Skip to content

Free lesson1 of 115 min video4 question quiz

Design JSM status pages that reach the right people

A 15-minute lesson that designs Jira Service Management status pages for a company with 3 products on one shared platform, builds them on a demo site, and runs an incident through them: one ticket, one update, and only the affected products marked. It also covers closing out, stakeholders, subscribers and webhooks, and a phased rollout.

Level
Intermediate
Products
Jira Service Management
Format
On demand, free, no sign-up

Watch the lesson Take the quiz

A page is an audience

Watch this from 0:41

Three products, one shared platform: Projects, Chat and Docs each with its own customers and status page, and one shared platform underneath all three

Most teams start with their systems and make a page for each one. Start with your audiences instead: ask who needs to hear about an outage, and where they already look.

The example in this lesson is a software company with 3 products: Projects, Chat and Docs. Each product has its own customers and its own status page. Underneath all 3 is a shared platform, with sign-in and notifications. That shape is common, and it is where status pages get hard.

One page meansSo
One web addressCustomers bookmark it and support links to it
One set of customersEverything on it is news that audience needs
One subscriber listEvery update reaches everyone subscribed to it

If two groups of customers never need the same news, they need different pages. To see the pages on a site, open Operations, then Status pages.

Pages, components and incidents

Watch this from 1:45

A group shows the worst status inside it: the Sign-in group at partial outage because Projects sign-in is at partial outage, while Chat sign-in and Docs sign-in stay operational

JSM builds every status page from 3 things.

PartWhat it is
PageThe audience: one address, one set of customers, one subscriber list
ComponentThe thing that broke, backed by a service in your services registry
IncidentOne record with one lifecycle, linked to one page and one Jira work item
  1. Open the page's menu and choose Edit page.
  2. Choose the Components tab. Each component points at a service in your services registry, so the page speaks the same language as your incidents.

Components can also be grouped. A group shows the worst status of anything inside it: if one component in the group is at partial outage, the group shows partial outage. That is the feature this whole design is built on.

Native status pages come with the Service Collection Enterprise plan. Source: Atlassian's status pages documentation.

One incident, one page

Watch this from 2:48

One incident copied to three product pages by three fixed-page automation rules, with the Docs page left open after the resolve

This is the rule that shapes everything else. Once an incident is linked to a status page, the page is locked: the Add status page button is gone, and Send update stays on the same page. Atlassian's docs say it plainly: you can link only one status page per incident. Source: Post stakeholder updates directly from incidents.

Teams often try to work around it with automation that copies an incident to every product page. We tested that on the demo site:

What we triedWhat happened
One rule that picks the pageThe automation action's page is a fixed picker, so every page needs its own rule
Copying one outage to 3 pagesAfter the resolve, one page was left with an open incident and a degraded component, and nothing closed it

So don't fight the rule. Design with it.

Build a platform page for shared services

Watch this from 3:48

The whole setup: one platform page, a group for each shared service, a component for each product named for the product, and a header link on every product page

The pattern: shared services go on one platform page. Each shared service is a group, like Sign-in, and inside the group is one component for each product it serves. Now one incident on one page can say exactly which products are affected. Each product page keeps its own components, plus a header link to the platform page.

  1. In Status pages, choose Create status page, then Create under Public.
  2. Type the name and a short, readable web address, then choose Create. JSM saves the new page as a draft, so nothing is public yet.
  3. Open the Components tab and choose Add component. Name it for the shared service, such as Sign-in.
  4. Choose Add subcomponents. That turns the component into a group.
  5. Name the first subcomponent for its product, such as Projects sign-in, and pick its service from the services registry. Choose Save components.
  6. Add the other products the same way, and a group for each other shared service. A service that never fails by product, like an admin console, stays a single component.
  7. Choose Back, open the page's menu and choose Publish page, then Publish.
  8. On each product page, open Edit page, then Appearance, then Page title. Choose Add link, type a label such as Core services status and the platform page's address. Save the draft and publish.

Watch out. Name every component for its product, like Projects sign-in, not just sign-in. Subscribers and webhooks see the component name on its own, without the group around it.

The whole setup
One platform page
A group for each shared service
A component for each product, named for the product
A header link on every product page

No routing automation anywhere.

Run an incident from one ticket

Watch this from 6:47

The public Example Core Services page during the incident: the Sign-in group at partial outage, Projects and Chat marked, Docs sign-in not affected

In the example, sign-in is failing for Projects and Chat, and Docs is fine.

  1. Open your incidents and choose Create. Type a clear summary that names the service and the products it affects, so anyone scanning the queue understands it. Set the affected service, then choose Create.
  2. Open the new incident, expand its Status pages panel and choose Add status page.
  3. Pick the platform page. The status starts at Investigating.
  4. Type a title that names what users notice, not the cause. Type the message customers will see: what is affected, and when they will hear next.
  5. Open the Components tab and expand the group. Set each affected product's component, here Projects and Chat, to partial outage. Leave the others operational. That is the only new step compared with a normal incident.
  6. Choose Send update.

If every product were down, set the group row instead. It sets every component in the group in one action.

