Datadog does not have a price. It has a price per meter, and most teams run five or six meters at once. That is why two companies with the same headcount can get invoices that differ by an order of magnitude, and why the most common Datadog thread on Reddit is a screenshot of a bill nobody saw coming.
Short answer: Budget Datadog as a sum of separate line items, not a seat price. Infrastructure monitoring starts at $15 per host per month, APM adds $31 per host, and logs are metered per GB. A 50-host production setup with APM and logs can land anywhere from a few thousand to $10,000 a month. If you have fewer than about 10 hosts, check whether the free tier or a self-hosted stack covers you first.
The figures below come from the pricing data we track on our Datadog profile, which we re-check periodically. Datadog changes meters and list prices fairly often, so confirm against their live pricing page before you sign anything.
How Datadog billing actually works
Most SaaS tools charge per user. Datadog charges per thing it watches. That sounds fair until you count how many things you have. Containers, hosts, log lines, traces, browser sessions and synthetic test runs each get their own meter.
| Product | Meter | Price we track |
|---|---|---|
| Infrastructure (Pro) | Per host per month | $15 |
| Infrastructure (Enterprise) | Per host per month | $23 |
| APM and tracing | Per host per month | $31 |
| Log Management | Per GB ingested | $0.10, plus an indexing charge |
| Real User Monitoring | Per 1,000 sessions | $1.50 |
| Free tier | Up to 5 hosts | $0, with 1-day metric retention |
Two things to notice. First, the host prices are the small part. A host running infrastructure plus APM is already $46 a month on Pro before a single log line is stored. Second, logs have two charges: you pay to ingest them and again to index (make them searchable) them. Teams that ship everything at debug level pay both on data they never look at.
What the tiers include
Free. Five hosts, one day of metric retention and core features. It is fine for kicking the tires or watching a side project. One day of retention means you cannot look back at last week's incident, which makes it a demo more than a monitoring plan.
Pro at $15 per host. This is the entry point for real use. You get 15-month metric retention and the full dashboard toolset. For a small team watching a handful of servers, it is reasonable money.
Enterprise at $23 per host. The jump from $15 to $23 buys machine-learning anomaly detection, custom metrics and SAML single sign-on. If your security team requires SSO, you are paying the Enterprise rate on every host whether or not you need the other features. That is a common source of unplanned spend.
Three example bills
These are rough illustrations built from the list prices above. Your usage will differ, and negotiated discounts for annual commits can shave a meaningful amount off.
Small startup, 8 hosts, infrastructure only. On Pro, that is 8 x $15 = $120 a month. Add APM on all 8 and you add $248, for $368 total. Reasonable, and the free tier does not cover you because you are over 5 hosts.
Mid-size product, 50 hosts, infrastructure plus APM. Infrastructure is 50 x $15 = $750. APM is 50 x $31 = $1,550. That is $2,300 a month before logs, RUM or synthetics. Switch to Enterprise for SSO and the infrastructure line becomes $1,150, so the total is $2,700.
The same 50 hosts with real log volume. If those services ship 2 TB of logs a month, ingestion alone at $0.10 per GB is about $200, which sounds cheap. The indexing charge is where it climbs, and it depends on how much you keep searchable. Add RUM on a busy site, say 500,000 sessions a month at $1.50 per 1,000, and that is another $750. This is how a 50-host environment ends up in the $5,000 to $10,000 range that users report.
Where the surprises come from
- ▸Verbose logging. A new release that logs every request body can multiply your ingestion overnight. Nothing breaks, so nobody notices until the invoice.
- ▸Dynamic infrastructure. Per-host pricing is hard to predict when autoscaling spins hosts up and down. Check how Datadog counts hosts for your setup (billing is based on usage over the month, not a fixed snapshot).
- ▸Meter creep. Each new product is a new meter. Adding synthetics, security monitoring and CI visibility feels like small decisions, but they add up.
- ▸SSO tax. Needing SAML pushes you to the higher tier across the board.
Datadog pros
- ▸Full-stack observability in a single platform: metrics, traces, logs, RUM and synthetics talk to each other.
- ▸600+ pre-built integrations cover nearly any technology you are likely to run.
- ▸Best-in-class Kubernetes and container monitoring.
- ▸Machine-learning anomaly detection helps cut alert noise on dynamic infrastructure.
- ▸Strong dashboard and SLO tooling.
- ▸Fully hosted, so there is nothing to patch or scale yourself.
Datadog cons
- ▸Extremely expensive at scale, because bills compound across products.
- ▸Log costs can spike without warning.
- ▸Per-host pricing is difficult to forecast for autoscaling fleets.
- ▸Steep learning curve to get full value from it.
- ▸Vendor lock-in once your dashboards and monitors are built.
- ▸Overkill, and often unaffordable, for small engineering teams.
How it compares to the alternatives
Grafana plus Prometheus is free and self-hosted, but you pay in engineering time. Someone has to run, upgrade and scale it. Past a certain size, that operational cost can exceed the Datadog bill, which is the argument Datadog's fans make. For a small team with one competent infrastructure engineer, the open-source route is often cheaper. You can read more on our Grafana profile.
New Relic sits in a similar price tier. Datadog generally has stronger container and Kubernetes support, while New Relic has a more generous free tier. If your main worry is the first six months of cost, that free tier matters.
Sentry covers error tracking and some performance monitoring at a much lower price. It does not replace infrastructure monitoring, but plenty of small teams pair it with a basic uptime tool instead of buying a full observability suite. See Sentry.
How to keep the bill under control
- Set usage alerts per product before you roll anything out, not after the first invoice.
- Filter logs at the source. Drop health checks and debug noise before they reach ingestion. Only index what you will actually search.
- Turn on APM selectively. Not every host needs tracing. Start with the services customers touch.
- Negotiate an annual commit once your usage is stable. Committed-use discounts are the main lever on list price.
- Review monthly. Look at the usage page, not just the total. One meter is usually responsible for most of any increase.
Who should buy it
Buy Datadog if you run dozens of hosts or a Kubernetes-heavy stack, you value one pane of glass, and your engineers' time is worth more than the license. It is genuinely good software, and for the right team the price is justified.
Look elsewhere if you have a handful of servers, a tight budget, or unpredictable log volume you cannot yet control. Start with the free tier or an open-source stack, and move up when the pain of running it yourself is greater than the invoice.
Our take: Datadog earns its reputation for quality and its reputation for bills. Treat the per-host price as the floor, model your log volume honestly before you commit, and put cost alerts in place on day one.
Pricing shown reflects the data we track and may not match Datadog's current list prices. Always confirm on the vendor's site. SaaSCompared may earn a commission from some links on this site, which does not affect how we rate products.