Datadog Pricing (2026): The Real Cost Breakdown

Datadog pricing starts at $15 per host per month for Infrastructure Monitoring (annual billing), but that number tells you almost nothing about your actual bill. Datadog is not one product with one price. It is roughly a dozen products, each metered differently, and the bill is driven by the ones that scale with your data: log ingestion and indexing, custom metrics, APM indexed spans, RUM sessions, and synthetic test runs. A mid-market team can start at a few hundred dollars a month and pass $8,000 a month within a year without changing much about how they operate.

This guide breaks down every pricing dimension in Datadog with current numbers, shows where the hidden costs come from, walks through what a real bill looks like at different team sizes, and covers how to cut your spend. At the end, I cover where Hyperping fits as a flat-rate option for the uptime, status page, and on-call slice of what teams use Datadog for.

In this guide:

• What Datadog actually costs, product by product, with current figures
• The hidden costs and billing mechanics behind a high bill (high-water-mark hosts, ingest-and-index, custom metric cardinality)
• A realistic bill for a 100-person engineering team
• How costs scale for solo developers, growing teams, and enterprises
• Practical ways to cut your Datadog bill, plus a quick spend audit
• Where a flat-rate alternative makes sense, and where it does not

Quick summary

• Datadog has no single price. Infrastructure Monitoring starts at $15/host/mo (annual) or $18 on-demand, APM at $31/host/mo, and logs at $0.10/GB ingested plus $1.70 per million events indexed. The bill is dominated by usage-based add-ons, not the base host fee.
• Four hidden costs drive most bills: high-water-mark host billing, the custom metrics tax (every Prometheus or OpenTelemetry metric counts), paying twice for logs (ingest then index), and APM indexed span overages.
• A 100-person engineering team commonly pays around $8,000 to $9,000 a month (roughly $100,000 a year) across products. Enterprises with 500+ services regularly report bills over $500,000 a year.
• Datadog is genuinely good at unified observability with metrics, logs, and traces in one place. The cost is the most common complaint, even from reviewers who rate it five stars.
• If you mainly use Datadog for uptime checks, synthetic monitoring, status pages, and on-call, Hyperping covers that slice at a flat rate from $29/mo, with no per-host or usage metering. It does not replace Datadog's tracing, logs, or APM.

Pricing figures are based on Datadog's published rates as of June 2026 and may change. Always check Datadog's pricing page for current numbers.

Is Datadog pricing actually worth it?

Datadog's pricing page lists clean per-host figures that make the platform look affordable. For a handful of hosts, it is. The problem is that the base host fee is the smallest part of most bills. The real money is in the usage-based products layered on top, and those scale with your data volume, your traffic, and your architecture, not with a number you pick on a plan page.

That is the gap this guide maps. Teams rarely get surprised by the $15 host fee. They get surprised when log indexing, custom metrics, and span overages show up on month two, after the data has started flowing and the annual commitment is already signed. The pattern is consistent enough that "I do not understand the Datadog pricing model" is a recurring thread on r/devops.

Datadog is worth it for teams that need deep, correlated observability across a complex distributed system and have the budget and the discipline to manage usage. It becomes a poor deal for teams that adopted it for a narrower job, like uptime checks and a status page, and are paying observability-platform prices for it.

Why listen to us?

I am Léo, founder of Hyperping. So yes, I have a stake here: Hyperping competes with a slice of what Datadog does. I am going to be straight about that throughout, including the parts where Datadog is the better tool and Hyperping is not a replacement at all.

To put this guide together, I went through Datadog's published pricing across every product, read dozens of G2 and Capterra reviews (the cost complaints are remarkably consistent), and pulled apart the billing mechanics that drive surprise invoices. Where different sources cite slightly different figures, I used Datadog's current published rates and noted the date. Hyperping sits at the simpler, lighter end of this market. It does uptime, server monitoring, status pages, and on-call well, and it does not do tracing, APM, or log analytics. I will not pretend otherwise, because the honesty is the point of a guide like this.

What is Datadog?

Datadog is a cloud monitoring and observability platform that brings infrastructure metrics, application traces, logs, security signals, and user-experience data into one interface. Founded in 2010 and public since 2019, it is one of the most widely used tools in the category, especially for cloud-native teams running containers, Kubernetes, and microservices.

Its appeal is breadth. With 650+ integrations and unified metrics, logs, and traces, it can replace a stack of separate tools and give DevOps and SRE teams a single place to detect and troubleshoot issues. That breadth is also what makes the pricing complicated: every one of those capabilities is a separately metered product.

Teams typically use Datadog for multi-cloud infrastructure monitoring, microservices and container observability, application performance troubleshooting, log analytics, security monitoring, and synthetic and real-user monitoring of frontend performance.

Datadog pricing breakdown

Datadog's model is modular and usage-based. There is a free tier (up to 5 hosts, 1-day metric retention), then two account tiers, Pro and Enterprise, that set your per-host rate and feature access. On top of that, you pay for each product you turn on.

Per-host account tiers

Pro includes 100 custom metrics and (on Infrastructure) 5 containers per host; Enterprise raises those allotments and adds machine-learning alerts, longer retention, and premium support. Neither number is your real bill. The products below are.

