Telegram AI Customer Service Team Core KPI Dashboard: Bot Resolution Rate, Human Handoff Rate, and CSAT Metrics Explained

Running a Telegram AI customer service team while only tracking daily message volume is like driving a car while only looking at the odometer without checking the fuel gauge—you know how far you’ve gone, but not whether the engine is healthy. What truly drives customer service efficiency and user satisfaction are the three core metrics: bot resolution rate, human handoff rate, and CSAT score. This article will guide you step by step in building a TG-Staff customer service dashboard, using data to drive team improvements rather than making decisions based on gut feelings.

Why Telegram AI Customer Service Teams Need to Focus on KPI Dashboards?

A common mistake among Telegram bot operators is only tracking daily session counts or message volumes, ignoring whether users are actually being resolved. A user might send 5 messages, and the bot auto-replies 5 times, which looks active, but none of the user’s issues may have been resolved—ultimately, users churn without the team knowing why.

The value of a KPI dashboard lies in:

• Quantifying the bot’s true efficiency: Know how many sessions the bot handled independently versus how many required human intervention.
• Identifying process bottlenecks: High handoff rate? Low CSAT score? Data tells you where the problem lies.
• Optimizing ad attribution: Combined with channel data from split links, you can see which sources generate the most complex user issues and which channels have the highest conversion rates.
• Supporting team decisions: Whether to add agents, optimize bot flows, or adjust handoff scripts—let data decide.

Three Core KPIs Every AI Customer Service Team Must Track

In Telegram customer service scenarios, three metrics form a complete evaluation triangle: efficiency (resolution rate), boundary (handoff rate), and quality (CSAT). None can be omitted.

Bot Resolution Rate — A Hard Metric for Automation Efficiency

Definition: The percentage of sessions independently resolved by the bot out of total sessions.

Formula: Bot Resolution Rate = (Sessions Resolved by Bot ÷ Total Sessions) × 100%

Industry Benchmark: For structured common issues (e.g., order inquiries, FAQs), the target resolution rate should be above 70%; for complex business scenarios (e.g., technical support, complaints), 50%–60% is considered good.

Why It Matters:

• Every 5% increase in resolution rate means you save 5% on human agent costs.
• High resolution rate = users get answers within 10 seconds instead of waiting in line for a human.
• Telegram-specific note: The definition of bot resolution must be clear—a session is considered bot-resolved only if it was not handed off to a human, the user did not repeat the question, and the session ended. TG-Staff’s session records can be filtered by status to help you count accurately.

Human Handoff Rate — Key to Measuring Bot Capability Boundaries

Definition: The proportion of sessions transferred from the bot to a human agent.

Formula: Human Handoff Rate = (Sessions Handed Off to Human ÷ Total Sessions) × 100%

Healthy Range: Generally, 20%–40% is a healthy range. Below 20% may indicate the bot is overextending, leaving users unsatisfied but with no way to reach a human; above 40% suggests the bot’s capabilities are insufficient or the routing rules need adjustment.

Trigger Scenarios in Telegram:

• User sends specific keywords (e.g., “human”, “agent”, “complaint”)
• Bot fails to understand user intent (e.g., multi-turn conversation failure)
• User actively clicks a “Talk to Human” button
• Session routing rules (e.g., under “Online First” mode, automatically hand off when an agent is online)

Interpreting Anomalies:

• Sudden spike in handoff rate → Check for new user influx, bot flow bugs, or misleading split link scripts.
• Handoff rate consistently below 10% → Check if users are truly resolved (combine with CSAT), or if the bot is “pretending to reply” without solving anything.

CSAT Score — Direct Feedback on User Satisfaction

Definition: A user’s rating of the service experience after a session ends, typically on a scale of 1–5.

Collection Method: In Telegram, the bot can send a rating card (e.g., “Please rate this service: 1 star—5 stars”) after the session ends, and users can tap to complete it. TG-Staff’s Professional version user profile feature automatically records CSAT scores for each session and aggregates them into an average score.

Correlation Analysis:

• High resolution rate + High CSAT = Bot working perfectly.
• High resolution rate + Low CSAT = Bot answered, but answer quality was poor or experience was bad (e.g., too robotic, repetitive).
• High handoff rate + High CSAT = Bot accurately identifies complex needs and hands off to human, good user experience (this is positive).
• High handoff rate + Low CSAT = Issues with routing rules or agent capabilities, requiring focused investigation.

How to Build a Telegram Customer Service KPI Dashboard with TG-Staff?

TG-Staff provides a one-stop tool from bot integration to data statistics. Follow these three steps to quickly build your dashboard.

Step 1: Integrate Bot and Configure Session Routing

1. In the TG-Staff Console, click “Add Project” and enter your Telegram Bot Token.
2. Go to Project Settings → Session Routing, and choose a routing rule:
   • Round Robin: Suitable for teams with few agents needing even distribution.
   • Online First: Suitable for peak hours; whoever is online takes the session, reducing user wait time.
3. To track KPIs by channel, create split links (available in Standard plan and above). Split links capture visitor IP, browser information, and URL parameters, which will be associated with session records.

Step 2: Enable User Profiles and Statistics

Professional plan users can enable “User Profiles and Statistics” in project settings. Once enabled:

• After each session, the system automatically records the bot’s handling result (Bot Resolved / Handed Off / Unresolved).
• Agent assignment records, first response time, session duration, and other basic data are immediately available.
• CSAT rating entry: Add a rating card in the bot’s reply at the end of the session (configurable in TG-Staff’s visual command flow).

