step is described in Datadog custom metric and ingest cost governance. If the answer is still no, plan the move in parallel: instrument with OpenTelemetry first, run both systems side by side, and compare a Grafana, Prometheus, Loki and Tempo design with your bill by product, as in Datadog to Grafana stack move planning. AceMQ can then run the replacement as managed Grafana and managed Prometheus, in your account or ours.

• Datadog list prices page (billing units, annual, month-to-month and on-demand columns), read 4 November 2026. The page shows no publication date.
• Datadog custom metrics billing, read 4 November 2026.
• Datadog usage details, read 4 November 2026.