How Many Conversations Can a Customer Service Agent Handle per Day on WhatsApp?
There's no single right number. Learn what drives agent workload, a formula to calculate it with your data, what sources say and how an AI agent helps.
It depends on the number of concurrent chats, the actual working time per conversation and how complex the questions are. Calculate it like this: actual shift minutes × occupancy rate ÷ average working minutes per conversation. A hypothetical example: 420 minutes × 75% ÷ 6 minutes = about 52 conversations, falling by half if the conversations are complex.
01The short answer: it depends on 5 factors
Any fixed number you read, such as "an agent handles 100 conversations a day", means nothing without context. An agent answering "Where's my order?" is not the same as one handling return complaints.
- Concurrency: How many conversations the agent has open at the same moment.
- Actual working time: The minutes the agent actually spends on the conversation, not its duration from first to last message.
- Complexity: A price inquiry, an order, or a complaint that requires checking systems.
- Channel type: WhatsApp is asynchronous; customers may reply after a minute or an hour, so conversations run longer while the agent's time in them may be short.
- Tools: Quick replies, customer context in front of the agent, and orderly handovers within the team.
02What do the available sources say?
LiveChat's Customer Service Report (data updated in 2024 from more than 35,000 companies) cites a global average of roughly 84 chats per agent per day, with an average chat duration of about eight and a half minutes. But it covers website chat and varies widely between industries, so don't apply it to WhatsApp as is.
In a case study published by CustomerThink in 2016, a contact center reduced the maximum from 3 concurrent chats to 2 per agent after measuring customer experience, and its NPS improved by 19%. The author herself says there is no single industry standard, and there shouldn't be.
Sources: LiveChat Customer Service Report (last updated 2024), and Jodie Monger in CustomerThink, September 2016.
The lesson from both sources: the numbers vary enormously, and increasing the load can raise the count while lowering satisfaction. Measure against your own data.
03Calculate it with your data: the formula
The formula: daily conversations per agent = (actual shift minutes × occupancy rate) ÷ average working minutes per conversation.
| Item | Hypothetical example 1: simple questions | Hypothetical example 2: orders and complaints |
|---|---|---|
| Shift length | 480 minutes | 480 minutes |
| After breaks and meetings | 420 minutes | 420 minutes |
| Target occupancy rate | 75% | 75% |
| Available working minutes | 315 minutes | 315 minutes |
| Average working minutes per conversation | 6 minutes | 12 minutes |
| Possible conversations per day | About 52 | About 26 |
- Pull the number of closed conversations per agent over two normal weeks from your reports.
- Divide each agent's active time in the system by their number of conversations to get the average working minutes.
- Classify conversations as simple, medium and complex, and calculate the average for each type.
- Review the result alongside first response time and customer satisfaction, not the count alone.
A 100% occupancy rate is unrealistic. Agents need time for documentation, learning and rest, and anything consistently above 85% wears the team out.
04Concurrency: how many conversations at the same time?
On WhatsApp, an agent can keep several conversations open because customers reply slowly. But every extra conversation raises the chance of mistakes and delays. These are starting points for testing, not a standard:
| Conversation type | Suggested starting concurrency | When to lower it |
|---|---|---|
| Repetitive questions: price, location, hours | 3 to 4 | If the second-reply time increases |
| Orders and bookings | 2 to 3 | If order errors increase |
| Complaints and technical support | 1 to 2 | If customer satisfaction drops |
The figures in the table are a hypothetical starting example, not the result of a study. Raise or lower them based on your own numbers.
05How does an AI agent change the workload?
An AI agent takes the repetitive questions and collects details before handing over, so agents get fewer conversations. But the ones left for them are harder, so their average time goes up. Don't expect the count per agent to double; expect the same team to handle a larger volume.
A hypothetical example: 1,000 conversations a day, and the AI agent resolves half of them. That leaves 500 for the team, but at an average of 9 minutes instead of 6, because they're the harder ones. Use the calculator to try your own numbers:
06Signs of burnout and what to do
- First response time rises more at the end of the shift than at the start.
- Shorter, curter replies, and copy-paste answers that don't fit the question.
- Repeated mistakes in orders or information.
- Customer satisfaction drops for a particular agent or on peak days.
- Frequent absences, transfer requests or resignations in the team.
The fix: lower the concurrency limit, distribute difficult complaints fairly, build quick replies for repetitive questions, and let an AI agent cover nights and peak times.
Key takeaways
- There is no single standard number; the count depends on concurrency, complexity and actual working time.
- Calculate it: shift minutes × occupancy ÷ working minutes per conversation.
- An AI agent reduces the count but leaves agents with the harder conversations.
- Watch response time, satisfaction and absences to spot burnout early.
Frequently asked questions
How many WhatsApp conversations does an agent handle per day?
There is no fixed number. In a hypothetical example, an 8-hour shift with 75% occupancy and an average of 6 working minutes per conversation gives about 52 conversations, dropping to about 26 if the average is 12 minutes.
How many conversations should an agent have open at once?
Start with 2 to 3 for regular conversations, and fewer for complaints, then adjust based on response time and customer satisfaction.
Do global benchmarks apply to WhatsApp?
Most of them are about website chat and vary widely between industries. Use them only for general comparison, and rely on your own data.
Does an AI agent replace human agents?
It handles a share of the repetitive conversations and works around the clock, but some conversations still need a human, so it usually reduces your hiring needs rather than eliminating them.
- #Customer service agent workload
- #Concurrent chats
- #Average handle time
- #Support agent burnout