Datadog Alternatives: 4 Tools That Cut the Bill (2026)
Summary
The best Datadog alternatives for a team trying to cut its observability bill are SigNoz, Grafana Cloud, New Relic and Better Stack. We priced all of them against one workload of 20 hosts, 400 GB of logs and traces and 20,000 metric series. SigNoz comes out closest to Datadog in scope and far cheaper, at roughly $293 a month against $920 for Datadog infrastructure plus APM alone.
Datadog is the default answer to "which observability tool", and the invoice is the usual reason people start looking at Datadog alternatives. SigNoz is our pick as the closest replacement: it keeps the single pane for logs, traces and metrics but bills by gigabytes instead of hosts. Grafana Cloud, New Relic and Better Stack win in narrower cases, covered below.
Why teams start looking for Datadog alternatives
You rarely leave Datadog because the product is bad. You leave because the bill has more levers than you can reason about. Infrastructure Pro is $15 per host per month on an annual plan, APM is $31 per host on top, and logs are billed per GB ingested and again per million events indexed. Custom metrics sit on a separate meter.
There is no permanent free tier either, only a 14-day trial. The pattern we see most: a team instruments a new service, adds a few thousand high-cardinality tags, and the next invoice jumps without anyone changing a dashboard.
That is the problem to solve. Everything else, from features to UI polish, is secondary.
What a 20-host bill looks like on each tool
Numbers beat adjectives, so here is one workload used for every tool: 20 hosts, 300 GB of logs and 100 GB of traces per month, and 20,000 active metric series scraped every 30 seconds. That is 86,400 samples per series per month, or about 1.73 billion samples in total. This is a model, not a measurement, and every vendor calculator will land slightly differently.
Datadog: Infrastructure Pro on 20 hosts is $300, APM on the same 20 hosts is $620. That is $920 per month before a single log is indexed and before custom metrics beyond the per-host allotment.
SigNoz Cloud: 400 GB of logs and traces at $0.30 per GB is $120. Metrics at $0.10 per million samples is about $173. Total: roughly $293 per month, with no host or seat multiplier.
Grafana Cloud Pro: the $19 platform fee, up to $130 for 20,000 series at $6.50 per 1,000, $120 for logs at $0.40 per GB and about $40 for traces. That is up to $309 before the free allowances are subtracted.
New Relic: 400 GB minus the free 100 GB is 300 GB at $0.40, so $120. Four full platform users on Standard cost $10 plus three at $99, so $307. Total: about $427, excluding the volume of metric data.
Better Stack: ingestion for logs and traces runs from $0.10 per GB depending on region, so roughly $40 to $60 for 400 GB, plus retention, metrics and one $29 responder licence. Call it $70 to $100.
The ranking changes with your shape. A team with few hosts and heavy logs gets a different picture than one with many small hosts and light telemetry. Run your own numbers before trusting ours.
Which alternative fits which reason for leaving
You want the same single pane, minus the host tax
This is SigNoz. It is OpenTelemetry-native, covers APM, logs, traces, metrics and exceptions in one UI, and it has an automated tool to import Datadog dashboards. Teams are unlimited on the cloud plan, and the open source edition gives you an exit if the vendor ever changes terms.
The honest limits: the integration catalog is smaller than Datadog's, and fine-grained RBAC and audit logs are not fully shipped yet. If you self-host, you also take on running a ClickHouse-backed cluster.
You already run Prometheus and OpenTelemetry
Grafana Cloud fits. The free tier is real (10K series and 50 GB each of logs, traces and profiles), billing is itemized, and nothing is proprietary at the agent level. Our earlier head-to-head, Datadog vs Grafana for a 5 to 50 engineer team, goes deeper on that pairing.
The catch is assembly. Correlating metrics, traces and logs takes more wiring than a single-agent SaaS, and active-series pricing has its own cardinality traps on Kubernetes.
Your team is two to four people
New Relic's permanent free tier (100 GB a month and one full platform user) is the most generous way to get real APM without paying. Pricing is by data and seats, not hosts, which suits small teams.
It stops being cheap when everyone needs a full seat. At $99 to $349 per full user, a ten-person team feels it quickly, and ingestion stops at 100 GB until you upgrade or the month resets.
You care more about uptime and on-call than deep APM
Better Stack bundles uptime monitoring, on-call, status pages, logs and error tracking, with spending alerts and limits built in. It can import Datadog and Grafana dashboards, and its error tracking accepts Sentry SDKs.
It is not a full Datadog replacement for deep tracing and profiling, and it is not HIPAA compliant at the time of writing. For a lean team that mostly needs to know when something is down and who is paged, that trade is often fine.
Migration cost nobody puts in the pricing table
Switching tools costs engineer hours, and those hours rarely appear in vendor comparisons. Three things decide the real cost.
First, instrumentation. If your services already emit OpenTelemetry, moving is mostly pointing a collector at a new endpoint. If you rely on the Datadog Agent's auto-discovery and proprietary integrations, expect a few weeks of re-instrumenting.
Second, dashboards and alerts. SigNoz and Better Stack both ship Datadog dashboard import, which handles the bulk of simple boards. Anything built on Datadog-specific query functions needs rewriting by hand.
Third, the overlap period. Run the new tool next to Datadog for at least one full billing cycle. You will pay twice for a month, and you will learn which alerts you actually trust before cutting over.
When staying on Datadog is the right call
Datadog is not a bad choice, just an expensive one with clear strengths. Agent auto-discovery gets working dashboards within minutes, the breadth across infrastructure, APM, logs, RUM and synthetics is hard to match in one query language, and Bits AI SRE pulls context together when an alert fires.
