“I'm an engineering manager at a 300-person B2B SaaS company and we're renewing our observability contract. Our current vendor charges per host and per GB ingested, and costs are unpredictable month to month. I want to move to something with usage-based predictable pricing, ideally self-hosted so we're not subject to vendor price increases. We already instrument with OpenTelemetry. Which full-stack observability platforms offer the most predictable pricing model and how do they compare to Datadog on cost at our scale?”

Given your constraints:

• ~300-person B2B SaaS company
• Already standardized on OpenTelemetry
• Primary pain is unpredictable invoices, not missing features
• Preference for self-hosted to avoid vendor pricing changes

I would optimize for pricing simplicity first, feature parity second. The instrumentation work (OpenTelemetry) is already the expensive part—you've preserved portability.

Here's how I'd evaluate the current market.

Why Datadog becomes expensive

Datadog's challenge isn't that any one SKU is expensive—it's that pricing compounds across multiple independent dimensions:

• infrastructure hosts
• APM hosts
• log ingestion
• log indexing
• custom metrics
• RUM
• synthetics
• database monitoring
• network monitoring
• security products

A Kubernetes autoscaling event can increase hosts, custom metrics, and log volume simultaneously. Finance sees a variable bill that's difficult to forecast. Datadog also bills hosts based on high-water-mark usage rather than simple monthly averages in many plans. betterstack.com

1. SigNoz (my first recommendation)

For companies already using OpenTelemetry, SigNoz is probably the closest thing to "Datadog without Datadog pricing."

Advantages:

• OpenTelemetry-native
• logs, metrics and traces in one product
• ClickHouse backend
• OSS Community Edition
• commercial support available
• self-hosted or cloud
• no per-host pricing
• no custom metric pricing

Cloud pricing is largely based on telemetry volume rather than infrastructure count, while self-hosted is effectively just your infrastructure cost plus operational overhead. betterstack.com

For a 300-person SaaS company I'd expect:

• 2–4 engineers able to operate it part-time
• ClickHouse cluster
• object storage for longer retention
• Kubernetes deployment

Migration is relatively straightforward because you're already emitting OTLP.

2. Grafana LGTM

If you have a reasonably strong platform/SRE team, this is arguably the safest long-term investment.

Typical stack:

• Prometheus / Mimir
• Loki
• Tempo
• Grafana

Pros:

• completely open source
• essentially unlimited scale
• no vendor lock-in
• enormous ecosystem
• best dashboarding

Cons:

• more operational complexity
• you're operating 3–5 distributed systems instead of one

The upside is pricing becomes mostly:

cloud infrastructure storage engineers

That's extremely forecastable.

3. OpenObserve

Worth considering if logs dominate your bill.

Its design emphasizes storage efficiency and a single binary deployment.

Compared to Grafana:

• easier operations
• fewer moving parts

Compared to SigNoz:

• somewhat less mature as a complete APM platform
• simpler operational story

How they compare to Datadog on cost

Without your exact telemetry volumes I can only estimate, but for organizations around your size, relative costs often look roughly like this:

Those aren't licensing quotes—they're typical total cost of ownership including infrastructure, assuming you already have engineers capable of operating Kubernetes and distributed storage. Community comparisons consistently report substantial savings from self-hosting, though operational labor needs to be included in the equation. apiscout.dev

Don't underestimate operational cost

This is where many Datadog comparisons become misleading.

You're effectively trading:

Datadog

• very little operational effort
• expensive licenses

for

Self-hosted

• cheaper software
• more infrastructure
• ongoing platform ownership

For a 300-person engineering organization, that's often a good trade. You likely already have platform engineers maintaining Kubernetes, databases, CI/CD, and internal infrastructure.

Feature gaps versus Datadog

Datadog still leads in several areas:

• Real User Monitoring
• Session Replay
• Synthetic Monitoring
• Security products (SIEM, CSPM, CNAPP)
• Network Performance Monitoring
• Database monitoring
• AI-assisted investigation
• breadth of integrations

If you're primarily consuming:

• metrics
• logs
• traces
• service maps
• dashboards
• alerts
• OpenTelemetry

the gap is much smaller.

