Taking on 2–3 new engagements in 2026 — EST & PST hours
Start a ConversationSupport operations glossary
Definitions for the metrics, workflows, and investigation concepts B2B support teams use to measure speed, quality, escalation load, and technical diagnosis.
After-call work is all work performed after a ticket interaction ends but before the agent is fully available for the next ticket, including documentation, system updates, follow-up scheduling, and knowledge base contributions. It is broader than simple wrap-up because it can include multi-step internal tasks.
Read definition →Agent occupancy rate is the percentage of an agent's paid working time spent actively handling tickets, including investigation, communication, and documentation, rather than idle time, training, or administrative work. It measures how full the agent workload is during scheduled time.
Read definition →Agent shrinkage is the percentage of scheduled agent time unavailable for ticket handling because of breaks, training, meetings, administrative work, absenteeism, or system downtime. It measures the gap between scheduled capacity and effective capacity.
Read definition →Agent utilization rate measures how much of an agent’s working time is spent on active support work versus idle time, meetings, or admin tasks. In technical support, utilization becomes dangerous when agents are busy but still blocked by manual investigation steps they cannot complete quickly.
Read definition →An AI agent is software that can interpret a goal, choose actions, call tools, observe results, and continue until it reaches a stopping condition. Unlike a plain chat interface that only returns text, an agent can read tickets, query APIs, inspect logs, open bug reports, and draft replies based on live system evidence. In production settings, the important question is not whether it sounds intelligent, but whether it can operate reliably against real permissions, failure states, and workflow rules.
Read definition →An AI support agent is software that handles support interactions autonomously, from reading tickets to investigating issues to drafting responses. Unlike chatbots, AI support agents connect to production systems.
Read definition →API error investigation is the process of diagnosing why an API request failed for a specific customer, integration, or time window. It typically involves checking request logs, authentication state, rate-limit headers, recent deploys, schema changes, upstream dependency health, and account configuration. In B2B systems, this work is rarely solved from the ticket text alone because the same HTTP status code can come from very different causes depending on customer setup and service state at the moment of failure.
Read definition →API rate limiting is the practice of restricting how many requests a client can send to an API within a defined window, such as per second, minute, or day. Platforms use rate limits to protect shared infrastructure, prevent abuse, and keep latency predictable under load. Limits may be enforced with token buckets, leaky buckets, or fixed windows, and are often scoped by API key, workspace, IP address, or endpoint class depending on how the service is designed.
Read definition →Average Handle Time is the average total time an agent spends on a ticket, including investigation, response drafting, customer communication, and post-ticket documentation. It shows how much labor is required to move one ticket through the full support workflow.
Read definition →Average Speed of Answer is the average time between a ticket being submitted and an agent providing the first substantive response, not an automated acknowledgment. It measures how quickly a human meaningfully engages with new demand.
Read definition →ClickHouse is a column-oriented analytical database built for very fast aggregation and scan-heavy queries on large event datasets. It is commonly used for API logs, product analytics, observability events, and customer activity history because it can process billions of rows with low latency when queries are written well. In support operations, ClickHouse is valuable because agents and automation can search recent request failures, workspace activity, and account-level patterns without pulling data into a separate warehouse first.
Read definition →Contact volume is the total number of support interactions received across all channels, including email, chat, phone, and in-app messaging, within a measurement period. It is the input signal for every staffing and queue-management decision.
Read definition →Customer Effort Score measures how easy it was for a customer to get their problem solved. In support, CES improves when the customer does not need to repeat context, chase updates, or wait through multiple internal handoffs.
Read definition →Customer escalation path is the documented sequence of people, teams, and channels used when a support issue exceeds normal handling scope or urgency. It defines who gets involved at each stage, what evidence must accompany the handoff, and how the customer is kept informed during the transition. A strong escalation path reduces confusion during outages, protects executive relationships, and prevents support teams from improvising under pressure when a large account or revenue-critical workflow is affected.
Read definition →CSAT measures customer satisfaction with a specific support interaction, typically on a 1-5 scale collected via post-resolution survey.
Read definition →Deflection effectiveness is a quality-adjusted measure of ticket deflection that accounts for whether deflected tickets were genuinely resolved or merely abandoned by frustrated customers. It distinguishes successful self-service from failed self-service.
Read definition →An escalation matrix is a predefined routing table that specifies who receives escalated tickets based on ticket type, severity, customer tier, and time-in-queue thresholds. It turns escalation from judgment-based behavior into a documented operating rule.
Read definition →An escalation path defines who should handle a support issue when the frontline team cannot resolve it, and in what order the handoff should happen. In B2B technical support, a clear escalation path prevents tickets from bouncing randomly between support, product, and engineering.
Read definition →First Call Resolution measures the percentage of support tickets resolved in the first interaction, without escalation, callback, or follow-up contact required from the customer. In B2B support, it is a direct test of whether the first responder had enough information and authority to actually solve the problem.
Read definition →First Contact Resolution measures the percentage of tickets solved in the first meaningful reply without follow-up handoffs or repeated back-and-forth. In B2B support, true FCR usually depends on whether the first response includes diagnostic evidence, not just empathy and next steps.
Read definition →First Response Time measures how long a customer waits between submitting a support ticket and receiving their first human response. It is the most visible support metric from the customer's perspective.
Read definition →Forward deployment is a go-to-market and engineering model where technical team members work directly with customers during implementation, onboarding, and issue resolution. Instead of throwing requirements over a wall to product teams, forward-deployed engineers configure integrations, debug live workflows, and feed product insights back into the roadmap. The model is common in enterprise software where each customer environment is different and time-to-value depends on hands-on technical problem solving during the first weeks of adoption.
Read definition →Handle time is the total elapsed time from when an agent opens a ticket to when they close it, including all active work, hold periods, and transfers. It represents the end-to-end ticket lifecycle from the agent's perspective rather than just the average across many tickets.
Read definition →Hold time is the total time a customer waits for a response after submitting a ticket across all wait periods in the lifecycle, including initial response, investigation holds, and between-reply gaps. It is the customer-experienced latency inside an otherwise “active” ticket.
Read definition →Incident management is the process of detecting, responding to, and resolving service disruptions. In B2B SaaS, incidents often surface first through customer support tickets before monitoring alerts.
Read definition →Issue taxonomy is a hierarchical classification system for support issues that organizes tickets into standardized categories, sub-categories, and root-cause types. It provides a stable framework for analysis, routing, and trend detection across the support org.
Read definition →Knowledge base deflection measures how often documentation helps a customer solve an issue without creating a support ticket. It is most valuable for onboarding questions, setup steps, and common workflow answers.
Read definition →LangChain is a software framework for building LLM applications that combine prompts, memory, tool use, retrieval, and workflow control in code. Developers use it to define chains and agents that move between model calls and external systems such as vector stores, SQL databases, ticketing tools, and web APIs. In support automation, it is commonly used to structure retrieval, tool calling, and response generation so the application can work with live customer and operational data rather than static prompt text alone.
Read definition →Mean Time to Acknowledge measures the average elapsed time between when an issue is created and when a responsible human or system explicitly recognizes ownership. In support and incident operations, acknowledgment is different from full diagnosis or first resolution update. It tells you whether the team saw the problem and started handling it. MTTA is especially important for high-priority tickets because customers interpret silence as inaction even when the team is already investigating behind the scenes.
Read definition →Mean Time to Resolution measures the average time from when a support ticket is created to when it is fully resolved. For B2B technical support, MTTR is the primary metric that drives customer satisfaction and retention.
Read definition →Model Context Protocol is an open protocol for exposing tools, prompts, and structured resources to LLM-based applications in a standardized way. Instead of wiring each model integration separately, MCP servers present capabilities such as database access, ticket search, or file retrieval through a common interface. That makes it easier to give an AI system controlled access to external context while preserving permission boundaries, typed inputs, and predictable transport behavior across different clients and models.
