Elastic Observability APM

Provenance unknown Report an error

Panel rating · 6 judges · How to read the stars

Category median

Sovereignty: 3 of 6 dimensions proven

0–5 in half steps. 5 means the rubric's top anchor is met on the evidence.

by Elastic (Elasticsearch B.V.) · www.elastic.co

Report an error on this page Is this your product? →

Link to this page

Elastic Observability APM: AI-assisted analysis on whats-best.ai Elastic Observability APM: AI-assisted analysis on whats-best.ai (dark)

Add ?theme=dark to the image address for dark backgrounds. The badge shows the category and the month of the latest evaluation, and EU-Made where that is the product’s provenance. It never shows a score.

Elastic Observability APM at a glance

Vendor
Elastic (Elasticsearch B.V.)
Headquarters
Not stated on the vendor’s pages as of 17 Nov 2026
Data hosting
Not stated on the vendor’s pages as of 17 Nov 2026
Free plan
Not stated on the vendor’s pages as of 17 Nov 2026
Panel verdict
The AI panel rates Elastic Observability APM 2/5, the mean of 6 judges. Elastic Observability APM is an application performance monitoring system built on the Elastic Stack. how scores are made
As of
Vendor pages captured 17 Nov 2026; evaluated by the panel 18 Nov 2026. all sources

Elastic Observability APM alternatives, European ones first →

Read this page as one judge. Each weighs the same scores by what they care about.

The panel's verdict

Elastic Observability APM is an application performance monitoring system built on the Elastic Stack. The bench scores it strongest on performance tracing, 4-6, and sovereignty, 5-8. Higher tracing scores rest on W3C Trace Context propagation, a trace waterfall linking spans to their errors and a managed OpenTelemetry intake endpoint; lower ones stop at undescribed percentile breakdowns, profiling and anomaly detection. Workflow and integrations is the weakest criterion, scored 2 with no spread: the documented integrations pull data in, and no public information appears on alert rules, notification channels or issue-tracker sync, so nothing shown routes an error to whoever fixes it. Error tracking clusters at 4-5, ending at grouping, with issue states, release tracking and source maps undescribed. Sovereignty is the genuine split: the reliability engineer scored 8 for the self-managed chain keeping events on customer-controlled servers; the SaaS CTO scored 5, finding the managed chain only partly verifiable — no EU region named, no data processing agreement found, no subprocessor locations shown. A 14-day free trial with all features and no credit card required is public; no plan prices appear.

Report an error

Speaks for it

  • Distributed tracing over W3C Trace Context works out of the box, with a trace waterfall that links a span to its errors and a managed OpenTelemetry intake endpoint.
  • Agents are documented for Go, iOS, Java,.NET, Node.js, PHP, Python, Ruby and real-user monitoring, capturing stack traces, a culprit field, service, host, process, URL and user context, automatic grouping and KQL search.
  • The whole stack runs self-managed as packages, Docker images and Kubernetes Helm charts on Linux, macOS, Windows, Docker or Kubernetes, with the default distribution including the full set of free features under the Elastic License.
  • SAML 2.0 single sign-on is documented in depth — signed requests, encrypted assertions, role mapping from identity-provider groups, automatic provisioning — and is included at no additional cost on Elastic Cloud Serverless on any tier.
  • Cloud deployments allow hosting-region selection on AWS, Azure or Google Cloud Platform with backups stored in the selected region by default, alongside a published transfer impact assessment and a stated no-backdoor commitment.

Report an error

Held against it

  • No public information appears on alert rules, notifications to email, Slack, PagerDuty or Teams, or issue-tracker sync, and the documented integrations pull audit and security data in rather than route errors to anyone.
  • Triage stops at grouped errors: no public information appears on issue states such as resolved or regressed, release and deploy tracking, or source maps and symbolication.
  • User email, ID and username are captured when supplied, field limits of 1024 and 10,000 characters truncate but do not anonymise, and no public information appears on scrubbing, IP anonymisation or deletion on request.
  • Audit logging is disabled by default, limited to certain subscription levels and stated to be unavailable on Serverless, with no public information on two-factor authentication, ISO 27001 or SOC 2, a public status page or an SLA.
  • The captured pages name no EU region, show no subprocessor locations, and give no public information on a data processing agreement or the jurisdiction of the contracting entity, elasticsearch B.V.

Report an error

Best for

  • You need every trace and error in your own chain — the stack runs self-managed on your Linux, Windows, Docker or Kubernetes infrastructure, with APM events indexed into your own Elasticsearch data streams.
  • You standardise on OpenTelemetry and want a managed intake endpoint that keeps instrumentation portable.
  • You already run Elasticsearch and Kibana and want errors, traces and logs searched together through KQL and Discover.
  • You must select your hosting region for data residency among AWS, Azure or Google Cloud Platform, with backups kept in that region by default.

