Filters and Eligibility
AI insights are not always generated for every conversation. This is usually intentional.
MiaRec supports filters and eligibility rules to ensure insights are:
- relevant
- accurate
- cost-effective
Why filters exist
Filters help avoid running tasks on conversations where an insight would be unreliable or meaningless, such as:
- extremely short calls (“Hello?” … hang up)
- internal/test conversations
- conversations outside the task’s scope (e.g., sales insight on support calls)
Common filter examples
Depending on your deployment and configuration, filters may include:
- Direction: inbound only or outbound only (voice calls)
- Duration: only conversations longer than a threshold (e.g., > 15 seconds)
- Channel: voice vs chat vs ticket (when enabled)
- Queue/team: only apply to specific business units
- Tags/metadata: apply based on custom business labels
Eligibility prerequisites (even without filters)
Some tasks require basic conditions:
- transcript/thread text must exist
- enough content must be present to infer the result
- language support (where applicable)
If prerequisites are not met, the task may be skipped or return “unknown / insufficient evidence.”
What to do if an insight is missing
From a user perspective, common reasons include:
- the conversation was too short
- the conversation did not match the task’s filters
- the transcript/thread is missing or incomplete
- the insight is not enabled for your organization
If missing insights are a recurring issue, an administrator should review:
- filters (are they too strict?)
- data quality (are transcripts available?)
- whether the insight is enabled
Next: practical use cases
To see how teams use filtered insights in real workflows: