Customer support glossary

Abandonment rate is the share of customers who leave a queue, chat, or callback line before ever reaching an agent. Contact centres track it alongside wait time because abandonment climbs sharply once queues pass a customer’s patience threshold, making it an early signal that staffing or deflection is not keeping pace with demand.

An agent scorecard rates a single customer interaction against fixed criteria. Categories typically include greeting and identity verification, accuracy of the resolution, compliance adherence, tone, and completion of required process steps. Each category carries a weight. Category scores combine into one percentage that feeds coaching sessions and QA reporting, and often rolls up into a team-level customer satisfaction index. Reviewed manually, scorecards cover only a small sample of conversations. OnClarity’s AI Agent QA applies a team’s existing scorecard rubric to every conversation across voice, chat, email, and WhatsApp.

Average handle time is the mean time an agent spends on a single contact, including talk or chat time and after-contact work. Because the definition varies between teams, comparisons are only meaningful when both sides confirm whether wrap-up time is included inside the calculation being reported.

Average speed of answer is the average time a contact waits in queue before an agent connects, timed from queue entry to answer and typically excluding IVR navigation. It differs from average wait time, which may include abandoned calls, and from service level, which reports the percentage answered within a target such as the common 80/20 standard. Because ASA is a mean, it can look acceptable while masking a long tail of extreme waits, so pair it with wait-time distribution or service level percentiles for a fuller picture. Example: 500 calls with a combined wait of 4,000 minutes yields an ASA of 8 minutes.

Backlog is the set of unresolved tickets carried past their target response window. A growing backlog signals that arrival rate has outpaced capacity, and because ageing tickets attract follow-up contacts, backlog tends to generate additional volume rather than simply waiting quietly to be cleared later.

Business rules are the conditions that decide how a contact is prioritised, routed, or escalated. Typical rules combine customer tier, channel, detected intent, and elapsed time to determine where a conversation goes and how quickly it must be answered under the applicable service level agreement.

Call barging happens when a supervisor joins a live customer call as an active participant. Both the agent and the customer can hear the supervisor once they join. This differs from listen-only monitoring, where the supervisor stays silent, and from whisper coaching, where only the agent hears the supervisor. In AI-assisted centers, barging can also mean a human stepping into a call that an AI voice agent is handling. For example, a supervisor sees a compliance phrase flagged in real time and joins the call to resolve the issue directly with the customer. Because barging grants access to a live conversation, permissions and logging matter. Teams should record who can barge, when, and why. OnClarity’s AI Agent QA can export QA scoring as part of a full audit trail, which matters for regulated contact centers that need to show reviewers exactly who touched a given call.

Call monitoring is the practice of reviewing live or recorded customer calls to check quality, compliance, and outcomes. Supervisors can listen in real time, or they can review recordings after the call ends. Traditionally, teams sampled only 5 to 10 percent of calls by hand. That left most conversations unreviewed. AI quality tools change this math. OnClarity’s AI Agent QA scores 100 percent of conversations against a company’s existing evaluation rubric within minutes of call end, instead of waiting days for a supervisor to catch up on a backlog. Recording consent rules vary by US state, and some states require consent from every party on the call, not just one. Recordings are commonly stripped of sensitive data through PII redaction before they are stored, so a reviewer can read a transcript without seeing a customer’s card number or social security number.

Code switching happens when a speaker alternates between two or more languages or dialects in one conversation. This can happen between sentences (inter-sentential) or inside a single sentence (intra-sentential). It includes Arabizi, Arabic commonly written using the Latin alphabet in digital messaging. In contact centers, this is routine on WhatsApp and chat across the GCC. A message may open in Gulf Arabic and close with an English product or brand name. Models trained on only one language often misread this mixing. The result is misclassified sentiment and a swollen “unknown topic” bucket. OnClarity’s Arabic-native capability and Agent Assist’s multilingual response drafting are built to handle this blending directly, rather than treating mixed-language messages as noise.

A contact centre is the team and tooling that handle inbound customer contacts across every channel, including voice, chat, email, and social. The term replaced call centre as support expanded beyond telephony, and it covers workforce management, quality assurance, and routing as well as the agents themselves.

