- Call auditing
- Reviewing customer calls against a defined set of criteria, such as script adherence, required disclosures, tone and resolution, to check quality and compliance. Traditionally done by listening to a small sample of calls; AI makes it practical to evaluate every call.
- Quality assurance (QA)
- The ongoing process of measuring customer conversations against agreed standards and feeding the results back into coaching, scripts and processes.
- QA sampling
- Reviewing only a subset of conversations, often picked at random, and extrapolating the results to the rest. It is cheap to start, but problems in the conversations nobody reviewed stay invisible.
- Evaluation questionnaire (QA scorecard)
- The list of questions a reviewer, human or AI, answers about each conversation. In Drellia a questionnaire can mix yes/no, multiple-choice and open questions, and each question can define which answers count as compliant.
- Compliant answer
- The answer to a questionnaire question that meets the standard, for example "Yes" to "Did the agent say the call is recorded?". A question that does not apply to a given conversation can be left without a compliance result instead of counting as a failure.
- Evaluation
- The result of applying one questionnaire to one conversation: an answer to each question, ideally with the reasoning behind it and a supporting quote from the transcript, plus an overall score.
- Compliance alert
- A notification raised when an answer signals a risk, such as a missing mandatory disclosure, so that a supervisor can act on it instead of discovering it weeks later.
- Mandatory disclosure
- Information an agent is required to give during a conversation, for example that the call is recorded, who is calling and why, or the terms of an offer. Which disclosures apply depends on the sector and the jurisdiction.
- Speech analytics
- Technology that transcribes recorded calls and analyzes the result to find topics, keywords, sentiment and patterns across many conversations.
- Transcription (speech-to-text)
- Converting call audio into text, usually separated by speaker. Accurate transcripts are the foundation of any automated evaluation.
- Conversation intelligence
- Using AI to turn conversations such as calls, chats and messages into structured information, like evaluations, trends, risks and coaching opportunities, instead of just storing recordings.
- AI voice agent
- Software that holds a spoken phone conversation with a person in natural language, for example to send a reminder, qualify a lead, schedule an appointment or follow up on a payment, and records the outcome of the call.
- Outbound campaign
- A planned series of calls to a list of contacts, run within set calling hours and retry rules, where every call ends with a recorded outcome.
- Call outcome
- The standardized result of a call, for example reached, callback requested, promise to pay or no answer. Consistent outcomes are what make results comparable across agents and campaigns.
- SIP trunk
- A connection that carries phone calls over the internet between a telephone carrier and a system such as a contact center platform or a voice agent. Connecting over SIP usually lets an organization keep its existing numbers and carrier.
- Agent coaching
- Targeted feedback to an agent based on specific conversations. When every conversation is evaluated, coaching can focus on recurring patterns backed by real examples rather than on one call picked at random.
- QA calibration
- Periodically comparing how different reviewers, or a reviewer and an AI, score the same conversations, then clarifying the criteria until they agree. It should happen before any score is used to make decisions.
- Average handle time (AHT)
- The average duration of a customer interaction, including talk time, hold time and after-call work.
- First contact resolution (FCR)
- The share of customer issues resolved in the first interaction, without the customer having to call or write again.
Want to run your questionnaire on every conversation?
Drellia evaluates calls and chats against your own questionnaires. Tell us about your team and we will show you how it works with your questions.