Customer service glossary

28 customer service terms defined in plain English, from CSAT and first response time to containment rate, RAG and MCP. Each entry opens with a one-sentence definition, gives the formula or a worked example where one exists, and links to the guide that goes deeper.

By , pen name of the Asks founderUpdated 28 terms

New to support metrics? Start with our customer service metrics guide, which shows how CSAT, response time and resolution fit together, then use the entries here to check a definition or a formula.

Browse by topic

A

  • After-hours support

    How customers get help outside business hours, from away messages to an AI agent.

  • AI agent

    Software that answers customers in natural language and takes actions to resolve their request.

  • AI hallucination

    A confident but false or invented answer from an AI model, like a made-up return policy.

  • AI resolution rate

    The share of conversations an AI agent resolves on its own; the definition of "resolved" decides it.

  • Average handle time (AHT)

    Average agent time per interaction: talk or chat time, hold time and follow-up work.

C

  • Canned response

    A saved reply agents insert with a shortcut, with placeholders for the customer's details.

  • Containment rate

    The share of bot or AI conversations that never reach a human agent; not the same as resolved.

  • Conversational commerce

    Selling through chat: product advice, carts, checkout links and order questions in one thread.

  • CSAT (customer satisfaction score)

    How satisfied customers were with one interaction, usually the share of 4 and 5 ratings on a 1–5 scale.

  • CSAT vs NPS

    CSAT rates one interaction; NPS measures loyalty to the company. Which to use, and when.

  • Customer effort score (CES)

    How easy it was to get help, usually agreement with "they made it easy" on a 1–7 scale.

D

  • Deflection rate

    The share of questions answered by self-service before they become tickets, and its blind spot.

E

F

K

  • Knowledge base

    An organized library of help articles, policies and FAQs used by customers, agents and AI.

L

  • Live chat

    A real-time chat window on a website or app, answered by people, an AI agent or both.

M

  • Macro

    One click that inserts a reply and updates the ticket: status, tags, priority or assignee.

  • Model Context Protocol (MCP)

    An open standard that connects AI assistants and agents to outside tools and data.

N

  • Net Promoter Score (NPS)

    Loyalty on a 0–10 "would you recommend us?" scale: % promoters (9–10) minus % detractors (0–6).

O

  • Omnichannel support

    Every channel feeds one inbox and one customer history, so context follows the customer.

P

  • Proactive support

    Reaching customers before they reach you: delay notices, outage alerts, reminders and tips.

R

S

  • Shared inbox

    One team queue, like support@ or a WhatsApp number, where conversations are assigned and tracked.

  • SLA (service level agreement)

    A promised response or resolution time, such as a first reply within 1 hour, and how often you meet it.

T

  • Ticket backlog

    Open requests not yet resolved, usually counted past an age threshold such as 24 hours.

W

FAQ

Questions, answered.

01What's the difference between CSAT, NPS and CES?

CSAT asks how satisfied a customer was with one interaction, NPS asks how likely they are to recommend you, and CES asks how easy it was to get help. Support teams usually run CSAT after conversations and leave NPS to company-wide surveys.

02Which customer service metrics should a small team track first?

First response time, CSAT and the topics customers contact you about. Add AI resolution rate and escalation reasons once an AI agent answers, and resolution time once you set SLAs.

03Are these definitions specific to Asks?

No. They describe how the terms are used across customer support. Where Asks handles something in a particular way, the entry says so in a separate "How Asks handles this" section.

04Who writes these definitions?

Ray, the pen name of the Asks founder, with AI-assisted drafting. Product facts are checked against Asks, and any outside fact links to its source with the date it was checked. Our editorial policy explains how we write and correct pages.

Every entry is written on its own, not generated from a template, and checked against the product and its sources; how we write and correct pages is in our editorial policy. Spotted a definition that's wrong or missing? Email hello@asks.app.

Fewer tickets to measure, faster answers to the rest.

Asks learns your website and help center, answers customers on chat, email and messaging apps, and hands the rest to your team in one inbox.

7-day free trial · card required · cancel anytime

Chat on WhatsApp