CCaaS stands for Contact Center as a Service. It is cloud-delivered software for managing customer interactions, and it covers call routing, IVR, omnichannel queues, recording, and reporting. A company buys CCaaS by subscription instead of building and running its own telephony hardware. CCaaS differs from UCaaS, which handles internal employee collaboration like chat and video meetings. It also differs from on-premise platforms, where a company owns and maintains its own infrastructure. Example: a bank migrates its voice and chat queues to a CCaaS platform, then connects AI agents through the vendor’s APIs to automate routine inquiries like balance checks and card replacements. AI vendors typically integrate on top of or alongside CCaaS rather than replacing it. This is one reason CCaaS sits near the center of most contact center AI terms: almost every AI deployment in contact centers has to plug into one.

Cost per contact is total fully loaded contact center operating expense divided by the number of contacts handled in the same period. The figure conventionally includes agent wages and benefits, supervision and QA staffing, telecom, technology and software licensing, and facilities overhead. Contact centers are labor-intensive, so personnel costs make up the bulk of the total and wage rates by geography drive much of the variance between centers. Example: a center spending $420,000 a month to handle 60,000 contacts has a cost per contact of $7.00. MetricNet benchmarking data puts average cost per inbound contact around $4.83, with a median near $3.73.

Data residency is the requirement that customer data be stored and processed inside a specific country or region. Regulated industries and public-sector buyers frequently mandate it, which makes deployment location a procurement question rather than a technical detail for any support platform handling personal data.

Data sovereignty is the principle that data is subject to the laws of the country where it is stored or processed. A foreign government may be able to compel access to that data regardless of who owns it. Data sovereignty differs from data residency, which only addresses where data physically sits. It also differs from data localization, which is a legal mandate requiring data to stay within a country’s borders. Example: a Saudi bank may require in-country processing to meet PDPL expectations. Storage location alone would not resolve which jurisdiction’s laws actually govern that data, so residency and sovereignty have to be considered together. OnClarity operates under SOC 2 Type II controls. OnClarity is also GDPR and HIPAA Ready, and it aligns with Saudi PDPL. OnClarity offers in-country data residency options for enterprises that need processing to stay within a specific legal jurisdiction.

Deflection is steering a contact toward a self-service or automated resolution before it reaches an agent queue. Effective deflection resolves the underlying question; unsuccessful deflection merely delays the contact, so teams measure it by whether the customer returns rather than by how many contacts were diverted.

A disposition code is the label an agent or AI applies to a closed contact to describe what actually happened. Disposition data drives volume reporting and root-cause analysis, so inconsistent or overly broad codes quietly destroy a support team’s ability to explain why contacts arrive at all.

Escalation is moving a contact to a more senior agent, a specialist team, or a human when automation stalls. Escalation paths are defined by business rules, and the rate at which contacts escalate is a direct measure of whether the first line is equipped to resolve them.

Escalation rate is the share of contacts transferred to a higher support tier, a supervisor, or from an AI agent to a human. Reporting should separate planned handoffs, where an AI agent routes a complex or high-risk case by design, from failure escalations, where the agent could not resolve the issue; blending the two obscures true automation performance. Escalation reason codes point to specific knowledge base gaps, and the metric moves inversely with containment and first contact resolution. Example: 120 escalations from 3,000 contacts gives a 4% rate.

First contact resolution is the share of issues fully resolved during a customer’s first interaction, with no follow-up contact about the same problem. It correlates strongly with satisfaction and cost, but only when measured over a defined reopen window rather than at ticket close.

Full resolution time is the total elapsed time from a customer’s first message to a confirmed resolution, including waiting, escalation, and overnight gaps. Unlike response time, it reflects the experience of waiting for an answer rather than the speed of the first reply alone.

Hold time is the duration a customer spends on hold after an agent has answered the call, typically while the agent researches an account, consults a colleague, or prepares a transfer. It is distinct from queue wait time, which happens before the call connects to an agent. Hold time is one part of average handle time, alongside talk time and after-call work. Teams usually report it as average hold time per call or total hold time across a contact. Frequent or lengthy holds are associated with lower satisfaction scores, since repeated waiting adds perceived effort even when the issue gets resolved, and repeated holds can drag down a whole customer satisfaction index.

A hybrid support model has AI handling routine, high-volume contacts while human agents own complex, sensitive, or high-value interactions. The design question is not what the technology can handle but where the cost of an error is high enough to justify keeping a person in the conversation.

