Custom Metrics Monitoring

Monitor the numbers that run your business.
Not just CPU and uptime.

Track business-specific and technical metrics with your own thresholds: queue depth, throughput, conversion, failed orders. Define incident triggers around your SLOs and catch async slowdowns before users feel them.

custom-metrics / failed-orders
OK
Failed Orders Last Hour
Behavior1h6h24h7d
Latest8.00 Orders
> 25 Orders
27.920.413.05.6-1.9
MIN
1.00 Orders
AVG
6.93 Orders
MAX
13.00 Orders
Orders / hr
Criteria
Alert when value is greater than 25 for 60s
Column value
> Greater than
25 Orders

Your stack, your rules.

Stop monitoring only what your infrastructure exposes. Push any number (from any source) and define exactly when it means something is wrong.

Criteria BuilderCustom Metric
failed_orders_per_hour
Column value
>
25
Orders
Alert preview
Alert when failed_orders_per_hour > 25 Orders

Your metrics, your thresholds

Define exactly what healthy looks like for your business. Push any numeric value (throughput, queue depth, conversion rate, error counts) and set the precise threshold that triggers an incident.

  • Alert when conversion, queue depth or throughput shifts
  • Column-value / count / threshold criteria
  • Units and human-readable labels

SLO-aligned incidents

Tie every alert to an objective. Sustained-breach windows let you ignore transient spikes and fire only when a threshold stays violated for the window you care about.

  • Define triggers around your reliability targets
  • Sustained-breach windows to avoid flapping
  • Route to the owning team

SLO Breach Window

Failed Orders / hrBreached
0Threshold: 2550
Breach window
Sustained for 60s
60s
Current value
Exceeds threshold
41 Orders
Incident triggered, paged on-call engineer
ALERT
Queue Depth
8,431
jobs
Threshold: 5,000
Queue depth has grown continuously for 47 minutes. No jobs drained in the last check window, worker process may be stuck.

Catch async slowdowns

Silent pipeline failures are the hardest to detect. Jobs back up, queues stop draining, and users see nothing, until it is too late. Track the numbers that reveal what is happening behind your HTTP layer.

  • Detect when queues stop draining
  • Catch silent pipeline backups
  • Correlate with incidents and AI root cause

Alert on what actually matters to your business.

Define your first custom metric in under a minute.

Create Your First Custom Metric