Quality assurance

Lexey checks the quality of every reply your support agent gives customers.

How it works

Each conversation is assessed shortly after its latest replies, usually within a minute or two, and again as it continues. Until then it shows QA pending. Each reply gets a QA score from 1 to 5 and may be flagged with a finding:

Poor response — The reply was unhelpful or wrong.
Missing knowledge — The agent didn't have the information to answer.
Tone mismatch — The reply didn't match your tone.
Missed escalation — The agent should have escalated and didn't.
Hallucination — The agent said something your knowledge base doesn't support.

A conversation's QA score is the average of its replies' scores. Preview conversations with your live agent are assessed too, marked Test, but left out of your performance figures. Shadow conversations and Preview against a draft aren't assessed.

Where you see it

Home shows your QA trend, and lists what needs your attention. Each item shows its severity, how many conversations it affected and why it was raised, with a Fix with Lexey button. The same items are Lexey's suggested next steps, and an item shows In progress while you work on it with Lexey. To clear an item that isn't worth fixing, select × beside it; Undo on the notice brings it back.
Conversations tags each conversation with its findings, and Filter narrows the list to a finding, a low QA score or negative feedback. Each message in a conversation shows its QA score, and each finding sits next to its message with Fix with Lexey.
Knowledge → Diagnostics groups missing-knowledge findings into topics, so you can see which articles to write.
Lexey's suggested next steps on Home, Conversations and Knowledge start from your findings.

Quality assurance

Lexey checks the quality of every reply your support agent gives customers.

How it works

Poor response — The reply was unhelpful or wrong.
Missing knowledge — The agent didn't have the information to answer.
Tone mismatch — The reply didn't match your tone.
Missed escalation — The agent should have escalated and didn't.
Hallucination — The agent said something your knowledge base doesn't support.

Where you see it

Home shows your QA trend, and lists what needs your attention. Each item shows its severity, how many conversations it affected and why it was raised, with a Fix with Lexey button. The same items are Lexey's suggested next steps, and an item shows In progress while you work on it with Lexey. To clear an item that isn't worth fixing, select × beside it; Undo on the notice brings it back.
Conversations tags each conversation with its findings, and Filter narrows the list to a finding, a low QA score or negative feedback. Each message in a conversation shows its QA score, and each finding sits next to its message with Fix with Lexey.
Knowledge → Diagnostics groups missing-knowledge findings into topics, so you can see which articles to write.
Lexey's suggested next steps on Home, Conversations and Knowledge start from your findings.