Report an error

Avoid if

  • You need an error to page someone — ask the vendor: the public pages we read do not show it
  • You triage frequent releases — the errors pages stop at frequency trends and top affected transactions, while release and deploy tracking, source maps, symbolication and issue states are undescribed on the captured pages.
  • Error payloads will carry personal data — user email, ID and username are captured when supplied, and field limits of 1024 and 10,000 characters truncate rather than anonymise.
  • You need audit logging on Serverless — the captured pages state it is unavailable there, and it is disabled by default and gated to certain subscription levels elsewhere.

Report an error

The scores

Panel median The median of the judges’ scores per criterion. Each label opens its reasoning below.

Error capture, grouping & triage

Show reasoning
How this is scored

What happens to an exception or crash: SDKs for the languages, frameworks and platforms a team runs (PHP and Laravel, JavaScript in the browser and Node, Python, Ruby, Java, .NET, Go, iOS, Android, Flutter, React Native), stack traces with source maps and symbolication, context (request, user, breadcrumbs, environment, local variables), grouping and custom fingerprints, deduplication, issue states (resolved, ignored, regressed), releases and deploy tracking, suspect commits, crash-free sessions and users for mobile apps, and search and filters over issues.

0 — Catches no application errors or crashes; a log viewer or an external check only.

3 — Errors from one or two platforms arrive with a stack trace and are grouped; context, issue states, releases and source maps are basic or not described.

5 — SDKs for the common web and backend platforms, stack traces with source maps, request and user context and breadcrumbs, automatic grouping with issue states and regression detection, release and deploy tracking, and issue search and filters.

8 — SDKs across backend, frontend and mobile with symbolication documented, custom fingerprinting and merge or split of issues, suspect commits or code owners, crash-free users and sessions, local variables or source context per frame, and every SDK documented with its configuration options.

10 — Error tracking a large engineering organisation standardises on: every platform it runs covered with parity stated, grouping rules editable and explained, release health with adoption and regression per version, and the full issue lifecycle reachable by API.

Report an error

The Agency Tech Lead

Agents for PHP, Node.js, Ruby, Java,.NET, Go, iOS and the browser are documented with W3C trace context, stack traces with a culprit field, service/host/user/url context, custom labels, automatic grouping of similar errors with frequency trends, a search bar over the errors page, and the top affected transactions per group. We found no public information on issue states like resolved or regressed, release tracking, source maps or suspect commits, so triaging forty client sites would still mean reading every group by hand. 6 7 9

Report an error

The SaaS CTO

Errors arrive as documents with stack traces, a culprit field, and service, host, process, URL and user context, grouped by a grouping key with an errors page showing frequency trends and the top five affected transactions, searchable with KQL; agents are documented for Go, Java,.NET, Node.js, PHP, Python, Ruby, iOS and the browser. We found no public information on issue states like resolved or regressed, release and deploy tracking, source maps and symbolication, or suspect commits on the captured pages, so I cannot yet see how a release gets tied to the break it caused. 6 7 9

Report an error

The Mobile Team Lead

The APM error pages show exceptions arriving with stack traces, a culprit field, grouping keys and rich service, host, process, URL and user context, with agents documented for Java, Go,.NET, Node.js, PHP, Python, Ruby, RUM and iOS plus a mobile Errors and crashes view and KQL search over issues. For my biweekly app releases, though, we found no public information on native symbolication or source maps, crash-free users or sessions per release, release and deploy tracking, or issue states like resolved and regressed. 6 7 8 9

Report an error

The Site Reliability Engineer

Agents for Go, Java,.NET, Node.js, PHP, Python, Ruby and the browser capture exceptions with stack traces, a culprit and full context (service, host, process, URL, user), similar errors are grouped by a grouping key, and the Errors page shows frequency, trends and the top five affected transactions with KQL search. The captured pages do not describe issue states such as resolved or regressed, release and deploy tracking, source maps, symbolication or custom fingerprints, so triage stops at grouping. 6 7 9

Report an error

The Data Protection Officer

Agents are documented for Go, Java,.NET, Node.js, PHP, Python, Ruby and real-user monitoring, with stack traces, grouping keys and grouped error views showing frequency, trends, the top five affected transactions, and error samples carrying culprit, trace and context that can be searched with KQL. We found no public information on issue states such as resolved or regressed, release and deploy tracking, or source maps and symbolication, so the lifecycle side of triage stays undocumented. 6 7 9