Read definition →n8n workflow automation is the use of n8n to orchestrate API calls, condition branches, database queries, and message handoffs across multiple systems without writing every step as custom backend code. Teams use it to connect help desks, CRMs, billing platforms, internal tools, and LLM services into event-driven workflows. It is especially useful for support operations because a ticket can trigger enrichment, severity checks, bug search, stakeholder alerts, and response drafting inside one graph-based execution path.
Read definition →NPS measures customer loyalty by asking how likely they are to recommend your product (0-10 scale). Promoters (9-10) minus Detractors (0-6) equals your NPS.
Read definition →Proactive support means identifying and addressing customer-impacting issues before the customer opens a ticket. In B2B SaaS, that can include noticing error spikes, failed webhooks, billing anomalies, or rollout regressions before they turn into escalations.
Read definition →Production AI refers to machine-learning or LLM-powered systems that run inside live business workflows with real users, real data, and operational consequences. A production AI system must handle permission boundaries, latency budgets, failure recovery, monitoring, and change management, not just answer a benchmark prompt correctly. In support operations, that means the AI needs to read tickets, query systems safely, return evidence-backed outputs, and fail in predictable ways when a dependency is unavailable or data is incomplete.
Read definition →Queue depth is the number of tickets currently open and awaiting agent handling, measured at a point in time or as an average over a period. It is a leading indicator of whether the team is about to miss service targets.
Read definition →Resolution rate is the percentage of tickets fully resolved within a given period, typically measured as daily or weekly throughput relative to incoming ticket volume. It shows whether the team is keeping up with demand or silently accumulating backlog.
Read definition →Root cause analysis in support context is the process of identifying the underlying technical reason for a customer-reported issue, rather than just addressing the symptom.
Read definition →Schedule adherence measures the percentage of scheduled working time during which support agents are actually available and handling tickets, compared with their published work schedule. It captures the gap between staffing plans and real queue coverage.
Read definition →Self-service rate is the percentage of customer issues resolved through self-service channels such as documentation, knowledge bases, FAQs, or chatbots without human agent involvement. It measures how often customers can solve the issue on their own.
Read definition →Service level is the percentage of tickets answered within a target time threshold, typically expressed as “X% of tickets answered within Y minutes.” Unlike an SLA, service level is usually an internal operational target rather than a contractual promise.
Read definition →A Service Level Agreement (SLA) is a contractual commitment between a vendor and customer that defines the maximum acceptable response and resolution times for support tickets by priority level. SLA breaches — where actual resolution time exceeds the committed window — trigger financial penalties including service credits, pro-rated refunds, or contract termination clauses. For B2B SaaS companies, SLA compliance is a direct revenue protection mechanism.
Read definition →SLA compliance rate is the percentage of tickets resolved within the response and resolution commitments promised in customer contracts. It is one of the clearest support metrics tied directly to retained revenue in enterprise B2B accounts.
Read definition →Support automation uses software to handle parts of the support workflow without human intervention. This ranges from chatbots answering FAQs to AI systems investigating technical issues by querying production databases.
Read definition →Support automation ROI is the economic return generated when automation reduces ticket handling cost, increases agent throughput, lowers escalations, or protects revenue through faster issue resolution. It should be measured against all-in costs including software, implementation time, maintenance effort, and failure risk. In technical B2B environments, the largest ROI usually comes from removing investigation labor and engineering interrupts rather than from shaving a few seconds off routing or canned replies.
Read definition →Support capacity planning is the process of forecasting ticket volume, allocating agent headcount, and scheduling shifts to maintain SLA targets across expected demand patterns. It converts ticket demand into staffing and queue coverage decisions.
Read definition →Support cost per ticket measures the fully loaded cost of resolving one support issue, including agent time, engineering escalations, tooling, and overhead. In B2B support, this metric jumps quickly when every technical ticket needs manual log analysis or product-engineering involvement.
Read definition →Support deflection is the broader practice of reducing agent-handled ticket volume through self-serve content, in-product guidance, automation, or workflow design. It covers knowledge-base deflection, chatbot routing, and other mechanisms that prevent low-value tickets from reaching humans.