IVR is an automated telephony system that plays recorded or synthesized prompts and routes callers based on their input. Touch-tone (DTMF) menus require keypad presses. Directed dialogue lets callers speak from a limited list of options. Natural language or conversational IVR interprets open-ended speech to identify intent. Success is typically measured by containment rate: the share of calls resolved without a human agent. Deep menu trees are a common source of customer friction, causing repeat prompts and abandoned calls. OnClarity’s AI agents work across chat and voice, handling voice interactions conversationally.

A knowledge base is the maintained source of truth that AI agents and human agents both answer from. Its accuracy sets a ceiling on the quality of every automated answer, which makes knowledge maintenance an operational requirement rather than a documentation project owned by nobody in particular.

A knowledge gap is a recurring customer question the knowledge base cannot answer, usually visible as repeat escalations or low-confidence fallbacks on the same intent. Systematically closing gaps is the highest-return maintenance activity available to any team running automated support at scale.

Live chat is real-time text support delivered in-product or on-site, fronted by a human agent, an AI agent, or both in sequence. Its defining constraint is synchrony: the customer is waiting, so response latency and handoff quality matter more than in asynchronous channels.

A macro is a saved reply or action set that agents apply to a common case, inserting standard wording and performing routine updates in one step. Macros reduce handle time and variation, but stale macros are a frequent source of incorrect policy reaching customers at scale.

Multi-tenancy is a software architecture where one application instance serves multiple customer organizations, called tenants. Logical controls keep each tenant’s data separate even though the underlying infrastructure is shared. Isolation models vary. Some platforms use a shared database with tenant ID columns distinguishing each customer’s records. Others use separate schemas per tenant. The strongest option is a fully dedicated instance per customer. Shared infrastructure is what keeps cloud contact center pricing low. Dedicated or on-premise deployment is the usual alternative for regulated buyers who need stronger separation between their data and everyone else’s. A procurement team evaluating an AI vendor should ask directly whether conversation data shares storage with other customers, and how encryption keys are separated between tenants. Both answers affect data sovereignty and audit requirements, so they are worth writing into a contract, not just a sales conversation.

Multilingual support means serving customers in their own language, including handling Arabic and English side by side within a single operation. Beyond translation it requires language-aware routing, quality scoring in each language, and knowledge content that is maintained rather than machine-translated once and forgotten.

Occupancy rate is the portion of an agent’s logged-in time spent actively handling contacts rather than waiting for them. Sustained high occupancy predicts burnout and attrition, so workforce planners treat it as a health measure rather than a productivity target to be maximised.

Omnichannel describes one continuous conversation and history across chat, email, voice, and social, so a customer who switches channel does not have to restart. The distinction from multichannel is continuity: offering many channels is not omnichannel unless context follows the customer between them.

On premise means software runs on servers the customer owns or controls, inside their own data center or private cloud, instead of the vendor’s environment. In an AI contact center, on-premise deployment gives regulated operators direct control over where conversation data is processed and stored. This matters when telephony investments, internal audit rules, or sector regulation make external hosting difficult to justify. Example: a utility keeps call recordings inside its own network, while the AI model itself runs in an isolated environment the utility controls directly. The trade-off is real. The customer takes on hardware cost, patching and upgrade cycles, and the engineering staff needed to run the stack. The vendor no longer carries that work. Many enterprises land on a hybrid model instead. They pair on-premise components with private cloud for everything that does not require strict data isolation, which keeps costs lower than a fully on-premise build.

Peak volume is the highest-contact period a staffing plan has to absorb, whether a daily rush hour, a seasonal spike, or an incident-driven surge. Because peaks drive headcount decisions, automation is usually justified on peak absorption rather than on average daily contact volume.

A perpetual license is a one-time software purchase that grants indefinite rights to use a specific version. Updates and support are typically sold separately, under an annual maintenance contract. It differs from subscription or term licensing, where access stops once payments stop. CX buyers weigh a higher upfront cost against predictable long-run spend. They should also check who hosts the platform, and whether newer AI model versions are included or billed separately. Example: a bank with strict KSA/GCC data residency requirements chooses a perpetual, self-hosted deployment to keep customer data within approved borders.