Report an error

The Skeptic

Errors arrive from a documented agent family covering Go, iOS, Java,.NET, Node.js, PHP, Python, Ruby and RUM, with stack traces, a culprit field, service/host/process/URL/user context, automatic grouping plus a KQL search bar — comfortably past a one-or-two-platform tool. But we found no public information on issue states like resolved, ignored or regressed, release and deploy tracking, source maps and symbolication, or crash-free sessions, so the triage lifecycle beyond the grouped view is undescribed. 6 7 9

Report an error

APM, tracing & profiling

Show reasoning
How this is scored

How the product shows where time goes: transaction and endpoint timings with percentiles, slow database queries and N+1 detection, background jobs and queues, external HTTP calls, distributed tracing across services, OpenTelemetry ingestion, code-level profiling, frontend performance and Core Web Vitals (real user monitoring), host or container metrics beside the traces, anomaly detection on throughput and latency, and the link from a slow trace to the error it caused.

0 — No performance data of any kind; errors only.

3 — Response times per endpoint or a simple performance view; tracing across services, query analysis and profiling are basic or not described.

5 — Transaction timings with percentiles per endpoint, slow query and N+1 detection, background job monitoring, distributed tracing across services, a link between traces and errors, and frontend performance or Web Vitals offered.

8 — OpenTelemetry ingestion documented, code-level or continuous profiling, real user monitoring with Web Vitals, host and container metrics beside the traces, anomaly detection on latency and throughput, and sampling of traces configurable with its effect explained.

10 — Full application performance management evidenced: traces, profiles, frontend and backend joined end to end, service maps, SLOs on endpoints, and every signal documented with its retention and sampling per plan.

Report an error

The Agency Tech Lead

Transactions carry timing, database statements on spans, cache and external HTTP monitoring, distributed tracing works out of the box via the traceparent header, there is a timeline waterfall with a link from a span to its errors, and an OpenTelemetry intake endpoint is documented for serverless. We found no public information on percentiles per endpoint, N+1 detection, background job timing, profiling or anomaly detection, so I can find a slow page but the captured pages do not show me proving why before the client calls. 7 8 9 10

Report an error

The SaaS CTO

Distributed tracing is the strong part: W3C Trace Context across all agents with out-of-the-box propagation, a trace timeline waterfall with a link from a span to its errors, transaction grouping by type and name, agent-controlled sampling, a browser RUM agent, and a managed OTLP endpoint for OpenTelemetry ingestion — that OpenTelemetry path is exactly what I want so instrumentation stays portable. We found no public information on percentiles per endpoint, slow-query or N+1 detection, background job monitoring, code-level profiling or Web Vitals on the captured pages. 7 8 9 10

Report an error

The Mobile Team Lead

Distributed tracing is documented as working out of the box over W3C Trace Context with a timeline waterfall that links straight to the error in the trace, and the product monitors response times, database queries, caches and external HTTP calls, with a RUM agent, configurable transaction sampling and a managed OTLP endpoint. We found no public information on percentile breakdowns per endpoint, N+1 or slow-query analysis, background jobs, or code-level profiling on the captured pages. 8 9 10 12

Report an error

The Site Reliability Engineer

Transactions and spans carry database statements, cache and external HTTP timings, distributed tracing works out of the box over W3C trace context with a waterfall view, errors link to their transaction by ID, an OpenTelemetry intake endpoint is documented, and agents expose sampling settings. We found no public information on percentiles per endpoint, code-level profiling, anomaly detection or Core Web Vitals in the captured pages. 8 9 10 6

Report an error

The Data Protection Officer

Transaction timings for requests, database queries, caches and external HTTP calls, distributed tracing over W3C trace context with a trace waterfall whose error view links a span to its errors, RUM transaction marks and a documented OpenTelemetry intake endpoint give a solid tracing core. We found no public information on percentile breakdowns per endpoint, slow-query or N+1 detection, code-level profiling or anomaly detection, so it stops short of full application performance management on these pages. 7 8 9 10

Report an error

The Skeptic

Transactions with agent-controlled sampling, distributed tracing over W3C Trace Context with a waterfall view, a documented link from a trace span to its errors, RUM agents and a managed OpenTelemetry endpoint put this well past a simple response-time view. We found no public information on percentiles per endpoint, slow-query or N+1 detection, background job monitoring, code-level profiling, or anomaly detection on latency and throughput. 7 8 9 10

Report an error

Alerting, integrations & developer workflow

Show reasoning
How this is scored

How an error reaches the person who fixes it: alert rules on new issues, regressions, frequency and performance thresholds, notifications to email, Slack, Microsoft Teams, PagerDuty or Opsgenie, issue tracker integrations (Jira, GitHub, GitLab, Linear) that create and sync issues, source code links to the repository, deploy and release hooks from CI, a documented REST API and webhooks, a CLI, AI features that explain an error or suggest a fix (and which model runs them), and digest or weekly reports.