Read definition →Support escalation rate is the percentage of tickets that get escalated from frontline support to engineering, senior agents, or management. High escalation rates indicate that L1 agents lack the tools or data to resolve technical issues.
Read definition →Support quality score is an internal measure of how accurate, complete, and policy-compliant support responses are. For technical teams, quality means the answer was not only polite, but actually correct and backed by evidence from the customer’s systems.
Read definition →Support tier definition is the written specification of each support tier's scope, response obligations, access permissions, and escalation criteria. It is the operating rulebook that makes a tiered support model work consistently.
Read definition →Ticket abandonment rate is the percentage of customers who submit a ticket or start a support interaction but withdraw it without resolution, either by closing the chat, deleting the ticket, or stopping responses. It measures unresolved demand that disappears from normal closure metrics.
Read definition →Ticket aging measures the time elapsed since a ticket was created without full resolution. B2B teams usually track aging in hours for P1 and P2 tickets and in days for lower-priority queues.
Read definition →Ticket backlog is the number of unresolved tickets waiting in the queue beyond the team’s normal operating window. In B2B support, backlog is especially dangerous when high-value customers with technical incidents are sitting behind slower investigative work.
Read definition →Ticket deflection rate measures the percentage of support requests that never become agent-handled tickets because customers solved the issue through self-serve help, product cues, or automation. For B2B teams, deflection is healthy only when it removes low-value questions without hiding unresolved technical incidents.
Read definition →Ticket deflection vs. investigation describes the operating choice between keeping a customer out of the support queue and helping a support team diagnose the issue after the ticket exists. Deflection works best for repeated questions with stable answers, such as setup steps or pricing rules. Investigation matters when the issue depends on live account state, recent deploys, API logs, or billing events. Confusing the two leads teams to build chatbot loops for problems that actually require system evidence and technical diagnosis.
Read definition →Ticket investigation is the process of diagnosing the root cause of a customer-reported issue by querying multiple internal systems - application logs, databases, billing platforms, bug trackers, and monitoring tools.
Read definition →Ticket reopen rate is the percentage of resolved tickets that are reopened by the customer within a defined window, usually 7 or 30 days, because the issue was not fully resolved. It measures how often a “closed” ticket was only temporarily quiet.
Read definition →Ticket tagging is the practice of applying structured labels to tickets so they can be categorized by topic, product area, root cause, and resolution type. Consistent tagging is what turns ticket history into usable operational data.
Read definition →Ticket triage is the front-end decision process that classifies a support request by severity, product area, customer impact, and routing destination. It answers questions such as whether the issue is an outage, a known bug, a billing state problem, or a misuse case. Good triage happens before deep diagnosis and creates clean queues, correct SLA priorities, and the right first owner. Bad triage sends urgent technical issues into slow queues or hands simple requests to expensive specialists.
Read definition →A tiered support model organizes support teams into levels: L1 (frontline agents handling initial contact), L2 (senior agents with deeper technical knowledge), and L3 (engineering teams handling the most complex issues). Each tier has defined scope and escalation paths to the tier above.
Read definition →Time to First Response measures how long it takes a customer to receive the first human reply after opening a ticket. It is closely related to first response time, but teams often use this wording in dashboards, customer reports, and executive reviews.
Read definition →Transfer rate is the percentage of tickets transferred from the initial handling agent to another agent, team, or escalation tier before resolution. It measures how often work changes hands before the customer gets an answer.
Read definition →A webhook is an HTTP callback triggered automatically when a system event occurs, such as a failed payment, closed ticket, or completed deployment. Instead of polling an API on a schedule, the receiving service gets a real-time POST request containing event metadata and identifiers. In support and operations systems, webhooks are used to trigger downstream actions like creating Linear issues, enriching Zendesk tickets, or starting investigation workflows the moment a customer-facing event happens.
Read definition →Wrap-up time is the time an agent spends completing documentation, CRM updates, and internal notes after a ticket is resolved and before beginning the next ticket. It is the immediate post-resolution portion of support work.
Read definition →