If you run any kind of production application, you eventually need a way to see what's actually happening inside it — which requests are slow, which service is throwing errors, where a bug is coming from. Datadog became the default answer to that for a lot of teams because it does almost everything in one dashboard. The catch shows up later: pricing scales with hosts and usage, and it can climb quickly once you're past a small setup, plus the free tier is very limited. That's why a lot of engineering teams go looking for alternatives — either to cut cost, to self-host and keep control of their data, or just because their needs are narrower than Datadog's everything-included platform. Here are three real ones, with what each actually costs and who it fits.
Honeycomb
Honeycomb is built around one idea: when something breaks in production, you shouldn't have to guess why. Its BubbleUp feature automatically finds the correlations behind an anomaly instead of making you dig through dashboards by hand, and it now includes native LLM/agent observability for teams running AI systems in production.
Pricing: free up to 20M events/month; Pro from $150/month up to 750M events/month; Enterprise custom. Every paid tier includes unlimited seats and query capacity — no per-seat pricing to budget around.
Best for: software engineers and SRE/platform teams doing deep production debugging who want a SaaS tool close to Datadog's polish.
Limits: no free self-hosted option — it's cloud/SaaS only — and event-volume pricing can still climb fast for high-traffic systems, which is the same trap Datadog sets, just on a different curve.
SigNoz
SigNoz is the option for teams whose real objection to Datadog is cost and control, not features. It's open-source (MIT license), OpenTelemetry-native, and you can self-host the entire thing with no data caps for free.
Pricing: Community Edition is free and self-hosted with no caps. Cloud Teams starts at $49/month plus usage ($0.30/GB for traces and logs, $0.10/million metric samples). Enterprise starts around $4,000/month.
Best for: platform engineers, DevOps and SRE teams who want to self-host and control cost directly rather than pay a usage-based SaaS bill.
Limits: self-hosting ClickHouse at scale is real ongoing ops work, not a set-and-forget deployment, and the full enterprise feature set still requires the paid tier.
Highlight.io
Highlight.io comes at this from the frontend side: it combines session replay, error tracking, logging and tracing in one self-hostable platform, so you can watch what a user actually did right before your backend threw an error.
Pricing: free up to 500 sessions/month; pay-as-you-go from $50/month; Business from $800/month; Enterprise and self-hosted Enterprise custom.
Best for: frontend and fullstack teams who want session replay tied directly to backend errors and traces, not just backend-only APM.
Limits: its session replay is less polished than pure-play specialists like FullStory, and its APM depth is shallower than dedicated tools like SigNoz or Honeycomb if backend tracing is your main need.
Side-by-side
| Tool | Pricing model | Self-hosted option | Best for |
|---|---|---|---|
| Datadog (for reference) | Per-host, from ~$15/mo/host | No | All-in-one, if cost isn't the constraint |
| Honeycomb | Free up to 20M events, then usage-based | No | Deep production debugging |
| SigNoz | Free self-hosted (MIT), or usage-based cloud | Yes | Cost control via self-hosting |
| Highlight.io | Free up to 500 sessions, then usage-based | Yes | Frontend session replay + backend errors together |
If cost and control are what's actually driving you away from Datadog, SigNoz is the most direct answer — free to self-host with no caps. If you want a SaaS experience close to what you're used to but with unlimited seats instead of per-seat pricing, Honeycomb is the closer match. And if your specific pain point is understanding what a user did before something broke on the frontend, Highlight.io solves that more directly than any general-purpose APM tool will.