0 — No notifications and no integrations; errors are only visible by logging in.

3 — Email or chat notifications for new errors and one or two integrations; alert rules, the issue tracker link and the API are basic or not described.

5 — Configurable alert rules for new, regressed and frequent errors, Slack and email and an on-call tool, issue tracker integrations with GitHub, GitLab or Jira, deploy tracking from CI, a documented REST API and webhooks.

8 — Two-way sync with issue trackers, source code and commit links, alerts on performance thresholds and anomalies with routing per team or project, a CLI for releases and source maps, AI explanations or fix suggestions with the model provider named, and scheduled reports.

10 — A workflow a large team runs on: ownership rules routing issues to teams, every setting reachable by API or infrastructure as code, alert noise controls documented, and every AI feature documented with its data handling and an opt-out.

Report an error

The Agency Tech Lead

The captured pages document a GitHub integration for audit logs, code and secret scanning, central APM server management with Fleet, Elasticsearch language clients, and support over forum and Slack — but we found no public information on alert rules, notifications to email, Slack or an on-call tool, issue tracker sync, webhooks or a releases CLI. For an agency this is the deciding gap: nothing on these pages shows how a new error reaches the developer who has to fix it. 5 10 11 12

Report an error

The SaaS CTO

The captured pages document an inbound GitHub integration that pulls audit logs, code scanning and secret scanning alerts, API access through the Elasticsearch language clients and curl, and central management of the APM integration with Fleet. We found no public information on alert rules, notification channels such as Slack, Teams or PagerDuty, issue tracker sync, deploy hooks from CI, webhooks, a CLI or AI assistance on these pages — as a buyer I cannot see how an error reaches the engineer who owns the service. 5 10 11

Report an error

The Mobile Team Lead

We found no public information on alert rules or notifications to email, Slack, PagerDuty or Teams, which is exactly how I would stop a bad store rollout, and no issue-tracker sync for Jira or Linear is described. What is documented is a GitHub integration for audit logs and code scanning, central management of APM Server through Fleet, and the Elasticsearch REST API via language clients and curl. 5 10 11

Report an error

The Site Reliability Engineer

Data ingestion is well covered — GitHub audit and security data, AWS S3, Azure, Google Cloud Storage, an OTLP endpoint and language clients — but we found no public information on alert rules, notifications to Slack, PagerDuty or email, or an issue tracker link that creates and syncs issues. For an on-call rotation this is the gap that matters: nothing documented here routes an error to the person who fixes it. 10 11 5

Report an error

The Data Protection Officer

What the captured pages show is data ingestion — GitHub code scanning, secret scanning and audit logs, Elasticsearch APIs and language clients, central agent management with Fleet, and community forum or Slack support — rather than the path from an error to the person who fixes it. We found no public information on alert rules for new or regressed errors, notification destinations such as email or on-call tools, issue tracker sync, deploy hooks from CI, or AI features that explain an error and which model runs them. 5 10 11

Report an error

The Skeptic

We found no public information on alert rules, notification channels for errors, webhooks or a CLI; the Slack and forum mentions are support channels, and the GitHub integration pulls security and audit data in rather than creating issues from errors. The documented APIs are the Elasticsearch data APIs, so how an error reaches the person who fixes it is left undescribed on these pages. 5 10 11

Report an error

Data scrubbing, retention & volume control

Show reasoning
How this is scored

What the product does with the personal and secret data an error carries and how volume is kept in check: scrubbing of passwords, tokens, card numbers and personal data in the SDK and on the server with configurable rules, IP address anonymisation, opting out of request bodies and user data, data retention per plan and deletion on request, sampling of errors and traces, rate limits, spike protection and quota caps per project, filters that drop events before they count, data export, and a self-hosted edition that keeps every event on the customer's own servers.

0 — No control over what is captured, no retention period stated, no limit on volume.

3 — Some default filtering of passwords and a retention period stated for one plan; server-side scrubbing, sampling and spike protection are basic or not described.

5 — Configurable data scrubbing in the SDK and default server-side scrubbing of common secrets, IP anonymisation, retention stated per plan, sampling and rate limits per project, a quota cap or spike protection, and event filters before ingestion.

8 — Custom server-side scrubbing rules with examples, per-project retention settings and deletion of individual events or users, spike protection documented with how overage is billed, data export by API, and a self-hosted edition or a documented option to keep data in the customer's chosen region.

10 — Data minimisation by design and evidenced: personal data off by default in every SDK, scrubbing rules auditable, retention and deletion controllable per data type, volume controls that never surprise the invoice, and a supported self-hosted edition with feature parity stated.

