Chatbot vs. Live Support: Which One Fits Your Business?

"Should we set up a chatbot or hire someone?" is usually the wrong question. The two aren't substitutes for each other. A chatbot handles frequent questions with known answers 24/7; live support handles conversations that call for a decision, a complaint or an exception. The real question is which conversation goes to which.

This post gives you a decision framework for making that split. I'm not giving you a fixed "this business gets that" table, because even within the same industry, the mix of questions varies from business to business. Instead, I explain how to decide by looking at your own conversations.

Three groups of questions

What goes to the chatbot

• Opening hours, address, shipping time, payment methods: the answer is always the same
• Order status, appointment time, stock: the answer is in the system, if the chatbot is connected to it
• Frequent questions that come in at night and on weekends

What stays with a human

• Return disputes, complaints, payment problems
• Special pricing and exception requests
• Sensitive topics and angry customers
• Questions the chatbot couldn't answer twice in a row

The third group may be small, but it's the most valuable: that's where the conversations that win or lose a customer happen. A chatbot trying to resolve these on its own is more risk than gain. I covered in more detail which questions should stay with a human in the chatbot guide.

First, run a 30-day message count

30-day message count

1. 1Step 1

   Collect the messages

   Put the last 30 days of WhatsApp, Instagram and website messages in one place.

2. 2Step 2

   Sort them into three groups

   You don't need to read hundreds of messages; even a sample of 100-150 messages shows the distribution.

3. 3Step 3

   List the 10 most frequent questions

   Recurring questions form the chatbot's first knowledge base.

4. 4Step 4

   Calculate the ratios

   If the first and second groups are large, a chatbot reduces the workload; if the third group is large, fix the process and team side first.

How does WhatsApp's 24-hour rule affect this decision?

According to Meta's pricing documentation, non-template replies can only be sent within an open 24-hour customer service window. The window opens when the customer writes; if there's no reply within 24 hours, the window closes and an approved template message is needed to restart the conversation.

For the cost side, also see the post What determines the price of a WhatsApp chatbot.

Handoff rules: when should the chatbot step back?

In a good chatbot, the handoff rules are written from the start. The conversation flow should look roughly like this:

1. 1The customer writesThe 24-hour window opens
2. 2The chatbot replies immediatelyInstant answer to a frequent question
3. 3A handoff trigger firesRequest for an agent, a complaint, two misses
4. 4A summary goes to a humanThe customer doesn't have to ask twice
5. 5The human gets back within 24 hoursBefore the window closes

Write these as handoff triggers:

• When the customer asks for an "agent," a "human" or something similar.
• When the topic involves a complaint, a return, a payment problem or legal wording.
• When the chatbot doesn't know the answer twice in a row.
• When the customer's message is clearly angry or desperate.
• When sensitive personal data is shared.

When handing off, it's important to send a summary of the conversation to the human; making the customer ask the same question a second time gives back the time the chatbot saved.

Checklist before setting up a chatbot

Am I ready for a chatbot?

0/5

• At least a third of customer messages are recurring questions
• The answers are written down and up to date: prices, promotions, opening hours
• It's clear who will pick up a handed-off conversation, and how quickly
• It's clear who will notice if it gives a wrong answer
• The personal data flow is defined with KVKK (Turkey's Personal Data Protection Law) in mind

Check off the items; every empty item is a job to solve before the chatbot.

Who should start with which, and when?

There's no hard rule, but I can sketch a general path:

• Low message volume, varied questions: start with live support and canned reply templates. There isn't enough repetition for a chatbot.
• High message volume, repetitive questions: a chatbot plus human handoff, together.
• Lots of messages at night and on weekends: the chatbot gives the first reply immediately; write strict handoff rules so that handed-off conversations reach a human within 24 hours.
• Mostly order and appointment messages: if the chatbot is connected to your system, this is where you gain the most.

How we work at DijiKoçum

In our AI Assistant service, we start by looking at your existing messages; we determine with numbers what the chatbot will handle and what it will hand off to a human. If you like, we can do this count together.

START YOUR FREE ANALYSIS

Frequently asked questions

Will a chatbot replace a customer service agent?

It takes over part of the routine questions; it needs to hand off conversations that require decisions and exceptions to a human. So your team's workload may shrink, but it won't disappear entirely.

Does a small business need a chatbot?

It depends on message volume and how often questions repeat. If you get a few messages a day, canned replies and a well-organized WhatsApp Business flow may be enough. To decide, count your messages first.

What happens if the chatbot gives a wrong answer?

There's a risk of wrong answers. To reduce it, keep the knowledge base up to date, make sure it says "I don't know, let me connect you with someone who can help" on topics it doesn't know, and read handed-off conversations regularly.

What happens to a customer who messages on WhatsApp at night?

When the customer writes, a 24-hour window opens. If the chatbot gives the first reply immediately, the customer doesn't wait; a conversation handed off to a human needs a reply before the window closes, that is, within 24 hours.

Is a chatbot compliant with KVKK, Turkey's data protection law?

It can be set up to comply, but the personal data flow (what data, where it's stored, where it goes) needs to be defined from the start. This decision is as much legal as it is technical; settle it before setup.

Sources

• WhatsApp Business Platform pricing (Meta for Developers)