A simplified breakdown of the core products

This is the table most people actually want: each product, how it is metered, the starting price, and the catch.

A few of these deserve a closer look, because they are where bills actually grow.

Infrastructure: per host, billed on a high-water mark

Datadog counts your hosts every hour, discards the top 1% of hours, and bills the month at the next-highest count. Scale to 200 hosts for a 5-day event and you pay for 200 hosts most of that month, not your usual 50. In a containerized setup the trap is worse: Pro includes only 5 containers per host (10 on Enterprise), and every container beyond that is billed extra at about $1 each per month.

Log Management: you pay twice

Logs are billed on two axes. First $0.10 per GB to ingest, then $1.70 per million events to index them so they are searchable (15-day retention; $2.50 for 30-day). A team generating 500 GB of logs a day pays roughly $1,500/mo just to ingest, before indexing. To keep the bill down you index a fraction, which means most of your logs are not searchable during an incident, exactly when you need them.

Custom metrics: the cardinality tax

You get 100 custom metrics per host, then pay about $5 per 100 metrics per month. The count is driven by cardinality: every unique combination of a metric name and its tag values is a separate billable metric. A standard Kubernetes cluster with Prometheus exporters can generate 10,000+ metrics before you write a line of your own instrumentation, and adding one high-cardinality tag (a customer ID, a session ID) can push the count into the thousands instantly. Crucially, metrics sent via OpenTelemetry are billed as custom.

APM: per host, plus indexed spans

APM is $31/host/mo, but the real cost is indexed spans. You get 1 million indexed spans per APM host, then $1.70 per million after that. A single busy microservice can generate billions of spans a month; even with 1% sampling, the overage across 30 services can run past $1,000/mo on top of the per-host fees.

Datadog's hidden costs and pricing caveats

After reading through the pricing mechanics and the review patterns, the hidden costs cluster into a handful of caveats worth knowing before you commit:

• High-water-mark host counting. A short scaling event sets your bill for the whole month.
• The custom metrics explosion. Prometheus, StatsD, and OpenTelemetry metrics all count. A cluster's standard exporters alone can generate ten thousand-plus metrics.
• Container density math. Efficient bin-packing that reduces node count can raise your bill if it pushes you over the per-host container allotment. You can be billed more for more efficient infrastructure.
• Log volume drift. Someone adds debug logging, traffic grows 30%, and the log bill doubles quietly because exclusion filters were never set.
• Cross-product duplication. The same request can be billed three times: once as a metric, once as a trace, once as a log line.

How a 50-host setup turns into an $8,000 monthly bill

Each caveat looks small on its own. The cost problem is that they compound. Here is an illustrative bill for a mid-market company with roughly 100 engineers, 50 services, and moderate traffic on Enterprise pricing, where no single line is shocking but the total is.

That is roughly $99,000 a year, and it is conservative: it leaves out Continuous Profiler, Cloud Security Management, Network Monitoring, Incident Management, and Session Replay. Enterprises with 500+ services regularly report annual bills over $500,000, and some cross $1 million.

What users say about Datadog's pricing

The pattern in reviews is unusually consistent: people like the product and flag the cost, often in the same breath. These are real reviews pulled from Datadog's G2 profile.

On unpredictability, from a 3.5-star review (Sep 2025): "The pricing model: costs scale quickly and unpredictably, especially with logs and additional features."

From a 4.5-star review (Jul 2025) that otherwise praised the platform: "Pricing can ramp up quickly, especially when enabling multiple features like log management, APM, and custom metrics."

From a 5-star review (Oct 2025), where cost was the only complaint: "The main downside of Datadog is its pricing. It can get quite expensive as your infrastructure and data volume grow. Managing costs can be tricky, especially when you're monitoring multiple environments."

And from a 5-star review (Jul 2025) on SMB affordability: "Compared to other competitive products, Datadog prices are a bit higher. Small and medium size businesses cannot buy this product easily."

The balanced read: reviewers who manage usage carefully (one 4.5-star reviewer specifically credited the "ability to fine tune custom metrics and logs ingested" for controlling cost) get real value. Teams that turn on products without a usage strategy get the runaway invoices.

Datadog pricing by team size

Most people reading a guide like this are the ones who own the bill: a CTO or VP of Engineering at a growing SaaS company, a DevOps or SRE lead evaluating the platform, or someone in platform or finance trying to forecast spend. The same price list lands very differently at each stage, so here is what to expect at three common ones.

Growing SaaS teams (the usual case)

This is the audience most of this guide is for, and the compound example above is the typical bill: 50 hosts, real log volume, custom metrics from Kubernetes exporters, APM across dozens of services, landing around $8,000 to $9,000 a month. The lever that bites is almost always logs and custom metrics, not the host count people budget for. This is also the stage where the bill starts growing faster than the infrastructure underneath it, which is the warning sign to watch.

Early-stage teams evaluating Datadog

A smaller team adopting Datadog for the first time usually starts on Pro at $15 to $18 per host and adds a couple of products. Ten hosts with light APM and some logs typically lands around $500 to $800 a month once it is set up. The free tier (up to 5 hosts, 1-day retention) is a trial, not a home for production. The thing to plan for at this stage is the trajectory: the products you enable now are the ones that compound later.

