Why 84% of teams are replacing Datadog with simpler monitoring

The observability bill nobody planned for

According to the 2026 Cloud Infrastructure Survey by Flexera, 84% of engineering teams are actively working to reduce their observability and monitoring costs this year. That number was 61% in 2024. Something changed.

What changed is that the bills arrived. Datadog charges per host, per custom metric, per log line ingested, per APM span, and per indexed log retained. Each dimension seems reasonable in isolation. Together, they compound in ways that surprise even experienced teams.

A 20-person startup running 15 hosts with moderate logging and basic APM can easily hit $3,000-5,000 per month. A mid-stage company with 50 hosts, microservices, and the kind of logging that distributed systems demand routinely sees $50,000 or more per year. That is before anyone turns on custom metrics or log analytics.

The cost scales with your infrastructure, which means it grows exactly when you can least afford it — during rapid growth when every dollar should go toward product and hiring.

The "observability tax" — paying $50k for uptime checks

Here is the uncomfortable truth most teams discover after a year on Datadog: the vast majority of their alerts come from two categories.

That is it. For most teams, $50,000+ per year is buying them uptime checks and log search. The APM dashboards look impressive in the demo. Six months later, two people on the team have opened them, and one of those was by accident.

This is the observability tax: the gap between what you pay for and what you actually use. And in 2026, teams are closing that gap.

Simple monitoring vs full observability — which do you need?

This is not a question with a universal answer. It depends on your architecture, your team size, and what your users actually need from you. Here is a honest breakdown:

Most teams — especially those under 50 engineers — fall into the first category. They adopted full observability because it was the industry default, not because their architecture demanded it.

Why usage-based pricing punishes growth

Datadog's pricing model is built around usage dimensions: hosts, metrics, log volume, APM spans. Each one makes intuitive sense — you pay for what you use. The problem is that all of these dimensions grow together, and they grow with your success.

More users means more hosts. More hosts means more metrics. More features means more services, which means more APM spans. More traffic means more log lines. Your observability bill grows superlinearly with your infrastructure.

The teams hitting $90k per year are not doing anything wrong. They are using the product as designed. The pricing model just assumes every team needs every feature at every scale — and charges accordingly.

When Datadog is genuinely the right choice

This is not a "Datadog is bad" argument. Datadog is an excellent product. For the right use case, nothing else comes close.

If you run a large-scale microservices architecture where a single user request touches 15 services, you need distributed tracing. If you have custom business metrics that drive real-time capacity decisions, you need a metrics platform with flexible querying. If you have an SRE team whose full-time job is maintaining service reliability across hundreds of hosts, Datadog's unified platform saves them real time.

Enterprise teams with the budget, the complexity, and the dedicated staff to leverage full observability should keep using it. The ROI is real at that scale.

The disconnect is not that Datadog is overpriced for what it offers. It is that most teams are buying capabilities they do not use. A tool can be both excellent and wrong for your situation.

What the 84% are switching to

The teams reducing their observability spend are not going without monitoring. They are unbundling. Instead of one platform that does everything (at a price that reflects everything), they are assembling focused tools that cover what they actually need:

The result is usually a 70-90% reduction in monitoring costs with zero meaningful loss in incident detection. The alerts that matter — "your site is down," "your SSL cert expires in 7 days," "your nightly backup did not run" — all come from simple, focused tools.

Where Uptrack fits

Uptrack is built for the teams in the first column — the ones who need reliable, fast uptime monitoring without the overhead of a full observability platform. It covers the checks that actually generate actionable alerts for most teams:

• HTTP endpoint monitoring with keyword matching — know your site is up and returning the right content
• 30-second check intervals on the free tier — the fastest free checks available
• SSL certificate expiry monitoring — no more emergency renewals at 2 AM
• Cron job monitoring — verify your scheduled tasks actually ran
• Public and private status pages — communicate reliability to your customers
• Multi-region checks — catch regional outages, not just global ones
• Alerts via Slack, Discord, email, and webhooks — where your team already works

The free tier includes 50 monitors — 10 at 30-second intervals and 40 at 1-minute intervals. For a small-to-mid team running a handful of services, that covers everything without spending a dollar.

If you later outgrow simple monitoring and genuinely need distributed tracing or custom metrics at scale, you should use a tool built for that. Uptrack is not trying to be Datadog. It is trying to be the monitoring layer that 84% of teams actually need.