Report an error

The Agency Tech Lead

Sampling is agent-controlled with settings, managed inputs answer with HTTP 429 under load with documented retry-and-queue behaviour for the shippers, retention can be customised per data stream through lifecycle policies, and the whole stack runs self-hosted on our own servers. We found no public information on scrubbing of passwords, tokens or personal data, IP anonymisation, or retention stated per plan — with shop customer addresses inside request bodies, that silence is the first thing I would have to close. 1 3 8 11 12

Report an error

The SaaS CTO

Volume behaviour is genuinely documented: managed inputs respond with HTTP 429 when capacity is exceeded, shippers retry with backoff and queue data, agents control transaction sampling, retention is adjustable through index lifecycle policies, and the whole stack runs self-managed on our own infrastructure under a free Basic licence with the full set of free features. We found no public information on scrubbing of passwords or personal data, IP anonymisation, event filters before ingestion, or per-plan retention and quota or overage terms on the captured pages. 1 4 8 11 12

Report an error

The Mobile Team Lead

Sampling is agent-controlled with settings to tune it, retention is customizable through ILM and data stream lifecycle policies, and the managed inputs push back with HTTP 429 under load while shippers retry with backoff and queueing, plus documented field length limits. We found no public information on scrubbing of passwords or personal data, IP anonymisation, or filters that drop events before they count. 8 11 12

Report an error

The Site Reliability Engineer

Volume control is documented in real depth — HTTP 429 responses with retry and backoff guidance for Beats, Elastic Agent and Logstash, cluster-capacity limits on hosted, automatic scaling on serverless — alongside agent-side sampling, field character limits and retention customisable through index lifecycle policies. We found no public information on scrubbing of secrets or personal data, IP anonymisation, event filters before ingestion or per-event deletion, though the self-managed edition keeps data on the customer's own servers. 12 8 11 1

Report an error

The Data Protection Officer

Volume handling is documented: agents control sampling, managed inputs answer with HTTP 429 when capacity is exceeded while shippers retry with backoff, retention is adjustable through index lifecycle policies, and a documented self-managed edition plus a selectable hosting region keep events in the customer's own chain. For the personal data an error actually carries, we found no public information on scrubbing of secrets or personal data in the agent or on the server, IP anonymisation, or deletion on request; the documented field limits of 1024 and 10,000 characters truncate but do not anonymise, and user email, ID and username are captured when supplied. 1 4 6 8 11 12

Report an error

The Skeptic

Volume behaviour is documented honestly — agents control transaction sampling, managed inputs shed load with HTTP 429 until volume drops, and shippers retry with backoff and queueing — and a self-hosted edition on the customer's own infrastructure is documented with the full set of free features. But we found no public information on scrubbing of passwords, tokens or card numbers, IP anonymisation, or a stated retention period for APM data; the only retention detail shown is ILM customisation on a GitHub audit data stream. 1 4 8 11 12

Report an error

Access control, security & the vendor's reliability

Show reasoning
How this is scored

Whether the service holds up in an audit and on the vendor's own bad day: organisations, teams and projects with roles, single sign-on (SAML or OIDC) and SCIM, two-factor authentication, audit logs, encryption in transit and at rest, security certifications and reports (ISO 27001, SOC 2), a security page and a vulnerability disclosure policy, the vendor's own public status page and incident history, a stated availability or SLA, and how ingestion behaves when the vendor is down (SDK buffering, no impact on the app).

0 — Nothing on access control, security or the vendor's own availability.

3 — User accounts with two-factor authentication and a security page; roles, SSO, certifications and a status page are basic or not described.

5 — Organisations and projects with roles, two-factor authentication, single sign-on offered on some plan, encryption in transit and at rest stated, a public status page, and a security page naming certifications or audits.

8 — SAML SSO and SCIM provisioning, audit logs, ISO 27001 or SOC 2 with its scope stated, an SLA with credits published, the incident history public, a vulnerability disclosure or bug bounty programme, and SDK behaviour during vendor outages documented.

10 — Enterprise-grade operation evidenced: granular roles per project and team, audit logs exportable to a SIEM, independent audit reports available on request, published uptime history over years, and ingestion designed and documented never to slow the customer's application.

