Paige Andrews
Anthony Rindone
Releasing new features is one of the highest-stakes moments in the software delivery life cycle. Even with CI/CD pipelines in place, plenty of things can still go wrong when a feature goes live for actual users. Most feature flagging tools operate in isolation from important observability tooling, forcing engineers to monitor changes across multiple disconnected systems to fully understand their impact. This slows down development and increases the chance of missing critical issues.
Datadog Feature Flags enables faster and safer releases by integrating flagging directly into the observability platform teams are using. Engineers can create feature flags and manage rollouts, correlate health metrics to rollouts, automate canary releases and rollbacks, and run advanced experiments, all in a single workflow.
In this post, we’ll explore how Datadog Feature Flags helps teams:
- Manage feature delivery directly in Datadog
- Monitor rollouts with built-in observability
- Automate releases and enforce safe deployment practices
- Run advanced experiments with modern statistical methods
Manage feature delivery directly in Datadog
The pressure to respond quickly when a new feature introduces problems can lead to chaos, especially if the system that catches the problem is separate from the system that can toggle the feature off.
Datadog Feature Flags allows teams to create and manage features within the same platform where they monitor and manage application health. Built on the OpenFeature SDK, Feature Flags enables teams to quickly create feature flags and variants using any data type—Boolean, string, number, JSON—for their configured environments, providing a vendor-neutral solution to feature flag management.
Within a newly created feature flag, teams can choose to add a flag to a particular codebase or repository, define targeting rules for rollouts in specific environments, add variants, and view real-time metrics. Read our documentation for a deeper look at configuring your SDKs and creating your first flag.
Troubleshoot with telemetry-linked flags
Traditional feature flag tools often operate independently from observability platforms, requiring teams to switch between dashboards to correlate application health or user behavior with a new rollout. This disconnect slows down triage and leaves teams guessing whether a new feature caused a problem.
Datadog Feature Flags closes this gap by tying each flag directly to your existing telemetry data: metrics, logs, traces, and business KPIs. There's no need to set up integrations or write custom scripts, as every rollout is automatically tracked against the health indicators you already monitor.
For example, when rolling out a new checkout experience, you can see in real time how it affects response times, error rates, and conversion metrics. This helps teams catch regressions early and validate success without jumping between tools. Teams can also pivot to RUM or Product Analytics events to dig deeper into errors or changes to business KPIs.
When a feature rollout triggers errors or performance issues, engineers can immediately locate and toggle the relevant flag from within Datadog, eliminating the need for new deployments or tool-switching. Every change is automatically logged for complete traceability and auditability.
For example, during a live incident, the on-call engineer can search for active feature flags, correlate spikes in errors directly to a new rollout, and disable the flag with a single click to contain impact. Feature access can be targeted by environment, service, or user group to mitigate risk. This integrated approach, following modern best practices modeled by tools like Eppo, empowers any authorized team member—not just developers—to keep releases safe and controlled without bottlenecks.
Automate releases and rollbacks
Manual rollouts put pressure on engineers to constantly watch dashboards and take action if something goes wrong. This isn’t scalable, especially as teams adopt AI-assisted development and ship changes more frequently.
Datadog Feature Flags lets teams define automated rollout strategies tied to application health. You can configure a flag to gradually release to users while Datadog continuously monitors key metrics; if performance degrades or errors spike, Datadog automatically halts or rolls back the release and notifies you to take action.
This automation enforces best practices like canary deployments and progressive rollouts without requiring engineers to supervise every change. And because these rules are defined in the same platform used to monitor application health, standardized best practices also become built into the release process.
As an example, imagine you lead a development team at a large online retailer with millions of online transactions. Your team has developed a new AI-powered feature that calls a machine learning service in real time to dynamically provide users recommendations on shipping options to increase conversion rates. You can't do a full immediate rollout (“big bang”) due to the complexity and risk of impacting checkout revenue, and you can't do a blue-green deployment because switching 100% of traffic would potentially still risk slowing down the experience for all users, even if your team could immediately switch off if needed. Therefore, the best option is a canary release because it's the only strategy that allows the team to safely observe the new, complex system's behavior with real user traffic before exposing it to the entire user base.
Scale experimentation safely across environments
Experimentation is essential for innovation, but it often introduces hidden risks. Every new variant, flag, or code path can impact reliability, increase complexity, and slow delivery. With Datadog Feature Flags, experiments are built directly on top of real-time observability data, so platform teams instantly see the impact of A/B tests and progressive rollouts on system health, latency, and user experience.
Datadog automates guardrails for every experiment: Rollbacks trigger instantly when SLOs are threatened, business metrics validate impact, and stale flags are removed before they increase tech debt. Whether orchestrating a canary release or segmenting users for targeted tests, experimentation workflows use the same safety net as production releases, with no manual oversight, blind spots, or fragmented dashboards.
This approach enables platform teams to deliver experimentation infrastructure that gives engineers freedom to try new things, all while maintaining the reliability and organizational standards the business demands.
Release quickly and confidently with Datadog Feature Flags
Feature flags have evolved from simple on/off switches into foundational tools for modern software delivery. With Datadog Feature Flags, they become part of a unified system that combines observability, automation, and experimentation.
As AI accelerates the pace of development, it's more important than ever to release features with built-in safety checks and clear measures of success. Datadog brings these capabilities together in one platform, helping teams ship faster, reduce risk, and learn continuously.
Check out how to get started with creating your first feature flag or sign up for a 14-day free trial.
