Frontend observability tools: platforms and open-source options compared

Frontend observability tools collect telemetry from real users' browsers: Core Web Vitals, JavaScript errors, resource timing, and traces of user interactions. The market splits into five categories, and which category fits depends on what you need to observe, whether self-hosting is a requirement, and what your organisation already runs. This page lists the tools in each category and the facts that separate them. For what frontend observability is and how to implement it, see the frontend observability guide.

The five tool categories

• Full-stack platforms: Honeycomb, Datadog, Dynatrace, New Relic. Frontend telemetry is one part of a platform that also covers backend services and infrastructure, which enables end-to-end tracing in one product.
• Error-tracking-first tools: Sentry. Built around errors and stack traces, extended with performance monitoring and session replay.
• RUM specialists: Embrace, SpeedCurve, RUMvision, DebugBear. Focused on real user performance data and Core Web Vitals rather than general observability.
• Open-source options: Grafana Faro, SigNoz, OpenReplay, Elastic, Sentry self-hosted, and OpenTelemetry with a backend of your choice. Required when telemetry must stay on your own infrastructure.
• Platform-bundled: Cloudflare Web Analytics, Vercel Speed Insights. Field data you get by virtue of where the site is hosted, with no separate contract.

Comparison table

Full-stack platforms

Honeycomb

Honeycomb ingests OpenTelemetry data natively and is built for querying high-cardinality data: you can group and filter by any attribute (user ID, device, release, country) without pre-aggregation. Its BubbleUp feature compares anomalous events against the baseline to surface which attributes differ. Frontend and backend spans join into single traces when trace context is propagated from the browser.

Datadog RUM

Datadog RUM is the browser component of the Datadog platform, alongside infrastructure monitoring, APM, and log management. It records session replays, tracks user journeys across pages, and links browser sessions to backend traces within the same platform. Its relevance rises when Datadog is already in use for backend monitoring, because frontend and backend then share one interface.

Dynatrace

Dynatrace instruments real user monitoring automatically, without manual SDK configuration per page. Its Davis AI performs automated root cause analysis across the full stack, and session replay is correlated with performance data. It traces from browser to backend within its own agent ecosystem. On 13 August 2026 Dynatrace announced the acquisition of Arize, an AI observability company, for $915 million in cash and stock, extending the platform towards monitoring AI models and agents.

New Relic

New Relic covers browser monitoring, APM, and infrastructure in one platform and offers a 100GB/month free data tier. Browser monitoring includes session traces and Ajax call analysis, and the platform provides a Vercel integration for Next.js deployments.

Error-tracking-first tools

Sentry

Sentry started as an error tracker and remains organised around the error workflow: source-mapped stack traces, release tracking, and issue assignment. It has added performance monitoring with Web Vitals, transaction tracing, and session replay that can record only when an error occurs. Its SDKs are open source and a self-hosted edition exists, which makes it one of the few options in both the commercial and self-hosted columns.

RUM specialists

These tools do not aim to be observability platforms. They specialise in real user performance data, which for teams whose primary question is "how fast is the site for real users, and why" is the entire requirement.

Embrace

Embrace is built on OpenTelemetry and started as a mobile observability platform, with SDKs for iOS, Android, React Native, Unity and Flutter. It added browser RUM in June 2025, covering Core Web Vitals, JavaScript exceptions and session timelines. Its web SDK is Apache-2.0 and exports to any OTLP-compatible backend, so the instrumentation is not locked to Embrace's own platform. The backend itself is SaaS only.

Embrace is the one vendor here that treats web and mobile as one product rather than two, which matters if the same team owns an app and a storefront.

• SpeedCurve combines RUM with synthetic testing, so field data and lab data sit in one product. Embrace acquired SpeedCurve in November 2025 and states the product continues to be supported. This site runs SpeedCurve's LUX RUM script.
• RUMvision specialises in Core Web Vitals monitoring and attribution, including INP debugging via the Long Animation Frames API.
• DebugBear monitors Core Web Vitals through both RUM and scheduled synthetic tests.

