← Back to Datadog Bill Estimator
A Datadog invoice is a sum of separately metered products: hosts, containers, custom metrics, log ingestion, log indexing, APM hosts and synthetic test runs. The useful question is not "how do I spend less" but "which line, changed by how much, saves the most". That is arithmetic you can do before touching production.
Log indexing is billed per million events and priced by retention (roughly $1.06 for 3 days up to $2.50 for 30 days at approximate list prices). Health checks, debug lines and successful access logs are the usual bulk. An exclusion filter that drops 30% of 800 million indexed events saves 240 million × $1.70, about $408 a month at the 15-day price.
Moving 15-day retention to 7 days cuts the per-million price by about a quarter, and archived logs can be rehydrated when an incident needs them.
Storage that is not indexed is much cheaper per event, but querying it costs compute. The estimator takes your blended per-million figure as an editable assumption.
Custom metrics beyond the allowance per host are billed per metric, and each unique tag combination counts. Removing user, request or pod identifiers from tags is usually the biggest single cut. Prometheus Cardinality Cost finds those labels from a scrape.
Host products are billed per host, so idle hosts and many small hosts cost the same per host as busy large ones.
Datadog lists a lower annual price on host-based products. If you are on month-to-month and stable, the difference is pure arithmetic: at list prices about 17% on those lines. Commitment is a business decision, so weigh your own risk.
Enter your usage in the estimator or from the CLI, and gate a budget in CI:
node cli.mjs estimate usage.csv --budget 4000
Contracts, committed-use discounts, how Datadog measures hosts across the month, overage rules and price changes are outside it. Prices here are approximate list prices and editable. The estimate is arithmetic on your inputs, never a bill reader.
For most teams it is log indexing, custom metrics and host counts, in roughly that order of surprise. Estimate your own mix with the estimator rather than assuming, because a team with few logs and many containers has a different top lever.
No. An exclusion filter stops events being indexed (and billed for indexing) but they can still be ingested and archived, and ingestion is billed separately. Dropping them at the agent also saves ingestion cost, but they are gone from Datadog.
No. Excluding logs and moving them to archive act on the same events, so each lever is calculated alone and the list is a ranking, not a plan total.