# Improve live chat support using my chat data

A metric set, weekly review routine, and improvement plan for a team running live chat on a website or online store.

- Topic: Customer Support
- Framework: Few-Shot
- Level: Standard
- Tags: Meetings, Customer Support, Fill-in Template
- Source: https://www.thinkwithprompt.com/prompts/improve-live-chat-support-using-my-chat-data

## Prompt

```text
Help me use our live chat data to improve support quality and sales.

**Business:** [TYPE and size - e.g. "online store, 8 chat agents"]
**Chat tool and data available:** [PLATFORM, exports, dashboards]
**Current numbers, if known:** [chat volume, first response time, resolution time, concurrent chats per agent, CSAT, missed chats, chat-to-order conversion]
**Hours of coverage:** [HOURS, time zones]
**Bots or automation in use:** [yes / no, what they handle]
**Biggest problem now:** [e.g. "slow replies at peak", "low CSAT on returns", "chats don't convert"]

Give me:
1. A metric set organized into four groups (speed, quality, efficiency, revenue), with a definition for each metric and the healthy direction. Note where definitions vary by platform.
2. Analyses to run: volume by hour vs. staffing, contact reasons from tags or transcripts, CSAT by reason and agent, missed and abandoned chats, chats that led to orders vs. those that didn't, and handoffs from bot to human.
3. How to tag chats so reasons are reliable (a starter tag list of 10-15 tags for my business).
4. Likely improvements based on my problem: staffing and concurrency limits, canned responses, proactive chat triggers on key pages, bot scope, and knowledge-base fixes.
5. A weekly review agenda (30 minutes) and a one-page dashboard layout.
6. A transcript QA sample method: how many chats to review per agent and what to look for.

Use only numbers I provide; put [ADD DATA] where needed. If analyzing transcripts, remind me to remove customer personal data first.
```

## What to fill in

- TYPE and size - e.g. "online store, 8 chat agents"
- PLATFORM, exports, dashboards
- HOURS, time zones
- yes / no, what they handle
- ADD DATA

Free to copy and use. From ThinkWithPrompt (https://www.thinkwithprompt.com/), a library of 2,663 curated AI prompts.