Datadog cost audit
Before you switch off Datadog, cut the bill.
Most of a Datadog bill is usage nobody decided on. A free, read-only audit in seven days shows what to cut and what it saves. Keep Datadog and pay less, or switch with a much smaller footprint to move.
A free 30-minute demo with an engineer. No commitment.
What would you save?
Set the sliders to your monthly spend.Ranges reflect results in production accounts, up to 40% in some. The free audit gives you an exact figure before you spend anything.
YC W22
Coinshift 
Where it grows
Where a Datadog bill grows.
Datadog bills by host, by series, by event and by gigabyte, and every default favours the meter. These are the six lines we open first.
Custom metric cardinality
One metric tagged with a request id or a pod name becomes thousands of billable series at $0.05 each beyond the allotment. Nobody queries most of them.
Log indexing by default
Every line ingested is indexed unless someone says otherwise, at $1.70 per million events at 15-day retention. Debug and access logs belong in the archive.
Agents and APM on every host
The agent gets installed everywhere for convenience. Infrastructure is $15 to $23 per host and APM another $31, including machines nobody looks at.
Retention set once
Index retention and metric retention were chosen on day one and never revisited. Longer retention on logs nobody searches is pure cost.
Committed counts from last year
The renewal locked in host and APM counts that matched last year’s peak. Usage fell, the commitment did not.
Products enabled and forgotten
A trial of a security or RUM product that was never turned off, billed monthly, showing up as a line nobody owns.
Switch or stay
Switch or stay? An honest read.
A migration is a quarter of engineering time and a second bill while both run. It is the right call for some teams and an expensive way to fix defaults for others.
Stay and cut the bill when
- The pain is cardinality and indexing defaults. Both are fixable in weeks without touching a dashboard.
- Your team relies on Datadog integrations, monitors and SLOs it would have to rebuild elsewhere.
- Nobody on the team wants to run an observability stack alongside the product.
- The renewal is close. Cutting usage first lowers the commitment whatever you decide afterwards.
A migration makes sense when
- The team already runs Prometheus and Grafana for part of the estate and knows the operational cost.
- The cost is structural: log volume that should never be indexed, or a footprint Datadog’s per-host model punishes.
- A platform decision is being made anyway, such as a move to Kubernetes or a new region.
- You have a quarter to run both in parallel and the engineers to rebuild dashboards, monitors and runbooks.
Either way, the audit comes first. A smaller footprint is cheaper to keep and cheaper to move, and the report tells you which case you are in. Compare Datadog and Grafana in detail.
Renewal timing
Cut usage before the renewal, not after.
Datadog renewals lock in host, APM and custom metric counts from the months leading up to the date. If the cuts land 60 to 90 days before, the committed counts drop and next year reflects what you actually run. Land them the month after and you carry the old commitment for another year.
We plan the audit and the changes around your renewal date, prepare the usage evidence, and join the vendor call so the new commitment is negotiated on numbers rather than last year’s estimate.
How it works
Read-only on day one. A number by day seven.
- Day 1
Read-only keys
An API key and an application key with read scopes only. Nothing to install and nothing changes.
usage_readmetrics_readlogs_read_configmonitors_readdashboards_readapm_read
- Day 7
The report
Every Datadog product ranked by annual savings and blast radius: which series to drop, which indexes to route to archive, which hosts lose the agent, and what to renew at. Yours to keep.
- From week 2
Iris applies, engineers approve
On a fixed monthly fee, Iris makes the changes in small, reversible steps and a senior engineer signs off on each one. Savings are reconciled against your invoice every month.
Questions
About the Datadog cost audit.
We are mid-contract with Datadog. Is there anything to do?
Yes. Committed spend still hides waste in custom metrics, indexed logs and host counts. Cutting usage now lowers what you renew at, and any on-demand overage above the commitment disappears immediately. We plan the work around your renewal date.
Will alerts break when you cut logs and metrics?
No. Iris reads which metrics and log indexes your dashboards, monitors and SLOs actually query before anything is excluded. Changes go out in small, reversible steps, and a senior engineer signs off on each one.
If we decide to move to Grafana, do you migrate us?
Yes, as a scoped project after the audit. Iris works across Datadog, CloudWatch and Grafana, so the same team can run the parallel period and take over the new stack under the managed plan.
What access does the audit need?
A Datadog API key and an application key with read scopes only: usage, metrics, log configuration, monitors, dashboards and APM. We never read log contents. Delete the keys and there is nothing else to remove.
Does the audit cover AWS too?
Yes. Most teams audit both at once. AWS needs a read-only cross-account role with Cost Explorer and CloudWatch access, and the report covers both bills side by side.
What does it cost?
The audit is free and the report is yours to keep. We only propose the managed plan when the audit finds annual savings of at least 3 times the fee. If it does not, you pay nothing.
See what’s in your Datadog bill.
Book a demo with an engineer and leave with the audit scoped. Bring last month’s invoice and your renewal date for a first read on the call.
Estimate first with the Datadog pricing calculator, or read the Datadog vs Grafana comparison.
You’ll pick a time on the next page.