Dashboard
Chats, hand-overs and response times for the days you pick. i
The reason the bot gave when it transferred.
How these numbers are counted
- A chat
- One visit of a customer. It starts with their first real message (not "hi", a thank-you or a rating). It ends when an agent closes it, when the quiet-chat timer closes it, or when the customer has been silent with the bot for the time set under Team settings, Timers. A rating tap, or a thank-you within 10 minutes of the end, belongs to the chat that just ended. Their next real message starts a new chat, also on the same day.
- Total chats
- Chats that started in the period (IST). A customer who comes twice counts twice.
- Human transferred
- Chats the bot handed over to people, whether or not an agent joined.
- Bot handled
- Chats with no hand-over to people.
- Open chats
- Handed-over chats an agent has replied to and not closed yet.
- Closed chats
- Chats that have ended and got an answer: closed by an agent, or ended with the bot having replied.
- Bot abandoned
- Chats that ended with the customer's last message unanswered.
- Transfer unanswered
- Handed-over chats where no agent has replied yet, for example a hand-over nobody picked up.
- Agent figures
- An agent is counted in every chat they replied in. A chat two agents replied in counts once in the total and once for each agent, so the agents can add up to more than the total. "Assigned, no reply": chats an agent was given but never replied in.
- What happened
- Worked out from the record when the chat ends: greeting only, dropped at confirmation (the bot asked "Just to confirm...?" and got no answer), resolved – customer confirmed (Satisfied, or a thank-you after the answer), bot answered, plan lead sent, transfer unanswered, agent handled, no reply.
- Avg agent wait
- From the hand-over to the first reply by an agent.
- Avg handling time
- From the agent joining to the chat ending.
Quality
The AI's nightly review of chats the numbers alone can't judge, and QA's verdicts. i
What customers asked that no knowledge base entry answers: correct "no answer" hand-overs, and bot answers that were not from the knowledge base. Write these entries first.
Executive summary
The numbers that matter, compared with the period before. i
What customers asked that no entry answers (from the AI analyser): add these to the knowledge base.
Agents
How each agent is doing with the chats handed to people, and who can sign in.
Closing reasons chosen by agents in this period.
Attendance
Check-ins, breaks, idle time and productive hours for every agent. Times are in IST.
Productive time is time checked in, minus breaks and idle time. productive break idle
Team settings
When people answer chats, how chats reach them, breaks, closing reasons and the customer rating.
When chats are handed to people. Outside these hours the bot tells the customer when the team is back. Times are in IST.
The number and hours the platform's own messages give customers: "I'm unable to connect you to a specialist right now... or call us ..." and the technical-issue line. Write the hours as they read inside the sentence. The bot's prompt has its own copy of these words: change them in Bots too.
How long the desk waits before it acts. Hover over or tap i for what each one does.
The same actions as the myBillBook admin panel's Change number and Swap Data/Plan: change mobile number, data swap, data merge, plan swap, data + plan swap. The old number is the one the customer chats from; another old number needs a screenshot of the app logged in on it, sent in the chat. Each change is prepared first: the agent sees myBillBook's preview of exactly what will change, then confirms it. Every change is recorded in the chat.
Agents put these on a customer's chat from the desk (for example VIP or Follow-up needed). A label stays with the customer's number, and Conversations can be filtered by it.
What agents choose from when they take a break.
Past the target a chat's timer turns red on the desk, and supervisors and admins get an alert. The Dashboard shows the share of first replies within the target.
Who gets a waiting chat. Languages and skills are preferences, never limits: an agent who matches is offered the chat first, and every agent still gets all other chats when they are free. Set each agent's languages, skills and own limit under Agents > Accounts and access.
A chat needs a skill when one of its words appears in the bot's hand-over note or in what the customer wrote. Separate the words with commas.
Sent on WhatsApp when an agent closes a chat, when the bot finishes a chat, and when a customer asks to rate. The customer taps "Satisfied" or "Not satisfied" (or replies 1 or 2). The message goes in the language the customer was writing in. Ratings show on the Dashboard and the Agents page.
An agent must pick a category and sub-category before closing a chat. Switch one off to stop offering it; chats already closed keep it.
My account
Your sign-in details and password.
Forgotten your password? An admin can reset it from Agents > Accounts and access.
Bots
Customers talk to the bot marked Live; try changes on another bot first. i
Knowledge base
What the bot answers from, and known issues (SOS) told to every customer who asks. i
A document must use headings like "Q17: How to create a sales invoice?" followed by the answer, with "TOPIC 3 — SALES" lines for sections.
Write each known issue like a knowledge base entry: the question customers ask, and the answer to give while it is broken (for example "We are aware of this issue and our team is fixing it. Please try again in 30 minutes."). Every issue is off until you switch it on. While it is on, every customer who asks about it gets this answer; other questions are answered as usual, and a customer who asks for a person is connected. Choose an end time so it switches off by itself.
API connectors
Let the bot call your systems during a chat, like looking up a plan or creating a lead. i
Ticket types
What people can raise, which team gets it, its targets, fields and rules. i
Teams
Who works on tickets, their lead, working days and hours (IST), and the holidays.
Paste many dates
Settings
Quick replies for agents, the AI model and what it costs, and system details.
Gupshup wallet. Replies stop reaching customers when it runs out.