For most local service businesses in India, no. A WhatsApp Business account with a good greeting message, saved quick replies and a genuinely fast human reply converts better and costs nothing. A chatbot earns its place in three specific cases: real after-hours enquiry volume, the same five questions eating hours every week, or enough traffic that routing needs a receptionist. Below roughly a hundred enquiries a month, the setup and upkeep cost more than the bot saves.
Chatbots are sold on a promise — never miss a lead again — that assumes you are currently missing leads to unanswered messages rather than to slow replies. For most of the businesses we work with, the second is the real problem, and a bot does not fix it.
What is the actual difference between a bot and a WhatsApp auto-reply?
They are solving different problems, and the comparison is usually made as though they were the same product.
| Website chatbot | WhatsApp auto-reply | |
|---|---|---|
| Catches | Someone who will not message you yet | Someone who already did |
| Answers questions | Yes, immediately | No — it buys time |
| Conversation lives | In a widget the visitor leaves behind | In a thread the customer keeps |
| Setup cost | Real — build, content, testing | Fifteen minutes, free |
| Ongoing upkeep | Yes — it goes stale | Almost none |
| Risk when wrong | Answers a pricing question incorrectly | Says nothing it should not |
In India the channel question usually settles it. WhatsApp is where customers already are, which is why the strongest setup for most local businesses is a prominent click-to-WhatsApp path with a disciplined auto-reply behind it — the full setup is in the WhatsApp Business guide.
When is a chatbot genuinely worth it?
Three situations, and they are specific enough to check against your own numbers this week.
- Real after-hours volume. If a meaningful share of enquiries arrive between 10pm and 8am and go cold by morning, a bot that captures the requirement and a phone number is doing something a human was never going to do.
- The same five questions, endlessly. Opening hours, location, whether you deliver to a particular area, whether a service is available, what documents are needed. If answering these consumes hours a week, automate exactly those five and nothing else.
- Enough traffic that routing is a job. A business fielding several hundred enquiries a month across three service lines gains from qualification and routing before a human ever reads the chat.
Notice all three are volume problems. Nobody has ever fixed a conversion problem with a chatbot. If enquiries are arriving and not converting, the gap is in the reply and the offer — the same diagnosis as in ads that spend money but bring no leads.
What does a chatbot actually cost to run?
The subscription is the smallest line. The real cost is in three places people do not budget for.
- Building the content. Every answer has to be written, in your voice, matching what you actually do. This is a day of work, not an afternoon.
- Testing it against real questions. Take fifty genuine enquiries from your inbox and run them at the bot. The failures are the entire brief.
- Keeping it current. Prices change, services change, hours change. A bot quoting last season's price is worse than no bot, because it has made a promise on your behalf.
Add the risk that the bot itself is the friction. A widget that opens uninvited, blocks the phone number behind it and asks for an email before saying anything useful loses more enquiries than it captures. If you already send ad traffic to a page, the landing-page fundamentals in landing page or website for ad traffic matter far more than any bot bolted on top.
When must a human take over?
Four triggers, and every one of them should be an immediate handover, not an escalation path.
- Any question about price. A bot guessing at price has made a commitment in your name, and unwinding that costs more than the enquiry was worth.
- Any question about timeline or availability. Only a person knows what this week actually looks like.
- Any sign of frustration. "I already asked this" is the last message before someone leaves.
- The second failure to understand. Once is tolerable. Twice reads as being stonewalled.
The framing that works: a bot holds a conversation for two or three exchanges and hands over warm. It does not close. That boundary is the same one that applies to AI agents generally — automation is fine right up to the point where it starts making promises.
What should you measure?
Not conversations. Conversations are what the vendor's dashboard shows, and they are not the number that decides anything.
| Metric | What it tells you |
|---|---|
| Conversations started | Whether anyone engages at all |
| Contact details captured | Whether it produces anything actionable |
| Handovers to a human | Whether it is qualifying or just deflecting |
| Enquiries that became customers | The only number that justifies the cost |
| Enquiries before vs after install | Whether it added leads or just moved them |
That last row catches the most common false positive. A bot reporting 200 conversations a month in a business that had 200 enquiries a month before it was installed has captured a channel, not created demand.
For context on the alternative: the Shivsagar Tours & Travels account we have run since February 2025 has produced 1,500+ leads with no chatbot anywhere in the funnel. Ads to WhatsApp, a structured opening message, a human replying fast. Across the 10 industries we work in, that pattern still outperforms the automated version for anything sold on trust — which is most of what a local business sells. Where a bot does fit, it belongs inside a properly built enquiry funnel, not bolted onto a broken one.
Key Takeaways
- For most local businesses in India, WhatsApp with quick replies and a fast human answer beats a chatbot and costs nothing.
- A bot earns its place on volume problems only: after-hours enquiries, the same five repeated questions, or routing at scale.
- Below roughly a hundred enquiries a month, setup and upkeep cost more than the bot saves.
- The subscription is the small cost. Writing the content, testing against real enquiries and keeping it current is the real one.
- Hand over to a human on price, timeline, frustration, or the second misunderstanding — every time.
- Measure enquiries before versus after install. A bot that reports conversations may only have moved them.
Before You Ask
Is an AI chatbot worth it for a local business?
For most local service businesses in India, no — a WhatsApp Business account with a good greeting message, quick replies and a fast human reply converts better and costs nothing. A chatbot becomes worth it in three situations: when you genuinely get enquiries outside working hours, when the same five questions consume hours every week, or when a receptionist would otherwise be needed to route enquiries. Below roughly a hundred enquiries a month, the setup and maintenance usually cost more than the bot saves.
Is a chatbot better than a WhatsApp auto-reply?
They solve different problems. A WhatsApp auto-reply buys you time on a channel the customer already prefers and keeps the whole conversation in one thread the customer can return to. A website chatbot catches someone who is not ready to message and would otherwise leave, and can answer questions immediately. In India, where WhatsApp is the default channel, the strongest setup for most local businesses is a prominent click-to-WhatsApp button with a disciplined auto-reply, and a bot only if the volume genuinely justifies it.
When should a chatbot hand over to a human?
Immediately on any of four triggers: a question about price, a question about timeline or availability, any sign of frustration, or the second time the bot fails to understand something. Every one of these is a moment where a wrong answer costs a customer, and a bot that guesses at pricing has made a commitment in your name. A good bot's job is to hold the conversation for two or three exchanges and hand over warm, not to close the sale.
*The 1,500+ leads figure is from the Shivsagar Tours & Travels account managed by Safar Spectrum Media, measured since the February 2025 handover. Volume thresholds quoted are our own working rules from client accounts, not published benchmarks. The 41+ brands, 10 industries and 25+ ad accounts figures are SSM's own, as of September 2026.