What I'd shortlist

If I were running your evaluation, I'd issue an RFP to just three options:

1. SigNoz Enterprise (self-hosted) — likely the strongest balance of Datadog-like experience, OpenTelemetry alignment, and predictable pricing.
2. Grafana LGTM (self-hosted) — best long-term control and lowest vendor lock-in if you have platform engineering capacity.
3. Stay on Datadog but renegotiate around committed usage only if your team depends heavily on advanced capabilities like RUM, session replay, and security products that would otherwise require multiple replacements.

One additional recommendation: before evaluating vendors, measure your current monthly OpenTelemetry output—GB/day of logs and traces, metric sample rate, cardinality, and retention. Since you've standardized on OpenTelemetry, those telemetry characteristics are now your real cost drivers. That makes it much easier to compare self-hosted infrastructure costs against Datadog's multi-dimensional billing model on an apples-to-apples basis.

You’re in a good position to move away from host-based pricing because you already have the hardest part done: OpenTelemetry instrumentation. That lets you evaluate backends independently instead of rewriting agents/instrumentation.

For a ~300-person B2B SaaS company, I’d separate the options into three buckets:

• Self-hosted OSS stacks → most predictable long-term cost, but you own operations.

• Usage-based managed platforms → predictable billing, less ops.

• Traditional SaaS observability vendors → easiest, but usually the least predictable at scale.

Shortlist: predictable-cost alternatives

PlatformDeploymentPricing modelCost predictabilityDatadog comparisonSigNozSelf-hosted or cloudInfra cost only if self-hosted; cloud priced by telemetry volume★★★★★Often dramatically cheaperGrafana Labs LGTM stackSelf-hosted or Grafana CloudInfra cost (self-hosted) or usage-based cloud★★★★★Usually lower, especially for logs/tracesHoneycombSaaSEvent/usage-based★★★★☆Can be cheaper if you prioritize tracesNew RelicSaaSData-ingest/user/usage models depending on plan★★★☆☆Often more predictable than host pricingDatadogSaaS onlyHost + feature modules + ingest★★☆☆☆Powerful, but cost tends to expand

SigNoz (+2 more citation)

1. SigNoz: closest “Datadog replacement” if you want ownership

SigNoz is probably the closest match to your stated goal:

• OpenTelemetry-native

• Logs + metrics + traces in one product

• Self-hostable

• No host-count pricing

• No vendor-controlled agent licensing

Their hosted pricing is based on telemetry volume rather than hosts, with published rates for logs/traces and metrics samples. Their self-hosted Community Edition removes SaaS ingestion fees entirely; your cost becomes compute, storage, and engineering time. SigNoz

For your company size, rough economics:

Assume:

• 100–300 production hosts/containers

• moderate SaaS workload

• 500 GB–2 TB/month telemetry after sampling

A self-hosted SigNoz deployment might look like:

CostEstimateCompute$300–$1,500/moStorage$100–$1,000/moOps time0.1–0.5 engineerVendor bill$0

The biggest variable is retention and log volume.

Tradeoff: you become responsible for ClickHouse scaling, upgrades, backups, and capacity planning.

2. Grafana LGTM stack: best if your team is comfortable operating infrastructure

The open-source stack from Grafana Labs is:

• Grafana dashboards

• Prometheus / Mimir for metrics

• Loki for logs

• Tempo for traces

It maps very naturally to OpenTelemetry. Grafana Cloud also offers usage-based pricing instead of host-based pricing. Grafana Labs

Typical architecture:

Applications | OpenTelemetry Collector | +--> Metrics → Mimir/Prometheus +--> Logs → Loki +--> Traces → Tempo | Grafana UI

Advantages

• Very low vendor lock-in

• Huge ecosystem

• Excellent Kubernetes/cloud-native fit

• Costs scale with infrastructure, not engineer count

Disadvantages

• It is a platform, not a single appliance

• Alerting, permissions, upgrades, and retention policies require ownership

For a 300-person engineering organization, this is often the “engineering maturity” choice.

3. Grafana Cloud: managed version of the above

If you want predictable pricing without operating storage clusters, Grafana Cloud is worth comparing.

Its model is consumption-based rather than host-based. Published pricing starts with a base subscription plus usage charges for telemetry processing/storage, with volume discounts.