Step 3: Export or View Core KPI Data

In the TG-Staff Console’s “Statistics” module, you can view:

Metric	View Location	Notes
Total Sessions	Statistics → Overview	Filter by time range
Bot-Resolved Sessions	Statistics → Session Records → Filter by “Bot Resolved”	Need to confirm bot resolution definition
Handed-Off Sessions	Statistics → Session Records → Filter by “Handed Off”	Includes auto and manual handoffs
Average First Response Time	Statistics → Agent Performance	First message response time for human agents
Average CSAT Score	Statistics → User Profiles → CSAT Summary	Professional plan feature
User Source Channel	Statistics → Split Links → Attribution Data	Ad/Social/Email sources

For export, the console currently supports CSV export (see official documentation for details).

Common KPI Dashboard Field Configuration and Best Practices

A comprehensive dashboard should include the following 6 fields, covering efficiency, quality, and channel attribution:

Configuration Notes:

1. Define the boundary between Bot resolution and manual resolution: In TG-Staff, if a conversation is first handled by the Bot, then transferred to a human agent and resolved by the agent, it should be counted as “Transferred to human” rather than “Bot resolved.” It is recommended that the team standardize the rule: any conversation with a human reply should be considered non-Bot resolved.
2. CSAT rating should be mandatory: Embed a rating card in the Bot message after the conversation ends, and set either a “default 4 points if not rated” or “exclude if not rated” strategy. The latter is recommended to avoid data distortion.
3. Attribution data for diversion links: If you run ads (e.g., Google Ads, Facebook), add the utm_source parameter to the diversion link URL. TG-Staff will capture these parameters and associate them with the conversation record. This allows you to compare resolution rates and CSAT differences between “ad-sourced users” and “organic traffic users.”
4. Regularly calibrate data definitions: Check monthly whether all agents have a consistent understanding of the Bot resolution definition to avoid statistical drift.

How to Interpret KPI Data and Drive Team Improvement?

Once you have data, the key is how to interpret it and turn it into action.

Data Interpretation Framework:

Scenario	Possible Cause	Improvement Action
Resolution rate < 60%	Incomplete Bot flow, common issues not covered	Add high-frequency issue branches in TG-Staff’s visual command flow
Transfer to human rate > 40%	Loose diversion rules, Bot cannot identify user intent	Check “online first” rules; optimize keyword recognition in Bot welcome message
CSAT < 3.5	Slow agent response, Bot replies too mechanical	Adjust diversion rules to allow more human intervention during peak hours; optimize Bot scripts
High transfer rate + High CSAT	Bot accurately identifies complex needs and transfers to human	Maintain status quo; this is ideal
Low transfer rate + Low CSAT	Users are “lost” in the Bot	Add “transfer to human” prompts; optimize Bot menu

Improvement Suggestions for Telegram Scenarios:

1. Optimize command flow: In TG-Staff’s drag-and-drop editor, analyze the content of transferred conversations to identify high-frequency issues, then add corresponding Bot reply branches. For example, if 5% of transfers are asking “how to reset password,” add an automated reply flow.
2. Adjust diversion link scripts: If a particular ad channel brings a very high transfer rate (e.g., over 60%), the ad copy may not match Bot expectations. Modify the guidance script on the diversion link page to inform users of what the Bot can and cannot do.
3. Train agents: For agents with low CSAT scores, review their conversation history and analyze reply quality. TG-Staff’s content moderation feature (Pro version) can detect risk words in agent messages, helping to avoid inappropriate replies.

FAQ

Q: How is the bot resolution rate calculated?
A: Bot resolution rate = Number of conversations resolved independently by the bot ÷ Total number of conversations × 100%. Independent resolution means the conversation was not transferred to a human agent and the user did not ask repeated questions, typically based on the conversation end state. You can view bot processing results in TG-Staff’s conversation records.

Q: How to collect CSAT ratings in Telegram?
A: After a conversation ends, the bot can send a rating message (e.g., 1-5 stars). Users can click to rate. TG-Staff Professional’s user profiling feature records CSAT scores for each conversation and aggregates them into an average score.

Q: What to do if the transfer-to-human rate is too high?
A: A high transfer rate usually indicates the bot cannot meet most user needs. Recommended steps: 1) Analyze transferred conversation content to identify frequent issues; 2) Add corresponding branches in TG-Staff’s visual command flow; 3) Adjust routing rules (e.g., enable “Online First” to reduce transfers caused by bot no-response).

Q: Which TG-Staff plan supports the KPI dashboard?
A: Basic conversation data (e.g., conversation count, agent assignment records) is available in the Standard plan; user profiling, statistics, CSAT score aggregation, etc., require the Professional plan (approx. $16.99/month). Check the official plan page for specific features.

Q: Can data from routing links be included in the KPI dashboard?
A: Yes. TG-Staff’s routing links capture visitor IP, browser info, and URL parameters. This data can be linked with conversation records to help analyze user conversion rates and customer service KPI differences across channels (ads, social media, email).

Summary & Next Steps

The core KPI dashboard is not a static report, but your team’s “instrument panel”. Start with bot resolution rate, then gradually add transfer rate and CSAT score. You can evolve from “how much was done” to “how well it was done”. This week, do two things: enable user profiling in TG-Staff console and configure a CSAT rating card.

Act now:

• Sign up for a free trial: https://app.tg-staff.com/
• Check official docs for Professional statistics features: https://docs.tg-staff.com/
• Contact @tgstaff_robot for 1-on-1 setup guidance

Drive your Telegram AI customer service team with data, starting today.