If your team is small, your host count is modest, and an hour of an engineer's time costs more than the monthly difference, staying put is defensible. The case for leaving gets stronger past roughly 50 hosts, or the first time custom metrics surprise you.
How we would choose
Start from your reason for leaving. Cost with the same breadth: SigNoz. Existing Prometheus habits: Grafana Cloud. A tiny team: New Relic. Uptime and paging first: Better Stack. Then price the 20-host model above with your own volumes, and run a trial in parallel before you cancel anything.
The bill is only half the decision. The other half is who on your team will own the new stack at 3 a.m.
At-a-glance
| SigNoz | Grafana | New Relic | Better Stack | |
|---|---|---|---|---|
| Pricing model | Pure usage: $0.30/GB for logs and traces, $0.10 per million metric samples, from $49/mo with $49 of usage included | $19/mo platform fee on Pro, then $6.50 per 1,000 billable series and $0.40/GB for logs | $0.40/GB of data beyond 100 GB free, plus per-user pricing for full platform seats | Per responder licence from $29/mo (annual), plus ingestion from $0.10/GB depending on region and retention |
| Modeled bill, 20 hosts | About $293/mo for 400 GB of logs and traces plus 20,000 series scraped every 30 seconds | Up to about $309/mo for the same workload, before free allowances are subtracted | About $427/mo for 400 GB ingest and 4 full platform users, excluding metric volume | Roughly $70 to $100/mo for logs and traces plus one responder licence, metrics extra |
| Free tier | No free cloud tier, but the open source edition is free to self-host | Free forever: 10K active series, 50 GB logs, 50 GB traces, 50 GB profiles | 100 GB ingest per month and one full platform user, permanently | 10 monitors, 1 status page, 3 GB logs and traces (3-day retention), 30 GB metrics |
| Seats and users | Unlimited teammates on the cloud plan, no per-user pricing | Platform fee plus usage, no per-host tax on dashboards or queries | Full platform users at $99 (Standard) or $349 (Pro), basic users are free | Unlimited team members for telemetry, responders pay for on-call |
| Self-hosting | Open source Community Edition, you operate the cluster | Open source LGTM stack (Loki, Tempo, Mimir) runs fully self-hosted | SaaS only | SaaS, with an option to host data in your own bucket as an add-on |
| OpenTelemetry fit | OpenTelemetry-native, built around OTLP from the start | Built on open standards: Prometheus, OpenTelemetry, Loki | Accepts OpenTelemetry data alongside its own agents | Collects metrics via Prometheus or OpenTelemetry |
SigNoz
- Logs, traces, metrics, exceptions and alerts in one OpenTelemetry-native tool
- No per-host, per-user or special custom-metric pricing
- Automated Datadog dashboard migration tool and an open source edition to fall back on
- Smaller integration ecosystem than a decade-old vendor
- Self-hosting means running a ClickHouse-backed cluster yourself
- Some enterprise features, such as fine-grained RBAC and audit logs, are still beta or coming soon
The closest replacement if your pain is the bill and your services already emit OpenTelemetry.
Grafana
- Generous free tier with 10K series and 50 GB each of logs, traces and profiles
- Open standards mean no proprietary agent lock-in
- Itemized usage billing with a fully self-hosted path if you outgrow the cloud
- Self-hosting means assembling Prometheus, Loki and Tempo yourself before the first dashboard exists
- Cross-signal correlation takes more manual wiring than a single-agent SaaS
- Active-series pricing has its own cardinality traps on Kubernetes workloads
Best when you already speak Prometheus and want cost control over polish.
New Relic
- Permanent free tier with the full platform, including APM, logs and synthetics
- Pricing follows data volume and seats, not host count
- One query language (NRQL) over all telemetry in a single data store
- Ingest stops at 100 GB a month on the free tier until you upgrade or the month resets
- Full platform seats at $99 to $349 add up quickly on larger teams
- The free tier includes only one full platform user
Good for a team of two to four, less so once everyone needs a full seat.
Better Stack
- Uptime monitoring, on-call, status pages and telemetry in one product
- Imports Datadog and Grafana dashboards, with Sentry SDK-compatible error tracking
- Spending alerts and limits, so ingestion cannot run away silently
- Not HIPAA compliant at the time of writing
- APM and profiling depth is lighter than Datadog
- Extras such as additional status pages, SSO and audit logs are billed separately
A strong fit for lean teams, but not a full Datadog replacement for deep APM.
Verdict
If the goal is to keep a single pane and cut the bill, SigNoz is the closest swap: same breadth, billed by gigabytes instead of hosts. Pick Grafana Cloud if you already run Prometheus, New Relic if your team is small enough for its free seat, and Better Stack if uptime and on-call matter more than deep APM. Stay on Datadog when time-to-dashboard is worth more to you than the invoice.
How we tested
This comparison is desk research, not a lab test. Prices come from each vendor's public pricing page and our product catalog, checked on 23 and 24 October 2026. The modeled bill uses one stated workload (20 hosts, 300 GB of logs and 100 GB of traces a month, 20,000 metric series scraped every 30 seconds) and shows the arithmetic, so you can swap in your own numbers. Customer ratings come from G2, Capterra and Gartner Peer Insights roundups. Scores run from 0 to 5 and blend those ratings with fit for cost-sensitive teams of 5 to 50 engineers. We have no affiliate link for any product on this page, so nothing here is ranked for commission.