Open-source options

Grafana Faro

Grafana Faro is the Grafana team's frontend observability SDK, now on the 2.x line. It collects Web Vitals, errors, and session metadata, and feeds the Grafana ecosystem: Tempo for traces, Loki for logs, Prometheus for metrics. It runs self-hosted or via Grafana Cloud.

SigNoz

SigNoz is built natively on OpenTelemetry and combines logs, metrics, and traces in a single product. Worth being precise about its role: SigNoz describes itself as an observability platform covering APM, distributed tracing, log management and infrastructure monitoring, and does not ship a browser RUM SDK of its own. It belongs here as a destination you send browser telemetry to, paired with OpenTelemetry or Faro instrumentation, rather than as the thing doing the measuring.

OpenTelemetry with your own backend

OpenTelemetry is a CNCF standard, not a product: it defines how telemetry is collected and exported, and any compatible backend can receive it (Jaeger, Tempo, Honeycomb, Grafana, or a custom store). The JS SDK 2.0 was released in March 2025, and the OpenTelemetry Browser SIG is building browser instrumentation with framework-specific support for React, Next.js, Vue, Svelte, and Angular. This route gives complete data ownership at the cost of operating the storage and analysis stack yourself.

Tools we have used in client work

The following tool usage comes from Iron/Out client engagements documented in our case studies, not from vendor materials:

• Dynatrace, Honeycomb, SpeedCurve, and ContentSquare in the Frasers Group engagement: INP optimisation across SportsDirect, Flannels, and House of Fraser, covering over 3 million sessions during a three-month rollout.
• RUMvision and the Long Animation Frames API in the Simyo engagement: mobile INP reduced from 413ms to 150ms, measured at the 75th percentile on origin level.

The market is consolidating

Three acquisitions in eighteen months have changed who owns what on this page, which is worth knowing before you sign a multi-year contract:

• LaunchDarkly acquired Highlight.io in April 2025 and shut the standalone service down on 28 February 2026. The repository remains Apache-2.0 for anyone self-hosting, but there is no supported hosted path. We have deliberately left it off the list above.
• Embrace acquired SpeedCurve in November 2025, putting a long-standing RUM specialist inside an OpenTelemetry-native platform.
• Dynatrace announced the acquisition of Arize in August 2026 for $915 million, extending a full-stack platform into monitoring AI models.

The pattern is that independent specialists are being absorbed into broader platforms. That is an argument for keeping your instrumentation portable: an OpenTelemetry-based setup can be pointed at a different backend without re-instrumenting your site, which a vendor SDK cannot.

Selection criteria

These are the questions that narrow the field, in the order they eliminate options:

1. Must telemetry stay on your infrastructure? Self-hostable options include Sentry, Grafana Faro, SigNoz, OpenReplay for session replay, the Elastic stack, and OpenTelemetry with a backend you run yourself.
2. Do you need traces that continue into the backend? If yes, the tool must support trace context propagation and your backend must be instrumented in a compatible format. The full-stack platforms and OpenTelemetry do this; the RUM specialists measure the browser side only.
3. What is the primary question you need answered? Error tracking points to Sentry; Core Web Vitals and page speed point to the RUM specialists; cross-layer debugging points to the full-stack platforms.
4. What does your organisation already run? Adding RUM to an existing Datadog, Dynatrace, New Relic, or Grafana installation reuses existing dashboards, alerting, and contracts.
5. What will the data volume cost? Browser telemetry scales with page views. Pricing models differ per vendor; the free tiers stated above are the ones vendors publish, and sampling reduces volume on high-traffic sites.

We select and implement observability tooling for clients as part of our observability service, using OpenTelemetry where vendor independence is required. Contact us if you want help making this selection for your stack.

Related resources

• Frontend Observability Guide: concepts, implementation, and best practices
• End-to-End Observability: linking browser telemetry to backend traces
• Observability Service: OpenTelemetry implementation by Iron/Out
• Core Web Vitals Guide: the metrics these tools measure

Frequently asked questions