Report an error

The Agency Tech Lead

Single sign-on is documented in real depth — SAML 2.0 with signed requests and encrypted assertions, role mapping from IdP groups or emails, enforced SSO, automatic provisioning, and lockout recovery through an owner-level API key — and audit logging of authentication failures, refused connections and access decisions is documented across self-managed, hosted and Kubernetes deployments, though gated to certain subscription levels and off by default. We found no public information on two-factor authentication, encryption in transit and at rest, ISO or SOC certifications, a status page or SCIM; the pages even state that removing a user in the IdP does not remove them in Elastic Cloud. 3 13 14

Report an error

The SaaS CTO

Single sign-on is documented in real depth — SAML 2.0 with role mapping from IdP groups, auto-provisioning, enforced SSO with an owner-level API-key fallback, encrypted assertions and clear tier rules (included on Serverless, Enterprise subscription on hosted) — and audit logging of authentication failures and configuration changes is documented down to its output file and dynamic setting, with a trust page publishing privacy datasheets, white papers and a no-backdoor statement. We found no public information on two-factor authentication, encryption in transit and at rest, ISO 27001 or SOC 2, a public status page, or an SLA with credits on the captured pages. 3 13 14

Report an error

The Mobile Team Lead

SAML single sign-on is documented in real depth — role mapping from IdP groups or emails, enforcement, encrypted assertions, tier gating — and Elasticsearch and Kibana audit logging is described down to the event types and per-node output files, alongside security enabled by default and a trust page with white papers and a published transfer impact assessment. We found no public information on two-factor authentication, a public status page, an SLA, or certifications such as ISO 27001 or SOC 2 being named, and audit logs are subscription-gated and unavailable on serverless. 3 4 13 14

Report an error

The Site Reliability Engineer

SAML 2.0 single sign-on is documented thoroughly — organisation and deployment level, role mapping from identity provider groups, enforcement, signed requests, encrypted assertions and session revocation — and audit logging of authentication failures and access decisions exists, though it is disabled by default, gated to certain subscription levels and not available on Serverless. Two-factor authentication, encryption statements, certifications such as ISO 27001 or SOC 2, a public status page and an SLA do not appear in the captured pages, and removing a user at the identity provider does not remove them from Elastic Cloud. 13 14 3

Report an error

The Data Protection Officer

Organisation-level roles, SAML single sign-on with role mapping from identity-provider groups, signed SAML requests, support for encrypted assertions, automatic provisioning and password invalidation on SSO login are documented in detail, and stack audit logging of authentication failures, refused connections and data-access authorisation is documented — disabled by default, dependent on subscription level, and with the captured pages stating it is unavailable on Serverless. We found no public information on two-factor authentication, encryption in transit and at rest, ISO 27001 or SOC 2 with their scope, a public status page with incident history, or an availability commitment. 3 4 13 14

Report an error

The Skeptic

Organisation and deployment roles, SAML SSO with encrypted assertions and request signing, role mapping by IdP group, an owner-lockout fallback API key and detailed audit-logging configuration are genuinely documented — though audit logs are disabled by default, available only on certain subscription levels, and unavailable on Serverless, and removing a user from the IdP does not deprovision them. We found no public information on two-factor authentication, encryption in transit and at rest, ISO 27001 or SOC 2 certifications, a public status page or an SLA with credits. 3 13 14

Report an error

European sovereignty panel opinion

panel disagrees Show reasoning
How this is scored

Who the contracting entity and the parent company are, where error events, stack traces, request data, traces, profiles, session data and account data are stored and processed, and who the subprocessors are (hosting, email and chat delivery, AI models that explain errors, support and payment tools). Independently sourced by the sovereignty pipeline. An EU data region the customer can choose counts for the product; a US default without a stated EU option does not. A self-hosted edition the customer runs on its own European servers counts as the customer's chain, whatever the vendor's seat.

0 — Non-EU vendor and contracting entity, hosting unstated, subprocessors unnamed, and no data processing agreement offered.

3 — A privacy policy and a data processing agreement exist, but hosting location is unstated or non-EU and the subprocessors are not listed.

5 — An EU data region available and an EU contracting entity or EU representative, but parts of the chain (AI features, email, support, payment) run through non-EU providers without a stated safeguard.

8 — EU hosting on named infrastructure, EU contracting entity, subprocessors listed with their locations, a data processing agreement published, and security certifications (ISO 27001, SOC 2) stated — or a self-hosted edition documented that keeps every event on the customer's own servers.

10 — Sovereign end to end and evidenced: European ownership, entity, hosting and every subprocessor including the AI models, or a self-hosted edition with no call home, data export and deletion documented, and certifications published with their scope rather than implied.

