Alerts
Told about the problem, not the noise
An alerting system earns its place by staying quiet. Cliqda watches seven specific failure modes across every connected account and speaks up when one of them actually happens.
Delivery
Where your team already is
Alerts go to Slack or email, with per-account muting so a known issue on one client does not train everybody to ignore the channel.
- Slack integration with per-account mute controls
- Email delivery with real failure handling — a bounced send is never marked as delivered
- An inbox in-app for the full history, filterable by account, type and status
- Incidents resolve when the underlying condition clears, not on a timer
Conversion tracking flatlined
14 min ago
Riverside Dental — 0 conversions recorded for 26 hours. Spend continued at $340/day.
Projected to overspend
2 hours ago
Northwind Outdoors is pacing to $41.2k against a $34k monthly budget.
Ad test reached significance
Yesterday
Variant B wins in 4 ad groups. Losing ads ready to pause.
Product illustration · Sample data, not customer results
Alert types
Seven things worth interrupting you for
Each one is a specific, checkable condition — not a generic anomaly score.
Budget
An account is pacing to miss its monthly budget, in either direction, with the projected month-end figure attached.
Over budget
Spend has already passed the ceiling you set. Separate from pacing, because by this point it is not a projection.
Conversion
Conversion volume moved enough to be worth looking at, with conversion lag accounted for so a reporting delay is not read as a drop.
Flatline
Conversions stopped being recorded entirely while spend continued — usually a broken tag, not a performance problem.
Delta
A metric moved sharply against its own recent baseline, so you hear about it the same day rather than at month end.
Bidding experiment
An experiment reached significance, or ran long enough that it is not going to.
Scheduled report
A scheduled report was delivered — or failed to be. Delivery failures are alerts, not silent log lines.
Why this matters
Silent failure is the expensive kind
An alerting system that fails open — marking a send as delivered when it bounced, resolving an incident because data stopped arriving, reading missing data as a flatline — is worse than no alerting at all, because you have stopped watching.
Cliqda’s staleness checks measure how much data actually arrived, not how long ago a sync last ran. A sync that completes in two seconds having fetched nothing is a gap, and it is reported as one.
Filters fail closed. If an alert cannot be evaluated, it is not quietly dropped from the results — the failure is surfaced. And a delivery that did not succeed is never stamped as sent.
Stop finding out at month end
Tell us about your accounts. We’ll help you explore the findings, the workflow and the results with a 14-day trial.
14-day trial · No credit card · Approval required by default