Analyze
Business impact
Version 1.0 · Last updated 2026-05-28
Deflection without satisfaction is just abandonment with extra steps. This tab shows whether customers are actually happy with the answers the agent gives.
The upcoming CSAT component
The Customer Agent will include a built-in CSAT component that appears at the end of a resolved conversation. Instead of redirecting customers to an external survey tool, the agent asks for a rating inside the same surface where the conversation happened. One tap, no context switch. The rating feeds directly into the satisfaction metrics on this tab.
The component adapts to the Moment and the Personality you configured. A Support conversation might ask “Did we resolve your question?” with a 1-to-5 scale. An Acquisition conversation might ask “Was this helpful?” with a thumbs up or down. The phrasing follows the tone you set in Personality, so the survey feels like part of the conversation, not a bolted-on form.
Key concepts
CSAT score
The average satisfaction rating customers give after a conversation. Shown as a number out of 5 and as a percentage of positive ratings (4 and 5). Always read alongside the response rate.
Response rate
The share of resolved conversations where the customer actually left a rating. A high CSAT on a low response rate is less reliable than a moderate CSAT on a high response rate.
Resolution time
Time from first customer message to conversation resolved. Shown as a median and as a trend over time. Lower is better, but not at the cost of satisfaction. A fast wrong answer is worse than a slower right one.
Sentiment trend
The agent reads the emotional tone of each conversation. The trend shows whether customers are getting more or less frustrated over time. Useful for catching problems before they show up in CSAT, because sentiment moves faster than survey responses.
Baseline comparison
A before-and-after view. Set the date you deployed Unless (or made a major configuration change) and the dashboard shows the delta in CSAT, resolution time, and sentiment between the two periods.
What you can do here
- See the current CSAT score and response rate
- Track resolution time as a median and a trend
- Monitor sentiment trends across Moments
- Compare current performance against a baseline period
- Break down all metrics by Moment
- Click into any metric to see the contributing conversations
When to use it
- Weekly check on customer satisfaction trends
- After a Personality or Procedures change, to measure the impact
- When building the case that the agent is not just deflecting, but resolving well
- Before a quarterly business review, to pair satisfaction numbers with cost and revenue data
When not to use it
- You want to read individual conversations. That is Conversations.
- You want to see cost savings or revenue. Those are Cost reduction and Revenue generation.
How it works
The agent collects ratings at the end of each conversation. Sentiment is analyzed per message during the conversation. Resolution time is logged from first message to resolution. The dashboard aggregates these into the scores and trends you see. Every data point links back to a conversation you can read.
Frequently asked questions
Where does the CSAT score come from?
From the in-conversation rating the customer gives after the agent resolves the question. The agent asks for a rating when the conversation ends. Not every customer rates, so check the response rate alongside the score.
CSAT looks high but I do not trust it. Why?
Low response rates skew CSAT high because satisfied customers are more likely to rate. Cross-check with negative-feedback counts in Conversations and look at the sentiment trend for a fuller picture.
What is the difference between CSAT and sentiment?
CSAT is an explicit rating the customer gives. Sentiment is the agent's reading of the customer's tone during the conversation. CSAT is what the customer says they feel. Sentiment is what the conversation sounds like.
How is resolution time calculated?
From the moment the customer sends the first message to the moment the conversation is marked resolved. Includes wait time if the agent escalates to a human.
Can I compare CSAT before and after deploying Unless?
Yes. Set a baseline date in the settings. The dashboard shows the delta between the baseline period and the current period.
Can I break down satisfaction by Moment?
Yes. The tab shows CSAT, resolution time, and sentiment per Moment. Support and Retention typically have the most data.
What do I do if CSAT drops?
Open Conversations, filter by negative ratings, and read the conversations. Common causes: the agent answered confidently but incorrectly (update Living Knowledge), the agent could not help and did not escalate (adjust de-escalation in Procedures), or the tone was off (adjust Personality).