Report an error

The Agency Tech Lead

The full stack — Elasticsearch, Kibana, agents, Logstash, Beats — is downloadable to run self-managed on Linux, Docker or Kubernetes on our own European servers, which is the chain I control, and the managed cloud lets me select the deployment region with backups kept in it. But the contracting entity appears only as elasticsearch B.V. with no stated seat, and we found no public information on a stated EU region, a published data processing agreement, or the named subprocessors with their locations — only that such lists exist. 1 2 3 4

Report an error

The SaaS CTO

Customers select the deployment region on AWS, Azure or GCP to meet data residency needs, backups stay in that region, the contracting entity is elasticsearch B.V., public subprocessor lists are stated to exist, and every product is deployable self-managed on our own European servers — that self-hosted path is the chain my enterprise customers accept. The captured pages do not name a specific EU region, publish a data processing agreement, or give subprocessor locations, so the managed chain is only partly verifiable from public evidence. 1 2 3 4

Report an error

The Mobile Team Lead

The whole stack can run self-managed on Linux, macOS, Windows, Docker or Kubernetes, which would keep our app events on servers we control in Hamburg, and the contracting entity is named as elasticsearch B.V. with selectable hosting regions and same-region backups on AWS, Azure and GCP plus a published transfer impact assessment and publicly available sub-processor lists. The captured pages do not name a specific EU region, give sub-processor locations, show a data processing agreement, or state the entity's jurisdiction. 1 3 4 14

Report an error

The Site Reliability Engineer

The entire stack — Elasticsearch, Kibana, APM server and agents — is published as packages, Docker images and Helm charts for Linux, macOS, Windows and Kubernetes under a source-available licence, with APM events documented as stored in the customer's own Elasticsearch data streams, which keeps every trace and error on the customer's own servers. Cloud deployments allow region selection with backups kept in the same region, the transparency statement is published and subprocessor lists exist publicly, though the captured pages state no data processing agreement, no subprocessor locations and no jurisdiction for elasticsearch B.V. 1 4 5 3 10

Report an error

The Data Protection Officer

The whole stack is documented as deployable self-managed on the customer's own Linux, Windows, Docker or Kubernetes infrastructure, which puts every event in the customer's own chain, and the managed cloud lets the customer select the hosting region with backups staying in that region by default on named AWS, Azure and Google infrastructure. The contracting entity is named only as elasticsearch B.V. with no stated seat, no EU data region is named, public subprocessor lists are referenced though the captured pages show no locations, and we found no public information on a published data processing agreement; the statements on no backdoors, disclosure only when strictly required by law, and a published transfer impact assessment are welcome but no substitute. 1 2 3 4

Report an error

The Skeptic

The self-hosted edition — packages, Docker and Helm on the customer's own Linux, macOS, Windows or Kubernetes infrastructure, with security enabled by default — makes the data chain the customer's own, and the managed service documents region selection for data residency with backups staying in the chosen region. We found no public information on the registered jurisdiction of Elasticsearch B.V., a published data processing agreement, or the locations of the infrastructure and support subprocessors; the subprocessor lists are said to be public but are not enumerated on the captured pages. 1 2 3 4

Report an error

Pricing transparency not rated — the vendor publishes no price

Show reasoning
How this is scored

Whether a buyer can compute the real annual cost from public pages alone: price per plan, what is included (errors or events, spans or transactions, replays, profiles, logs, seats, projects, retention), what overage costs per unit and whether it is capped, which features (SSO, audit logs, longer retention, EU region, AI) need a higher plan, host or seat based pricing for suites, the price of the self-hosted edition, annual versus monthly terms, and VAT.

0 — No public prices at all; every customer is a sales call.

3 — A starting price exists, but what volume it covers and what overage, retention and add-ons cost is unclear.

5 — Plans are public with their price and main volume limits, but at least one commonly needed piece (overage per event or span, longer retention, SSO, the EU region or AI features) is unpriced or "contact sales".

8 — Every plan priced publicly with its volume limits and overage rates, a price calculator or table per unit, add-ons and retention priced, the behaviour at the quota (cap or overage) stated, and VAT and billing terms stated.

10 — Complete price computability: annual cost derivable for a given volume of errors, traces, seats and projects, with every add-on, retention step, support level and the self-hosted licence priced and the cancellation terms published.

Report an error

The Agency Tech Lead

The pages document a 14-day free trial with all features and no credit card, a post-trial free Basic licence, and which features sit on which tier — single sign-on needs Enterprise on hosted deployments and is included on serverless, audit logs depend on the subscription level. We found no public information on prices for any paid plan, volume limits, overage rates, VAT or a calculator, so I cannot even begin to compute what one client's cron job throwing a million exceptions would cost me. 1 5 9 13 14

Report an error

The SaaS CTO

The only public numbers are trial terms: a 14-day trial with all features and no credit card, and a one-month local licence that reverts to a free Basic tier, with some gating stated in prose such as SSO requiring an Enterprise subscription on hosted and audit logs depending on subscription level. We found no public information on plan prices, volume limits, overage rates, retention pricing or support levels on the captured pages, so I cannot compute a real annual cost from public pages alone. 1 5 13 14