Enterprises and platform teams

At 500+ services with high traffic, multi-region infrastructure, security monitoring, and long retention, bills run into the hundreds of thousands per year, sometimes past $1 million. At this scale the spend justifies a dedicated effort to manage it, which is why large Datadog customers often have someone in platform or FinOps whose job is partly cost control.

How to reduce your Datadog bill

You can cut a Datadog bill meaningfully without leaving the platform. The biggest wins:

• Commit annually. Annual billing runs about 20% cheaper than on-demand for the per-host products.
• Set log exclusion filters and index selectively. Ingest broadly if you must, but only index the logs you will actually search. This is the single biggest log-cost lever.
• Sample APM spans deliberately. Decide which spans matter and sample the rest. Do this from day one, not after the first overage.
• Control custom metric cardinality. Use Metrics without Limits to choose which tags stay queryable, and drop high-cardinality tags you do not query on.
• Watch your container-to-host ratio. If it is above the included allotment (5:1 Pro, 10:1 Enterprise), you are paying container overages.

A quick spend audit

If you are already on Datadog, this 5-minute check usually surfaces 20% to 30% in savings:

1. Usage page: Plan & Usage, then Usage, for per-product consumption.
2. Custom metrics: Infrastructure, then Metrics Summary, for your active metric count.
3. Logs: the Estimated Usage dashboard, to see your ingest-to-index ratio.
4. Containers: compare container count to host count; above 5:1 means overages.
5. APM: APM, then Ingestion Control, to check sampling rates.

A flat-rate alternative: Hyperping

A note on alternatives in general before I talk about ours. The observability market has real competition now: self-hosted open source, managed platforms , and focused tools that cover one slice well. The right move depends on which slice of Datadog you actually use.

That is the honest frame for Hyperping. It is not a Datadog replacement for tracing, APM, or log analytics. If that is why you run Datadog, keep it. But many teams pay Datadog prices for a narrower job: uptime and availability checks, synthetic monitoring, a status page, and on-call alerting. For that slice, Hyperping does the same work at a flat, predictable rate.

The pricing contrast is the point. Hyperping is flat-fee, not per-host and not usage-metered:

A Hyperping plan bundles uptime monitoring, server monitoring, status pages, and on-call into one number. There is no high-water-mark host billing, no ingest-then-index double charge, no per-metric cardinality tax, and no span overages. For comparison, replacing the legacy stack Datadog-style teams often run alongside it (Pingdom for uptime, Statuspage.io for status, PagerDuty for on-call) costs roughly $650/mo; the Hyperping Business plan covers all three for $249/mo on annual billing.

This maps directly onto the Datadog caveats above. The per-host and high-water-mark mechanics do not exist here because pricing is not per host. The ingest-and-index log trap does not apply because Hyperping is not a log platform. Seat costs are bounded and explicit ($12/mo per extra seat on Business), not an open-ended usage line.

Choose Hyperping if you use Datadog mainly for uptime, synthetic checks, status pages, and on-call, and you want a bill you can predict to the dollar.

Stay on Datadog if you need correlated metrics, distributed tracing, log analytics, and APM across a complex system. Hyperping does not do those, and that is by design.

If the predictable-pricing case fits, you can start free (20 monitors, no time limit) or schedule a demo.

Frequently asked questions

How much does Datadog actually cost?

Infrastructure Monitoring starts at $15 per host per month on annual billing, but most bills are driven by usage-based products: logs ($0.10/GB ingested plus $1.70 per million events indexed), custom metrics ($5 per 100 over your allotment), APM ($31/host plus span overages), and synthetics. A 100-person engineering team commonly pays around $8,000 to $9,000 a month.

Why is Datadog so expensive?

Because costs scale with data, not with a fixed plan. High-water-mark host billing, per-metric cardinality charges, paying separately to ingest and index logs, and APM span overages all grow as your traffic and architecture grow. The base host fee is the smallest part of a typical bill.

Is Datadog free?

There is a free tier covering up to 5 hosts with 1-day metric retention, which is suitable for testing or very small projects. Production use generally needs the Pro tier ($15 to $18 per host) plus whatever products you enable, so the free tier is best treated as a trial.

What is the cheapest way to use Datadog?

Commit annually (about 20% off the per-host products), index only the logs you will search, sample APM spans, and control custom metric cardinality with Metrics without Limits. A quick audit of the Usage page usually finds 20% to 30% in savings.

What is a cheaper alternative to Datadog?

It depends which part of Datadog you use. For full observability, SigNoz, Grafana, and New Relic are common alternatives. If you mainly need uptime monitoring, synthetic checks, status pages, and on-call, Hyperping covers that at a flat rate from $29/mo with no per-host or usage metering. It does not replace Datadog's tracing, logs, or APM.

How is Datadog billed for hosts?

On a high-water mark. Datadog measures your host count hourly, drops the top 1% of hours, and bills the month at the next-highest count. A short scale-up event can therefore set your bill for the entire month.