Single event check

The Single Event Check is a way for QA to look at a conversation or interaction and determine whether a specific event occurred - just one, not multiple. Think of it as asking, “Did this one particular thing happen during the call or chat?”

To perform this check accurately, the system needs to understand:

• What is the event? This could be something like a customer asking to cancel, an agent providing a refund, or a mention of a specific product.
• How might it be phrased? Customers and agents may word things differently. So we need to provide the QA with example language to recognise - like “I want to cancel” or “Your refund has been processed.”
• What does not count? Some statements might sound like the event but don’t actually meet the criteria. These are called “anti-events.” For example, “I thought about cancelling but changed my mind” shouldn't be counted as a cancellation event.
   

Multiple event check

A Multi-Event Check looks at a single interaction and asks: Did all of the specified events happen?

This is different from checking for just one event - here, we’re looking for a combination of things to occur during the same conversation.

Multi-event checks are powerful for verifying quality standards - especially when several actions must take place to meet a goal. This gives you deeper insight into agent performance and helps automate processes like coaching alerts or customer follow-ups.
 

Conditional or order-sensitive events

Conditional checks are ideal for evaluating how agents respond to specific scenarios. They help ensure that once a customer mentions something important, the appropriate next step is always taken.

This leads to better compliance, customer outcomes, and automation of follow-up actions - all without needing manual review.

This check helps our QA understand contextual logic within conversations. It’s not just about whether something was said - it’s about when and why it was said.

For example:

• If the customer says they’re cancelling → the agent should explain the cancellation process.
• If the customer doesn’t mention cancelling → we don’t need to check anything further.
   

Weighted / Risk-based evaluation

A Weighted Event Check helps assess if enough meaningful actions took place in a conversation to meet a goal, or if the level of risk crossed a threshold.

Not all events are equal, some matter more than others. This check gives each event a weight, and if the total score meets or exceeds a set threshold, it’s a pass.

Each event is given a percentage weight based on how important it is. QA reviews the interaction and adds up the weights of all the events that occurred. Then it compares the total score to a threshold (like 75pts) to determine:

• Pass – enough positive actions occurred
• Fail – not enough key actions occurred
   

Context dependent assessment (outcome-based check) 

An Outcome-Based Check helps us evaluate whether the overall behavior or result of an interaction aligns with a desired outcome, even if there’s no single action or phrase that proves it.

This type of check relies on context, judgment, and broader behavioral cues, rather than exact keywords or events.

Instead of looking for one or more fixed events, this check looks at the general direction or tone of the interaction. QA determines if the behavior observed is “on track” with the intended outcome.

Examples of outcomes could be:

• The customer feels reassured
• The issue is fully resolved
• The agent shows ownership and empathy
• There are multiple valid ways to reach the outcome, and this check allows for that flexibility.

Outcome-based checks are essential for situations where flexibility, judgment, and a human-like understanding of intent matter most. They’re especially useful when you care about how something was handled - not just what was done.