PII redaction is the automatic removal of personally identifiable information from transcripts, logs, and training samples. It lets support conversations be analysed, stored, and used to improve models without carrying customer identifiers into systems that were never scoped to hold them.

QA calibration is a recurring exercise in which multiple evaluators independently score the same customer interaction against the same agent scorecard, then compare results and reconcile any gaps. Sessions typically run on a regular cadence to catch rubric ambiguity before it spreads and to keep evaluators aligned as scorecards evolve. Success gets measured by inter-rater reliability: how closely scores converge across reviewers. Calibration is necessary because manual QA relies on small, sampled reviews, where individual judgment drifts over time and across reviewers. Applying one rubric automatically across every conversation, as OnClarity’s AI Agent QA does, removes evaluator-to-evaluator drift by design.

QA coverage is the percentage of customer service conversations reviewed and scored against a quality rubric. It measures how much of an operation’s actual customer interaction volume actually gets evaluated. Manual QA programs typically sample only a small slice of contacts. Industry data puts the standard at 2 to 5 calls per agent per month, often a low single-digit percentage of total volume. Everything outside that sample is a blind spot, where compliance breaches, coaching opportunities, and recurring issues go undetected. OnClarity’s AI Quality Agent scores 100% of conversations within about five minutes of call end, roughly 20 times typical manual coverage. That shift turns QA from a sampling problem into a question of rubric consistency and audit-trail completeness.

Quality assurance is the structured review of conversations against a scorecard, applied to human agents and AI agents alike. Traditional QA samples a handful of conversations per agent each month; automated QA scores every conversation, which changes the exercise from spot-checking into actual measurement.

A quality assurance score is the rating a customer service conversation receives after evaluation against a defined rubric, typically covering accuracy of information, process and compliance adherence, and communication quality such as tone, empathy and clarity. A simple rubric might weight ten criteria at 10 points each, with a call losing points for a missed disclosure or an inaccurate answer to land at 82/100. The score is only as reliable as the rubric’s clarity and the sample size behind it: manual QA programs conventionally review roughly 5 to 10% of conversations, so most interactions never get scored. OnClarity’s Agent QA scores 100% of conversations within five minutes, cutting QA operations cost by roughly 70%, and keeps scoring criteria visible so teams can calibrate evaluators against each other.

A queue is the ordered set of contacts waiting to be handled, usually segmented by channel, language, skill, or customer tier. Queue design determines who waits and for how long, making it one of the most consequential and least visible decisions in a support operation.

Resolution rate is the share of contacts closed with the customer’s issue actually solved, as distinct from contacts merely closed. The gap between the two is where reopened tickets and repeat contacts live, so the measure is only trustworthy when reopens are tracked against it.

Response time is how long a customer waits for a reply once they have made contact. Teams report it as a first-response figure and as an average across subsequent replies, because a fast opening reply followed by long silences produces a poor experience despite good headline numbers.

Routing is deciding which agent, team, or AI agent receives a contact, based on intent, language, skill, customer tier, and availability. Accurate routing is the cheapest available reduction in handle time, because a misrouted contact pays the cost of being read and reassigned twice.

Schedule adherence is the percentage of scheduled time an agent is available and performing the specific activity planned for that interval, such as taking calls when calls are scheduled or being on break when a break is scheduled. Most contact centers target 85 to 95% adherence, per industry benchmarks, with schedules built to allow for shrinkage from training, meetings, and absences. Example: an agent scheduled for 480 minutes who logs 456 minutes matching the schedule scores 95% adherence.

Self-service is help a customer can complete without contacting support at all, such as a help centre article, an account action, or a guided flow. Its success is measured by task completion rather than page views, since a read article that fails to resolve still generates a contact.

Service level is the percentage of contacts answered within a set time threshold, written as two numbers such as 80/20: 80% of calls answered within 20 seconds. The threshold is a business decision rather than a fixed standard, and while 80/20 is the most commonly cited target, its origin is debated and it lacks rigorous research backing. Staffing to hit a target service level typically relies on the Erlang C queuing formula, and service level moves together with abandonment rate, since more callers hang up as wait times grow. Example: if 850 of 1,000 calls are answered within 20 seconds, service level is 85%.

A service level agreement sets the committed response and resolution targets for a customer, tier, or contract, often with financial penalties attached. Support operations design queues, routing, and escalation around SLA thresholds, which makes the agreement a structural constraint rather than a reporting line.

