Glossary
Contact center QA glossary
Plain definitions of the terms used in contact center and BPO quality assurance. The definitions are the general industry meanings. Where a line describes how Merivex handles a term, it is labelled as such.
Terms
Quality assurance terms, A to Z
- Agent
- A person who handles conversations with the people contacting a contact center, by phone, chat, email or messaging. Also called an advisor or representative.
- In Merivex: Merivex evaluates the work of human agents. Its own AI agents, the specialist software components that perform the analysis, are a separate thing.
- AHT (average handle time)
- The average time an agent spends on a contact, usually talk or chat time plus hold time plus after-contact work.
- Auto-fail
- A criterion serious enough, usually a compliance breach, that failing it sets the whole evaluation to fail regardless of the other scores.
- Calibration
- Having several evaluators score the same conversations and comparing results, so the scorecard is applied the same way by everyone.
- In Merivex: Reviewer scores are imported as CSV and compared with AI scores, overall and per criterion.
- Call transcription
- Turning a recorded phone call into text, so it can be read, searched and scored like a chat or an email.
- In Merivex: Upload MP3, WAV, M4A, OGG, FLAC or WebM recordings, up to 50 at once, 50 MB and 2 hours per call. Each is transcribed and evaluated automatically; audio is kept 90 days.
- CES (customer effort score)
- A survey measure of how easy it was for a person to get their issue handled.
- Coaching plan
- A set of targeted development steps for an agent, usually built from QA findings and followed up in coaching sessions.
- In Merivex: Plans are generated from patterns across evaluated conversations, with trackable steps and scheduled sessions.
- Compliance monitoring
- Checking conversations for required disclosures, identity checks and prohibited statements set by regulation or internal policy.
- Criterion
- One item on a QA scorecard, describing a behaviour or outcome to be scored.
- CSAT (customer satisfaction score)
- A survey measure of satisfaction with an interaction, often reported as the share of positive responses.
- Dispute
- A formal challenge by an agent or team lead to a QA score, resolved by a reviewer or QA lead.
- Evaluation
- The scoring of one conversation against a scorecard, with a score per criterion and an overall score.
- In Merivex: Each criterion score carries a rationale and a citation to the transcript turn it rests on.
- FCR (first contact resolution)
- The share of issues resolved in the first contact, without the person having to get in touch again.
- Incremental sync
- Importing only the records that are new or changed since the previous import, tracked with a cursor such as the last update time.
- In Merivex: Every pull connector (Zendesk, Salesforce Service Cloud, HubSpot, Intercom, Genesys Cloud, Talkdesk, Five9, Aircall, Twilio) syncs incrementally after the first import.
- Intent
- The reason a person made contact, such as a billing question or a cancellation request.
- In Merivex: Conversations are classified by intent automatically.
- Inter-rater agreement
- How closely different evaluators score the same conversations. Calibration exists to raise it.
- Interaction
- One conversation between a contact center and the person contacting it, in any channel.
- Interaction mapping
- Attaching the who and what to a conversation: the customer, the agent who handled it, the topic and the contact reason.
- In Merivex: For call recordings, the Call Transcription Agent fills these in. It never overwrites what you entered and never creates people.
- NPS (net promoter score)
- A survey measure of how likely a person is to recommend a company, reported as promoters minus detractors.
- QA (quality assurance)
- The process of evaluating conversations against a standard and using the findings to improve performance.
- QA scorecard
- The weighted list of criteria used to evaluate conversations. Also called a QA form or evaluation form.
- In Merivex: A workspace has one active, versioned rubric; earlier scores keep the version they were made under.
- Sampling
- Reviewing a subset of conversations, often a few per agent per month, and treating the results as representative of the rest.
- Sentiment analysis
- Estimating the emotional tone of a conversation, from its words, its audio or both.
- In Merivex: Sentiment is derived from text only. There is no acoustic analysis.
- Speech-to-text
- Software that converts spoken audio into written text. Accuracy depends on audio quality, accents, crosstalk and vocabulary.
- In Merivex: Merivex uses Whisper-based speech recognition with an automatic fallback, with English and French language hints.
- Transcript
- The text of a conversation. For calls, produced by transcribing the audio.
- In Merivex: Merivex transcribes the call recordings you upload, and also evaluates transcripts imported from Genesys Cloud, Talkdesk, Aircall, Twilio or a file.
- Webhook (HMAC-signed)
- A URL that receives data pushed from another system. An HMAC signature lets the receiver check that each request came from the expected sender and was not altered.
- In Merivex: Merivex accepts conversations on a signed webhook on every plan, including from n8n workflows.
- Workspace brief
- A written description of an operation's business, standards and rules, given to software as context so its judgments reflect that operation rather than generic norms.
- In Merivex: In Merivex, every AI agent, the evaluation and the coaching read the workspace brief. It is context, not instructions: it cannot change permissions or safety rules.
Questions
About these terms
What is the difference between QA and calibration?
QA is evaluating conversations against a scorecard. Calibration is checking that the evaluators, people or software, apply that scorecard the same way.
Are CSAT and QA scores the same thing?
No. CSAT comes from the person who made contact, through a survey. A QA score comes from an evaluator applying a scorecard to the conversation. They often disagree, and both are useful.
When this glossary says agent, does it mean an AI agent?
No. Here, an agent is a person working in a contact center. Merivex also has AI agents, the software components that analyze conversations; where they are meant, they are called AI agents.
Keep reading
See the terms in practice
Run an evaluation on a sample conversation and see criteria, citations and review in one place.