TL;DR

The one-question rating shown when a chat ends. Short surveys get answered; long ones get closed, which is why almost every implementation is a single tap.

Key Points

✓

The survey is the delivery mechanism; customer satisfaction score (CSAT), customer effort score (CES), and net promoter score (NPS) are the metrics it can carry ^[1].

✓

Response rates fall sharply with length — a single-tap rating typically outperforms a multi-question form by a wide margin.

✓

Timing matters: shown at the moment the chat closes, while the experience is fresh, rather than emailed hours later.

✓

An optional free-text box after the rating is where the actionable detail lives, even though most people skip it.

✓

Ratings are self-selecting — unhappy and delighted customers answer more often than satisfied ones, so the average overstates both tails.

✓

Pairing survey results with a Conversation Tag turns a score into a diagnosis: which topics produce low ratings, not just how many.

What to ask

Most teams ask a satisfaction question — 'How would you rate this conversation?' — on a two-point (thumbs), three-point, or five-point scale, and compute customer satisfaction score (CSAT) as the share of positive responses ^[1]. The alternative worth considering is a customer effort score (CES) question: 'How easy was it to get your issue handled?' Effort correlates better with future loyalty than satisfaction does, because customers rarely churn over a pleasant conversation that solved nothing. Net promoter score (NPS) is a poor fit here — it asks about the company as a whole, and a support interaction is the wrong moment to sample it. Whichever you choose, ask one question, keep the scale consistent over time so trends mean something, and put the optional comment box after the rating rather than before it.

Reading the results honestly

A post-chat survey measures the customers who chose to answer, not the ones you served. Silence is the majority case and it is genuinely ambiguous — a visitor who got a fast answer and closed the tab looks identical to one who gave up. That makes the absolute number less useful than the trend and the breakdown. Slice by topic using a Conversation Tag, by operator, and by hour of day; a score that drops only outside Business Hours is telling you about coverage, not quality. It is also worth reading low ratings alongside the Chat Transcript, because the rating says something went wrong and the transcript says what. Where volume is thin, Sentiment Analysis over every conversation gives a broader if noisier signal than a survey answered by a fraction of visitors.

Implementation details that matter

Show the survey when the conversation genuinely ends, not when the widget closes — a visitor who minimises the chat and returns has not finished. Keep it inside the widget rather than redirecting anywhere. Do not gate it behind a login or ask for an email, since both convert an easy tap into a decision. Handle the Offline Message case separately: rating a conversation that never got a reply measures availability, and mixing that into the same score muddies both. Finally, close the loop internally — a survey nobody reads is worse than none, because it spends customer goodwill for nothing. The useful cadence for a small team is a weekly scan of every low rating with its transcript, which is a handful of minutes and consistently surfaces the First Contact Resolution (FCR) failures worth fixing.

Sources & References

Measuring customer satisfaction - Nielsen Norman Group ↗

Stop Trying to Delight Your Customers - Harvard Business Review ↗

Last updated: August 12, 2026

Related Terms

Chat Transcript

A chat transcript is the complete, timestamped written record of a live-chat conversation between a website visitor and a support operator or chatbot. It captures every message in sequence, usually alongside metadata about the session, and can be archived, exported, or emailed for later reference.

Conversation Tag

A conversation tag is a short label attached to a support conversation (or an individual message within it) that classifies its topic, intent, or status so the team can search, filter, route, and report on it later.

Sentiment Analysis

Sentiment analysis is the automatic classification of text by the emotion it expresses — typically positive, negative, or neutral. In customer support it is applied to chat messages and transcripts to flag frustrated customers, route conversations, and measure how people feel about a product without waiting for a survey.

First Contact Resolution (FCR)

First Contact Resolution (FCR) is the percentage of customer issues fully resolved in a single interaction, with no transfer, callback, or follow-up needed. It measures how often a support team solves a problem the first time the customer reaches out.

Resolution Time

Resolution time is the elapsed time from when a customer first raises an issue until that issue is fully resolved and the conversation or ticket is closed. It is also called time to resolution (TTR) or mean time to resolution (MTTR) when averaged across many cases.

Put live chat on your site in 30 seconds

Eloqra gives you a lightweight live chat widget that routes every visitor message to your Telegram — free forever for one project.

Get Started Free

Post-Chat Survey