The True Cost of Datadog in 2026
Datadog's public pricing is just the starting point. Here is a complete breakdown of what you actually pay, including the custom metrics trap, container billing nuances, and the AWS networking costs that most teams forget to budget for.
The complete bill breakdown
Datadog's pricing has multiple dimensions. Understanding each component is essential for accurate cost forecasting.
Infrastructure monitoring
| Tier | Annual | On-Demand | Custom Metrics/Host |
|---|---|---|---|
| Pro | $15/host/mo | $18/host/mo | 100 included |
| Enterprise | $23/host/mo | $27/host/mo | 200 included |
APM pricing
- APM: $31/host/month (annual), $36 on-demand
- APM Enterprise: $40/host/month (includes Continuous Profiler)
- AWS Fargate: $2-2.60 per concurrent task
- Span allocations: 150GB ingestion + 1M indexed spans per APM host/month
- Overage: $1.70/million indexed spans (15-day), $0.10/GB ingested spans
Log management
- Ingestion: $0.10/GB (uncompressed)
- Standard indexing: $1.70/million events (15-day retention)
- Flex Storage: $0.05/million events/month
Additional products
| Product | Price |
|---|---|
| RUM | $1.50/1K sessions |
| Synthetics (API) | $5/10K test runs |
| Synthetics (Browser) | $12/1K test runs |
| Database Monitoring | $70/host/month |
| Network Monitoring | $5/host/month |
| Container Overage | $1/container/month over allocation |
Custom metrics: the cost trap
This is where Datadog bills become unpredictable. According to SigNoz research, custom metrics can constitute up to 52% of total billing at scale.
How cardinality explodes costs
Consider a simple business metric: api_request_duration
You add reasonable tags for analysis:
- 10 endpoints
- 3 status code categories (2xx, 4xx, 5xx)
- 3 customer tiers (free, pro, enterprise)
That is 10 x 3 x 3 = 90 unique metric timeseries from one metric name.
Now add customer_id as a tag for debugging specific accounts:
- 1,000 active customers
- 90 x 1,000 = 90,000 unique timeseries
At $5/100 metrics overage (Enterprise tier), that one metric could cost$4,500/month in custom metrics charges alone.
OpenTelemetry metrics are custom metrics
This is the part that surprises teams migrating to OTel: all OpenTelemetry metrics are billed as custom metrics in Datadog. Your standard OTel instrumentation can quickly generate thousands of custom metrics.
The container billing trap
In Kubernetes environments, billing depends on how you deploy the Datadog agent:
- DaemonSet (correct): Agent runs per node, billed per node
- Per-pod (incorrect): Each pod billed as a separate host
We have seen teams accidentally deploy agents per-pod in their Helm charts, turning a 50-node cluster into a 500-host bill.
High-water-mark billing
Datadog uses 99th percentile billing for hosts. This means:
- A 5-day traffic spike from 100 to 200 hosts
- Results in paying the 200-host rate for the entire month
Autoscaling for resilience becomes autoscaling your observability bill.
The hidden networking costs
When shipping telemetry to any SaaS provider, you also pay AWS networking fees. These are often forgotten in observability budgets.
| Component | AWS Cost |
|---|---|
| NAT Gateway Processing | $0.045/GB |
| Internet Egress (first 10TB) | $0.09/GB |
| Total per GB | $0.135/GB minimum |
For a company shipping 1TB of telemetry per day:
- 30TB/month x $0.135 = $4,050/month in networking alone
- Annual networking cost: $48,600
This is not included in your Datadog contract. It is a separate AWS bill that often surprises finance teams.
Real-world bill estimates
| Company Size | Hosts | Monthly | Annual |
|---|---|---|---|
| Startup | 10-50 | $2K-15K | $24K-180K |
| Mid-Market | 100-500 | $20K-150K | $240K-1.8M |
| Enterprise | 1,000+ | $100K-1M+ | $1.2M-12M+ |
| Large Enterprise | 10,000+ | $500K+ | $10M+ not unusual |
Notable public examples
- Coinbase: $65 million annual Datadog spend in 2021 (confirmed in Datadog Q1 2023 earnings call by CFO David Obstler)
- OpenAI: Reportedly $170 million annually on Datadog
- ClickHouse: Found self-hosted observability was 200x cheaper than Datadog equivalent for their internal usage
How to audit your Datadog spend
- Review custom metrics: In Datadog, go to Metrics Summary and sort by cardinality. Identify which metrics are driving costs.
- Check container billing: Ensure agents are deployed as DaemonSets, not per-pod.
- Calculate networking costs: Check AWS Cost Explorer for NAT Gateway and Data Transfer charges. Filter by the instances running Datadog agents.
- Review log indexing: Are you indexing logs you never query? Consider using Flex Storage for cold data.
Frequently asked questions
Why is my Datadog bill higher than expected?
The most common causes are custom metrics cardinality explosion, container billing (each pod billed as a host if agent deployed per-pod), high-water-mark billing (5-day spike means peak rate for the month), and AWS networking costs not included in observability budgets.
What percentage of cloud spend should observability be?
Industry benchmarks suggest observability typically represents 25-35% of public cloud costs (Chronosphere estimate). If your observability spend exceeds this range, it may be worth evaluating alternatives.
How do custom metrics costs spiral out of control?
A single metric with dimensional tags can explode exponentially. 10 endpoints times 3 status codes times 3 customer tiers equals 90 unique timeseries. Add customer_id with 1,000 values and you have 90,000 metrics from one business metric. All OpenTelemetry metrics are billed as custom metrics.
The alternative
Sampleless uses flat annual pricing: $100K-275K based on cloud accounts, not data volume. No custom metrics surcharges. No per-host fees. No networking costs because data stays in your cloud.
Send us your current Datadog bill. We will show you exactly what Sampleless would cost for the same infrastructure.
Get a cost comparison for your Datadog bill
Send us your invoice. We will calculate your Sampleless pricing and show the math.