On the public platform page, the Sign-in group shows partial outage, Projects and Chat are marked, and Docs is still green: one incident, one update, and only the right products. The product page's header link sends customers straight there.

StatusWhen to post it
InvestigatingThe first update
IdentifiedYou know the cause
MonitoringA fix is in and you are watching it
ResolvedIt's over

Each later update uses Send update in the same panel and goes to the same page.

Close it out cleanly

Watch this from 9:12

Close it out in three steps: send a Resolved update, set every component back to operational, resolve the work item
  1. Choose Send update, set the status to Resolved, and type a closing message that says what was fixed and that it's over.
  2. Open the Components tab, expand the group and set each affected component back to operational. Choose Send update.
  3. Resolve the work item itself: open its status, choose Resolve, pick a resolution such as Done, and choose Update.

Watch out. Resolving the incident does not reset the components. Skip step 2 and a page stays red after the outage is over. Put all 3 steps in your runbook.

Internal audiences use stakeholders

Watch this from 10:25

Two audiences, one incident: the public update reaches customers, and stakeholder email from the incident reaches support, sales and leadership

Support, sales and leadership need more detail than customers, and sooner. Since an incident can't reach a second page, use stakeholders: they get updates by email, straight from the incident.

  1. On the incident, open the Stakeholders panel.
  2. Add the people or teams who need to know.
  3. Choose Notify stakeholders and write the internal update.
ChannelAudienceFrom
Status page updateCustomersThe incident
Stakeholder emailSupport, sales, leadershipThe same incident

Stakeholders are on the Premium and Enterprise plans, and they don't need a license or access to the incident. Atlassian allows customers and partners as stakeholders too, so treat them as a separate email channel rather than an internal-only one. Sources: What is stakeholder communications? and Send notifications to incident stakeholders.

Plan subscribers and in-product status

Watch this from 11:05

Show status inside the product: one update on the platform page, webhook JSON with the affected components, the product app keeps its own components and shows a banner until the incident resolves

Before you move anyone, count your subscribers. Each site supports up to 10,000 subscribers in total, across every page combined, and active, pending and inactive subscribers all count. Export the counts from your current pages first: they decide your migration order. Source: Manage status page subscribers.

The best way to stay inside the limit is to show status inside your product:

  1. Each product app subscribes once to the platform page by webhook.
  2. JSM sends every update as JSON, with the affected components by name and status.
  3. The app keeps only the components with its own name.
  4. It shows a banner until the incident resolves.

One subscription reaches every user of that product.

To subscribe a webhook, open the public page, choose Subscribe to updates, then Webhook. It asks for the endpoint and an email for failed deliveries. Subscriptions are email or webhook only. Source: Subscribe to a public status page using webhooks.

For business customers, subscribe one operations address, not every member of staff.

Before and after, gaps, and rollout

Watch this from 12:09

The same outage, two designs: one page per product took 2 Jira incidents and 6 status page updates, a platform page took 1 incident and 3 updates

On the demo site, the same shared outage hitting 2 products took:

DesignJira incidentsStatus page updates
One page per product26
Platform page13

These are counts on demo data, not timings. As more products are affected, one page per product grows and the platform page stays flat.

GapCover
Product pages stay green during a shared outage, because it shows on the platform pageThe header link and the in-product banner
JSM's own pages don't offer component subscriptions yet, so a subscriber gets the whole pageWebhooks filtered by component name

Roll it out in 4 phases, each with a gate:

PhaseWorkGate
PrepareConfirm who owns each page and count subscribersThe counts fit the limit
BuildThe platform page, the header links and the runbookA test incident run end to end
PilotA few products on real incidentsThe runbook is tuned
ExpandThe rest of the products in wavesEach old page retired as its product moves

Check what stuck

4 questions on this lesson. Pick an answer to see whether it is right, and why. Nothing is sent anywhere and there is nothing to sign up for.

  1. Question 1 of 43 products share a sign-in service. Where does the sign-in outage go?

    The answer: On one platform page, with a component for each product inside a Sign-in group.

    One incident reaches one page, so shared services live on one platform page. The components inside the Sign-in group mark exactly which products are affected.

  2. Question 2 of 4Can one JSM incident post to 2 status pages?

    The answer: No.

    Once an incident is linked to a page, the page is locked and Add status page is gone. Atlassian's docs say you can link only one status page per incident, so design with that rule.

  3. Question 3 of 4You've sent the Resolved update. What's left?

    The answer: Set every component back to operational, then resolve the work item.

    Resolving the incident does not reset the components. Set each one back to operational, then resolve the work item so the page and the ticket agree.

  4. Question 4 of 4How do you tell your support team more than customers see?

    The answer: Add stakeholders to the incident.

    An incident can't reach a second page. Stakeholders get email updates straight from the incident, so your team can get more detail, sooner.

Who teaches it

Riley Venable

Riley Venable

VP of Services, Atlas Bench

Full-stack engineer, Atlassian Certified Instructor, and Atlassian Community Champion who works on the layer underneath the tools: identity, infrastructure, and the systems that decide whether AI can be trusted with them.

Plan your move to JSM status pages.

The pages in this lesson run on demo data. An Atlas Bench architect can design your pages, components and rollout around your own products and subscriber counts.