Diabolocom

Diabolocom exposes several object types through its API: phone calls, campaigns, contacts, users, groups, wrap-up codes, and tags. This integration focuses exclusively on phone calls, which constitute the fundamental unit of work for contact center agents and the only object type that supports per-agent attribution at the granularity required for performance management. Five KPIs were retained, selected against three criteria: ability to attribute the outcome to a specific agent, resistance to gaming, and balance between leading and lagging indicators.

Activity and efficiency: the volume-quality tension

Calls Handled counts the total number of interactions processed by an agent in a given period. Average Handle Time measures the average duration of those calls. Neither indicator is meaningful in isolation. A high Calls Handled count may reflect genuine productivity, or it may reflect rushed handling where agents terminate interactions prematurely to inflate throughput. A short Average Handle Time amplifies this ambiguity: it can indicate efficiency or inadequate resolution. Read together, the two indicators constrain each other: an agent who handles many calls at an unusually low average duration warrants closer examination of their outcome quality, whereas an agent with a rising Average Handle Time on stable volume may be encountering increasingly complex interactions or systemic obstacles.

Quality indicators: automated scoring and outcome measurement

AI Quality Score is Diabolocom's AI-generated rating assigned to each call based on transcript analysis. Because the rating is produced automatically from the recorded interaction, it is not susceptible to self-reporting bias. This makes it the most structurally reliable lagging indicator in the set: it measures what actually happened in the conversation, not what the agent or supervisor reports. When AI Quality Score declines for an agent whose Calls Handled is rising, the pattern suggests volume pressure is compressing interaction quality, a dynamic that is invisible when examining either indicator alone.

First Contact Resolution Rate is computed as the share of calls not followed by a callback from the same contact within 48 hours. As a proxy for resolution quality, it captures whether the agent genuinely resolved the customer's issue. An agent who handles high volumes quickly while producing a low First Contact Resolution Rate is generating repeat contacts downstream, which represents a real cost to the operation invisible in their individual activity metrics. This cross-indicator dependency is what gives First Contact Resolution Rate its diagnostic value: it is the only KPI in the set that measures the durability of an agent's work.

Process compliance: Wrap-Up Rate as a data quality signal

Wrap-Up Rate measures the share of calls for which an agent applied a disposition code at the end of the interaction. It is a leading indicator of data discipline rather than of customer outcomes: agents who consistently categorize their calls enable the downstream reporting and analysis that makes the other KPIs reliable. A sustained low Wrap-Up Rate for an agent degrades the interpretability of that agent's outcome metrics, since calls without categorization are absent from segmented analyses. Wrap-Up Rate is therefore not a performance indicator in the direct sense, but rather a prerequisite for the validity of the rest of the set.

Scope and limits of the integration

Diabolocom's API records call-level events and AI-generated ratings, but does not capture customer satisfaction directly. CSAT and NPS data, when collected, typically reside in a separate survey tool and are not available through the phone calls endpoint. The AI Quality Score is an automated proxy for interaction quality, not a direct satisfaction measurement, and the two should not be conflated. Outbound campaign metrics, which would apply to agents working in an outbound context, were excluded from this integration: campaign contacts reached carries a high Goodhart risk in pure outbound settings and is not a universal indicator across inbound and blended teams.

The reliability of these KPIs depends on consistent platform usage. Agents who handle calls outside the platform, pause recordings, or skip wrap-up steps will produce incomplete data. First Contact Resolution Rate, which depends on contact identity matching across calls, is additionally sensitive to the quality of contact data in the system. As with all activity-based integrations, the indicators measure what is recorded in the tool, not the totality of work performed.