Shrinkage is the share of paid agent time unavailable for handling contacts, covering breaks, training, meetings, absence, and system downtime. Workforce plans that ignore shrinkage systematically understaff, because scheduled headcount and available headcount differ by a predictable and often substantial margin.

SIP trunking, short for Session Initiation Protocol trunking, carries voice calls over an IP network between a contact center platform and a telephony provider. It replaces physical PRI or analog phone lines with a virtual connection, sized by concurrent call capacity and codec choice. For an AI voice agent, the trunk is where audio enters and leaves the system. Codec quality, jitter, and capacity limits all directly affect how natural the agent sounds and how quickly it responds. Trunk health ties directly to latency, since a congested trunk adds delay before the agent even starts processing audio. Example: provisioning a dedicated test trunk and phone number lets a team run a proof of concept with OnClarity’s AI agents answering live calls, before wider rollout across the rest of the queue.

A ticket is the record of a single customer issue and everything done about it, including messages, internal notes, actions taken, and final disposition. The ticket is the unit most support reporting counts, which is why inconsistent ticket creation quietly distorts every downstream volume metric.

Time to first response is how long a customer waits before any human or AI reply arrives. It is the most visible service measure to the customer and the easiest to improve with automation, which is why it should always be read alongside resolution quality.

Transfer rate is the percentage of contacts moved from the first agent or queue to another agent, team, or queue before resolution. A warm transfer includes context passed directly between agents; a cold transfer drops the customer into a new queue with no handoff, often forcing them to repeat information. Elevated transfer rates usually signal routing misconfiguration, gaps in skill-based mapping, or missing knowledge access rather than individual agent fault. Each transfer represents an unresolved first contact, so transfer rate moves inversely with first contact resolution and inflates average handle time across the interaction chain. Example: 80 transfers out of 1,000 calls yields an 8% transfer rate.

Triage is sorting incoming contacts by intent, urgency, and value before they are worked, so the most consequential conversations reach the right place first. Automated triage now performs the classification step, leaving prioritisation rules and exception handling as the main human contribution.

A unified inbox is one workspace where conversations from every channel land, so agents work a single prioritised list instead of switching between tools. It reduces context-switching cost and makes cross-channel history visible at the moment an agent is drafting a reply.

A warm transfer connects a customer to another agent or team only after the receiving party has been briefed on who the customer is and what they need. The customer never has to repeat themselves. This contrasts with a cold or blind transfer, where the call is passed along with no introduction, and the next agent has to start from zero. Example: in an AI voice deployment, a warm transfer means a mortgage query is handed to a human specialist along with the customer’s verified identity, an intent summary, and the full transcript so far. The trade-off is brief. Two agents are occupied during the handoff instead of one, but that small cost is offset by fewer repeat-explanation complaints from customers. See also routing and triage, two related terms already defined on the OnClarity glossary.

White label describes a product built by one vendor and sold or deployed under another company’s brand. The originating vendor stays unnamed to end users. In CX, this shows up three ways: a BPO presents analytics as its own service, a platform embeds an analytics layer into its product, or a partner resells software under its own name. Buyers should confirm branding scope (UI, emails, domains), tenant data separation, support ownership, and who carries contractual liability. Example: an outsourcer issues auto-QA scorecards and audit-trail exports to a client under its own brand, with the underlying provider unnamed.

Workforce Management, or WFM, is the discipline of forecasting contact volume, calculating staffing requirements, building schedules, and tracking adherence and shrinkage. The goal is simple: have the right number of agents available for predicted demand. WFM is narrower than workforce engagement management, which also covers quality scoring, coaching, and performance management. When an AI agent resolves routine contacts on its own, forecasts must split deflected volume from escalated volume, since escalations tend to run longer and more complex than average. Example: once automation clears most password resets, the remaining queue holds fewer calls, but harder ones. Planners raise their average handle time assumptions, even as overall ticket volume drops, and reset staffing levels instead of reusing last quarter’s model unchanged.

Wrap-up time is the after-contact work an agent performs before taking the next conversation, including notes, disposition coding, and system updates. It is frequently excluded from handle time reporting, which is the most common reason two teams’ average handle time figures cannot be compared.