Zendesk

Zendesk exposes several object types: tickets, ticket metrics, satisfaction ratings, users, organizations, and groups. This integration focuses on tickets and their associated metric and satisfaction data, which together provide the most direct indicators of individual agent performance. Other objects, such as organizations or groups, are useful for segmentation and load analysis but do not support per-user attribution in a meaningful way. Five KPIs were retained, selected against three criteria: ability to attribute to an individual agent, resistance to gaming, and balance between leading and lagging indicators.

Resolution volume and quality

Tickets Solved counts the number of tickets an agent has resolved in a given period. It is the most direct measure of contribution volume. Taken alone, however, it is the KPI most exposed to gaming: an agent can artificially inflate this count by resolving tickets prematurely or without fully addressing the underlying issue. Two indicators are paired with it specifically to neutralize this risk.

Ticket Reopen Rate measures the proportion of solved tickets that were subsequently reopened by the customer. It functions as a first-contact resolution proxy: a high reopen rate on an agent with a high Tickets Solved count exposes a pattern of superficial closures rather than genuine resolution. CSAT Rate measures the proportion of rated tickets that received a positive customer assessment. It is the only indicator in this set that captures the subjective dimension of the interaction — whether the customer felt the issue was handled well, not merely that a status was changed. Reading these three indicators together distinguishes productive throughput from inflated throughput.

Speed and efficiency

Average First Reply Time measures the elapsed time between ticket creation and the agent's first response. It is the only leading indicator in this set: it reflects current agent responsiveness before any resolution outcome can be measured, making it actionable on a daily or weekly basis. Average Full Resolution Time measures the total calendar time from ticket creation to final closure, including all intermediate states such as pending and on-hold. These two indicators are in a relationship of tension that reveals process bottlenecks. A short Average First Reply Time combined with a long Average Full Resolution Time indicates that agents acknowledge tickets quickly but then stall during resolution — a pattern consistent with escalation delays, unclear ownership of complex issues, or heavy dependency on third parties. Conversely, a long Average First Reply Time with a relatively short Average Full Resolution Time suggests a triage bottleneck at intake that resolves once work begins.

Scope and limits of the integration

Zendesk records status transitions and satisfaction responses, but does not capture the nature or complexity of the work performed. A ticket marked solved may represent a two-minute password reset or a multi-day investigation requiring cross-functional coordination; the API data treats both identically. CSAT Rate is also subject to a sampling limitation: surveys are not sent for all ticket types, and response rates vary by customer segment and ticket volume. The denominator is restricted to tickets where a rating was actually submitted, which introduces selection bias on agents handling ticket types where surveys are less likely to be returned.

More broadly, this integration only reflects activity that is recorded in Zendesk. Customer interactions handled through informal channels — direct emails, phone calls outside the platform, or messaging tools not connected to Zendesk — remain invisible. The accuracy of all five KPIs depends on consistent ticket assignment: if tickets are regularly left unassigned, reassigned without closing the original, or resolved in bulk without per-agent attribution, the data loses individual-level reliability.