Alerts and incidents

When something starts going wrong on a site, Monitor opens an incident and decides whether to tell anyone. An incident is one event, such as “40 pages started returning 404”, however many pages it touches. The Alerts tab lists a site’s incidents, and alert emails are sent about them.

Before you start: Monitor must be enabled for your workspace. It is set up by arrangement, so if you do not see the Monitor icon in the top bar, contact us. To also receive alerts in Slack, Slack must be connected in your workspace settings.

One problem, one incident

Monitor bundles findings: one check failing on 100 pages is one incident with 100 pages, not 100 alerts. A single check round sends each recipient at most one email per workspace, however many incidents or sites it covers. The same bundling applies to changes, so a template change that rewrites every title opens one incident.

How an incident behaves

It opens when a check starts failing, or when a watched value changes. See Issues and severity and Changes and diffs.
It grows as more pages are affected, and pages drop off as they recover.
It closes when every page has recovered. It also closes after 24 hours with no newly affected and no newly recovered pages, even if some pages are still affected. The finding is still listed on the Issues tab.
It stays listed. Closed incidents remain on the Alerts tab for a week.

If a finding that closed this way is still there, Monitor does not email about it again unless it has grown.

Open an incident to see how many pages are Affected and how many have Reverted, when it was Detected, when it Closed or had its Last change, the Was and Now values for a change, and What to do.

What emails you by default

Every new site starts with these alert rules:

Rule	Why it is on
The site is not responding	The site is down
robots.txt blocks search engines from the entire site	The whole site is leaving search
The TLS certificate expires soon	An outage with a known date
A page that earns traffic has broken	A page that brings in search traffic has broken
Not indexable	Pages leaving the index. Set so one deliberate `noindex` stays quiet but a template-wide one does not
Analytics tracking removed	Measurement has stopped on a page
Any incident	Anything else severe enough, at the least sensitive setting

On top of the rules, the site-wide setting Only when it is at least is warning by default, so Info findings are never emailed unless you lower it.

Alerts go to everyone in the workspace until you choose recipients.

Nothing is emailed while a site’s first pass is running. Page problems found during it are never emailed: they are the site’s existing condition. Whole-site problems, such as the site not responding, are emailed when the first pass finishes if they are still open.

Change who is told and when

All alert settings are on the site’s Settings tab, under Alerts.

Add or edit an alert rule

A rule decides, for one kind of finding, how widespread it must be before anyone is emailed.

Under Alert rules, select Add rule, or Edit on an existing rule.
In Alert on, choose a check, or Any incident.
In Scope, choose Whole site or a segment.
Move Sensitivity between only big events and almost anything. The caption under it tells you how many pages it takes to fire on this site. A few important pages count as much as many unimportant ones.
Optional: in Recipients (optional), enter email addresses. Leave it empty to use the site’s recipients.
Optional: select Also post to Slack. Each recipient also gets a Slack direct message, in addition to the email, never instead of it.
Select Save rule.

A rule for a specific check takes priority over Any incident. To remove a rule, select Delete.

Change the site-wide settings

Under Fallback: when to send:

Setting	What it does
Email someone when this site has an incident	Turn off to stop all alert emails for the site. The site is still monitored and incidents are still listed
Only when it is at least	The lowest severity that is emailed. Applies to every incident, including ones handled by a rule
And affects at least	Pages that must be affected. Used only for incidents that no rule covers
And one of them is at least this important	Minimum page importance, from 0 to 10. 0 ignores importance. A value above 10 is not saved. Used only for incidents that no rule covers
Hold non-critical alerts overnight	Quiet hours, with From, Until and a timezone. Defaults to 22:00 to 07:00 when turned on
Send non-critical alerts once a day	A daily digest at the time you set in At. Defaults to 09:00 when turned on
Who to tell	The site’s recipients. Leave empty to email everyone in the workspace

Every new site has an Any incident rule, so And affects at least and And one of them is at least this important only take effect for page findings if you delete that rule.

Critical incidents are never held by quiet hours or the daily digest.

Select Save alert settings when you are done.

Stop an alert you do not need

If an alert is about something that is fine on your site, stop reporting that finding on the page, in a segment, or across the site. See Issues and severity.