Why didn't my alert fire?

Summary#

An alert has four chances to go missing: it was never set up the way you think, nothing was actually due, the background job did not run, or the delivery failed. Work through them in that order, because each one rules out the ones after it. Most missing alerts are settled in the first two steps, and you can check both yourself in under a minute.

Symptoms#

• A competitor page visibly changed but nothing appeared in the notifications feed.
• A keyword dropped several positions and no rank alert arrived.
• No email arrived from a scheduled report, or the report never ran.
• The Competitor Monitor card reads Watchtower alert channels / Not configured — changes detected but no alerts sent.
• The Rank Alerts screen promises a Slack ping and none arrives.
• A scheduled social post is still marked Scheduled well after its time.
• The notifications feed reads All caught up — no new alerts. on an account that has monitoring set up.
• An AI visibility drop alert never emailed, even though the chart shows a fall.
• A monitored URL shows a last-scan time from days ago.

Diagnosis#

Step 1 — Is it set up the way you think?#

The most common cause is not a fault. It is that the alert you expected has no delivery channel attached, or a preference is switched off.

Start with the two switches. Dashboard → Account & Team → Account → Preferences carries four notification switches. Two of them silence in-app items, and two gate the only two automation emails that exist.

See Notification preferences.

Then check the delivery channel on the item itself. This is where expectations most often outrun the setup.

Finally, confirm the item exists at all. Open the tool and look:

• Competitor Monitor — is the URL in the Monitored URLs tab?
• Rank Alerts — is the keyword in Tracked Keywords?
• Scheduled Reports — is the schedule in Your Active Schedules?
• Social Dashboard — is the post in Scheduled posts? The card is hidden entirely when the queue is empty.
• AI Tracking Settings → Alerts — is the alert listed, with the threshold you expect?

Step 2 — Was there anything due, and anything to report?#

An alert that correctly decided nothing had happened looks identical to one that never ran.

The first check never alerts. Every monitor stores a baseline before it can compare anything.

The threshold may not have been crossed.

A keyword that moved from #4 to #7 with a threshold of 5 is working correctly and will not alert.

The interval may not have elapsed. Check the elapsed time against the real cadence, which is on Automation limits and frequency. The short version: monitored URLs every 6 hours, rank alerts every 24 hours, AI visibility about every 22 hours, scheduled reports weekly, social posts at their set time.

The item may not be in the current rotation. AI visibility measures the five most recently used brands per pass. Track more than five and the least recently used stop being measured until you run a tool on them again.

Step 3 — Did the background job run?#

Every automation records when it last touched an item, and the app shows those timestamps. This is the fastest evidence that the machinery is alive.

If one item is stale and the rest are current, the scheduler is fine and the problem is with that item. A URL that returns an error is still stamped as checked so it does not block the queue, so a URL that consistently reports no changes may be failing to load. Open it in a browser: a page behind a login, a bot challenge or geo-blocking cannot be read.

If everything is stale at once, the background scheduler is not reaching the platform. Nothing you can change on your side will fix that, and nothing you set up will be lost — the jobs pick up all outstanding work as soon as they run again. Contact support with a screenshot of the stale timestamps. See Contacting support.

Step 4 — Did delivery succeed?#

Only two automation emails reach a customer today: scheduled report ready and AI Visibility drop. If you are diagnosing a missing email that is not one of those two, go back to Step 1 — it is not a delivery problem.

For a scheduled report specifically, see My scheduled report never arrived.

For a scheduled social post that reads Failed, the queue does not show which channel refused it. Press Publish on the same text to get the per-channel result immediately, which names the real cause, usually an expired connection. See Scheduled social posts.

Resolution#

Work down the list and stop when the answer appears.

1. Open the notifications feed first. Track & Report → Notifications & Alerts Hub, filter set to All. If the alert is there, this was never a detection problem, only a delivery expectation. See Notifications and alerts hub.
2. Turn the two feed switches back on. Account → Preferences. Competitor Alerts and SEO Score Alerts each silence a whole class of item.
3. Confirm the item exists in its own tool, and that its threshold and severity gate are what you expected.
4. Give it one full cycle. Six hours for a monitored URL, 24 hours for a keyword, about a day for AI visibility. New items need two cycles, because the first stores a baseline.
5. Check the last-run timestamps listed in Step 3. One stale item points at that item; everything stale points at the scheduler.
6. Test the source page. Open the monitored URL in a private browser window. If it needs a login or shows a challenge page, it cannot be monitored.
7. Add a recipient address to a schedule that only shows in-app only, and confirm Email Notifications is on.
8. Reconnect a social channel whose posts fail, then use Publish once to confirm it works before scheduling more.
9. Re-run by hand to prove the pipeline. Competitor Monitor and Rank Alerts both have an immediate check; running one confirms detection independently of the schedule.
10. Contact support with the tool name, the item, the timestamps you read in Step 3 and what you expected. See Contacting support.

Quick reference by symptom#

Prevention#

• Decide where each alert should land before you create it. Competitor changes and rank drops land in the app. Scheduled reports can email a named recipient. AI visibility drops email you. Anything else is not a delivery path that exists today.
• Put a recipient address on every schedule you actually want emailed. A blank field is the single most common reason a report "never arrived".
• Leave the notification switches on unless you are deliberately quieting something, and note what you switched off. A switch flipped six months ago is invisible when you are debugging today.
• Set thresholds to the movement you care about, not the smallest movement you can detect. A threshold of 3 positions on a volatile keyword produces daily noise, and noise gets muted, and muted alerts get missed.
• Keep the tracked brand set at five or fewer if you want each one measured every day.
• Check the connected channels on the Social Dashboard before scheduling a batch. An expired connection only announces itself after a post has already failed.
• Read the last-scan figures occasionally. They are the cheapest health check in the product, and they make a real outage obvious within a day rather than a fortnight.
• Do not look for the Webhooks or No-Code Automation screens. Neither is built and neither is in the menu any more. For webhooks that do fire today, see Webhooks.