Overview
An SLA (service level agreement) sets expectations for how reliably a service performs. This article explains how SLA adherence is measured in Stax.ai CX: the signals used to confirm core workflows are healthy, what counts as impact, and what happens when an issue is detected. Use it as a reference when reviewing your SLA terms or discussing an incident with support.
Key terms
Term | What it means |
Heartbeat metric | A high-volume signal that shows whether a core workflow is working normally. |
SLA adherence | How consistently the service meets the levels defined in your SLA terms. |
RPO (Recovery Point Objective) | The acceptable data-loss target for incident scenarios. |
RTO (Recovery Time Objective) | The target time to restore service after a major availability event. |
What is a heartbeat metric?
A heartbeat metric is a high-volume indicator that shows whether a core workflow is functioning normally. Because these actions happen constantly, a sudden change stands out quickly. Examples include:
Messages created and delivered from Inbox workflows
Replies saved and sent successfully
Task updates written and reflected in workflow views
Automation runs completing without failure
File upload and retrieval success rates
SLA categories
Teams commonly track two categories:
Core platform workflow SLA — whether core work paths (Inbox, tasks, automations, files) remain usable.
AI workflow SLA — whether AI-assisted actions respond and complete within expected service levels.
When impact counts against SLA adherence
Impact counts against adherence when:
A complete outage prevents use of a core workflow.
A material degradation significantly slows or blocks expected actions.
A critical feature path fails repeatedly during normal usage windows.
Note: Because of distributed architecture and how issues are detected, calculated impact can reflect the worst-case behavior observed during an incident window.
How issues are detected and handled
CX monitoring uses anomaly detection — comparing current behavior against expected patterns — so both outages and slowdowns can be caught early. When a heartbeat metric regresses, the typical response flow is:
Alerting triggers on the heartbeat metric regression.
Incident response is activated and ownership is assigned.
Recent risky changes may be rolled back to stabilize service.
Root-cause analysis and remediation actions are tracked.
Planned maintenance
When maintenance may affect workflow availability, you should be notified in advance, following the communication policy in your SLA terms. Planned maintenance is handled separately from unplanned incidents.
How to report a suspected SLA issue
If something in CX seems slow or unavailable, gather details before you reach out — it speeds up triage:
Note the timestamps when the issue started and (if it has) ended.
Note which workflows were affected, such as Inbox, tasks, automations, or files.
Gather example records, like a task or message that failed to save.
Contact your Stax.ai support team with these details.
Related articles
Dashboards Overview
Understanding Filters, Views, and Reporting
Milestone Reports