Report an error

The Mobile Team Lead

A free starting point is public: the self-managed distribution "includes the full set of free features" under the Elastic License, the local trial reverts to "Free and open - Basic", and the cloud trial is 14 days with all features and no credit card. Beyond that we found no public prices for the hosted plans or paid licences — subscription levels such as Platinum and Enterprise gate features like single sign-on, but no figures, volume limits or overage rates are given on the captured pages. 1 5 11 13

Report an error

The Site Reliability Engineer

A starting point is public: a 14-day all-features trial with no credit card, a one-month local trial licence that reverts to 'Free and open - Basic', and feature gating that is stated — single sign-on requires an Enterprise subscription on Elastic Cloud Hosted while it is included at no additional cost on Serverless, and audit logs are limited to certain subscription levels. No plan price, volume limit, overage rate or billing and VAT terms appear anywhere in the captured pages. 1 5 13 14

Report an error

The Data Protection Officer

A free Basic licence after a one-month local trial and 14-day all-features cloud trials with no credit card are public, and subscription tiers are named — single sign-on is included on Serverless at no additional cost while hosted deployments require an Enterprise subscription, and audit logs sit behind certain subscription levels. We found no public information on any monetary price, volume limits, overage rates, retention pricing or VAT terms, so a real annual cost for a given event volume cannot be computed from these pages. 1 5 13 14

Report an error

The Skeptic

The only figures on the captured pages are free: a "14-day free trial" with all features and no credit card, and a self-hosted "Free and open - Basic" licence after the one-month local trial. Plan names Basic, Platinum and Enterprise appear in the SSO documentation — with the Enterprise subscription required for SAML on Elastic Cloud Hosted stated — but we found no public information on what any plan costs, what volume it includes, overage, retention or add-on pricing, so a real annual bill is not computable from these pages. 1 4 5 13

Report an error

European sovereignty — proven facts

3 of 6 dimensions proven

Built only from facts shown on the vendor's own pages. A dimension we could not prove is left open, not scored as zero.

Hosting location Not determined — uncited Report an error
Hosting options Self-hostable 1/1 pts 14 Report an error
Open source Source available unverified
Why unverified?

The Elastic License is a source-available licence rather than OSI-approved open source, and the GitHub repository additionally lists AGPLv3 as a supported licence.

0/1 pts 4 Report an error
Data processing agreement Not determined — uncited Report an error
Transfer impact assessment Published 1/1 pts 3 Report an error

Questions about Elastic Observability APM

Is Elastic Observability APM European?

Not determined: the vendor’s pages as captured do not say where Elastic Observability APM is based.

Where does Elastic Observability APM store data?

Not stated on the vendor’s pages as of 17 Nov 2026.

Does Elastic Observability APM have a free plan?

Not stated on the vendor’s pages as of 17 Nov 2026.

What are the best alternatives to Elastic Observability APM?

As of 18 Nov 2026, the panel rates Sentry (3.5/5), Datadog APM & Error Tracking (3.5/5) and Dynatrace (3/5) highest among the other 19 Error Tracking & APM products evaluated. Ratings that look equal are rounded to half stars; the order uses the unrounded rating, then the name. The highest-rated European alternative is Tideways (EU-Made). Ratings are the AI panel’s opinion, formed from public evidence: how scores are made. The full list, with European products first: Elastic Observability APM alternatives.

Where this could be wrong

What we left out

A claim that does not survive our checks costs us the claim, not the page. This is what was taken off this one.

Sources (14)

The pages every claim on this page was read from — each one checked, dated, and kept verifiable.

  1. 1 Vendor page www.elastic.co Checked 17 Nov 2026 Details →
  2. 2 Legal notice — found from the homepage www.elastic.co Checked 17 Nov 2026 Details →
  3. 3 Privacy policy — found from the homepage www.elastic.co Checked 17 Nov 2026 Details →
  4. 4 Self-hosting / on-premises installation docs — found from the homepage www.elastic.co Checked 17 Nov 2026 Details →
  5. 5 Code repository (open-source licence) — found from the homepage github.com Checked 17 Nov 2026 Details →
  6. 6 Error capture, grouping & triage — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →
  7. 7 Error capture, grouping & triage — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →
  8. 8 APM, tracing & profiling — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →
  9. 9 APM, tracing & profiling — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →
  10. 10 Alerting, integrations & developer workflow — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →
  11. 11 Alerting, integrations & developer workflow — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →
  12. 12 Data scrubbing, retention & volume control — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →
  13. 13 Access control, security & the vendor's reliability — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →
  14. 14 Access control, security & the vendor's reliability — found from sitemap www.elastic.co Checked 17 Nov 2026 Details →