Why status pages matter and how to set one up

Build a service status page that stays reachable during incidents. Separate the publishing path, choose useful components, and turn monitoring evidence into customer updates.

Best practices
Illustration of regional services around a shared status dashboard

Checkout is timing out in one region. The on-call engineer can see failed probes, but customers are already asking whether their orders went through. A status page gives support and customers the same place to check what is affected, when the team last verified it, and when another update is due. The incident team can keep the diagnostic details in its own channel.

A useful status page needs two working paths. Customers must be able to read it when the application fails, and an authorized responder must still be able to publish. Work out both paths before choosing a design or a host.

Keep the page reachable during an outage

status.example.com gives customers an address they can remember. Whether it survives a production failure depends on the systems behind it. A status site on a separate host can still share an authoritative DNS provider, a CDN account, or the same identity provider responders use to sign in. Check the read path from outside your network and rehearse publishing during a simulated production or sign-in outage. Keep an approved backup communications channel in the runbook.

A custom build gives you control over the publication workflow and leaves the team responsible for hosting, updates, and certificates. A self-hosted open-source project can provide a starting point, while you still operate it. A hosted service takes on its own infrastructure; you still own the domain setup, accounts, and communication process. Compare those responsibilities and test how each option handles the failures you're trying to isolate. Atlassian's incident communication guidance

Show the parts customers can act on

Name components after work customers recognize, such as Checkout, Order History, and the API. Add a region or product distinction when it changes what someone can do. If payment attempts fail while order history works, showing both components is more useful than a single red banner. Keep the current impact and the latest update easy to find on a phone. Branding can help people recognize the page; readability and a small page payload matter when people are trying to check it quickly.

Decide which information belongs on a public page and which needs a restricted customer or internal channel. Publish impact, an approved workaround when there is one, and the next update time. Keep account-specific findings, request identifiers, and internal diagnostics in the appropriate channels. An unlisted URL is still shareable. Use a product that supports the access controls your restricted audience needs.

Turn monitoring evidence into an update

Consider a fictional checkout incident. Two consecutive one-minute HTTP probes from Singapore fail at 10:01 and 10:02 UTC, while checks from Tokyo pass. In this example the failures page a responder. By 10:04 UTC, production request data confirms checkout timeouts in Singapore and normal order history access. The communications lead can now publish the impact. A probe result on its own describes a particular check and location.

Time (UTC) Example public update
10:05 We're investigating checkout timeouts in Singapore. Order history is available. Next update by 10:20 UTC.
10:20 We've applied a mitigation and are checking checkout results. Next update by 10:35 UTC.
10:35 Checkout has recovered in Singapore. We confirmed 15 minutes of normal checkout results through 10:35 UTC. Contact support if you're still seeing problems.

The 15-minute cadence and recovery check are choices for this scenario. Set yours according to the impact and what users need to decide. Assign a communications lead and a backup before the first page. The first note can go out with the facts already confirmed, and it can promise another update time even when there is no reliable restoration estimate. Google SRE on incident roles and stakeholder updates

Separate component status from the incident record

Guance HTTP synthetic tests provide probe results to investigate. For a custom Guance status page, the publisher needs Status Page configuration permission and read access to the service-level objective (SLO). Before publishing, set at least one service to display publicly and link it to an enabled SLO without grouping dimensions. Service health on that page follows the associated SLO attainment and thresholds. A published maintenance record can put its service in maintenance on the specified business date. An incident record explains what users experienced and what the team did.

For an incident record, link a public note you can read in the same workspace. If the source note is already linked to a published status-page record, saving edits to its title or body updates the public page. Changes to the business date, affected services, or display settings require an updated publication. Check the text and the page preview at the corresponding step. For an outage like the one above, the note can tell customers checkout is affected even when a longer SLO window still meets its target.

Make the page easy to find before you need it

Put the address in the help center and footer, and give support the same link to use in replies. Test that a responder can publish a draft update under your access policy and find it from an external connection once published. Keep history for a defined period and describe which components and time window an uptime chart covers. Track the chart’s display window and the retention policy separately.

Service status page basics · Status page design examples · Incident update templates

Get a tailored plan

Contact us

Join the community

Scan with WeChat
to join the community

Try Guance

Start online and pay only for what you use.

Get started

Choose a Guance plan

Code hosting