How to build a WhatsApp chatbot for your grocery store

The first four are the bot's. The last two stay with people, and the bot's job there is to collect the details so the person does not have to ask again.

Customers will not change how they write a grocery list for your bot, so the bot has to fit the way they already do it. A flow that works:

1. Receive anything. Typed text, a photo of a handwritten list, a voice note. All three are valid orders.
2. Acknowledge immediately: "Got your list — we'll confirm the total once it's packed." This single message removes most "did you get my order?" follow-ups.
3. Ask for the slot with a list message of today's and tomorrow's open slots.
4. Confirm the address — for a returning customer, show the last one with [Same address] [Change].
5. Hand the list to the picking team in the shared inbox, tagged with the slot, so the earliest slot is picked first.

What the bot does not do is price the list. Loose items are weighed, brands get substituted, and a total quoted by a bot and corrected later is worse than a total that arrives a few minutes later from a person.

A large share of grocery orders repeat. The same family buys the same milk, bread, eggs and vegetables every few days, and typing it all again is the part they least enjoy. A reorder flow removes it:

1. The customer types "same as last time", "usual", or taps [Order again] on the last delivery message.
2. The bot shows the last order as a short list and asks: [Order this] [Change something].
3. "Change something" lets them reply in free text — "no bread this time, add 1 kg sugar" — which goes to staff with the original list attached.
4. The bot asks for the slot and confirms, exactly as in a new order.

Put the [Order again] button on the delivered message. It is the moment the customer has the store in mind and the shelf at home in front of them.

Notice what the business side does: it answers with the specific detail (time, reference, next step), it asks for one thing at a time, and it gives the customer a one-word way to reply. Those three habits matter more than how clever the bot is.

WhatsApp gives a bot two interactive formats, and a grocery flow uses both:

• Reply buttons — up to three per message. Right for yes/no and small choices: [Replace] [Leave it out], [Same address] [Change], [Order again] [Something else].
• List messages — a menu of up to ten rows behind one button. Right for delivery slots and for "what would you like to do?" menus.

Tapping is faster than typing for everyone, and it removes spelling from the problem — no bot has to understand "tmrw evng" if the customer tapped "Tomorrow, 5–7 pm".

Slots are where a grocery bot most often promises something the store cannot deliver. Before launch, write down the rules the bot works from:

• Cut-off per slot — for example, orders for the 5–7 pm slot close at 3 pm, so there is time to pick and pack.
• Capacity per slot — how many orders one rider can take. A full slot should disappear from the list, not accept a twelfth order.
• Same-day and next-day — which slots are offered after the last same-day cut-off.
• Closed days and festivals — the bot should know the store is shut on Diwali evening before a customer asks for that slot.
• Areas with different slots — farther areas may only get one delivery run a day.

When the rules change — a rider on leave, a festival rush — change them in one place, and the bot changes with them.

Reputoo supports both, and most grocery stores end up using both:

• Keyword flows — a fixed menu or trigger word ("MENU", "BOOK", "STATUS") leads down a set path. Predictable, easy to audit, right for anything with a fixed answer.
• AI chatbot — reads the customer's own words and answers from your knowledge base (website, price list, policies). Right for the long tail of questions nobody wrote a menu item for.

A sensible start for a grocery store: keyword flows for the top two or three tasks, AI for everything else, and a person behind both.

Grocery customers write in the language they speak at the counter, which in much of India means mixing two. A keyword flow built only on "ORDER" will miss "saman bhejna hai", "list bhej rahi hoon" and a photo with no text at all. Two fixes:

• Give keyword triggers the words your customers really use, in every language you hear at the counter.
• Let the AI chatbot take anything that does not match a keyword, since it reads the meaning rather than the exact word — and replies in the customer's language.

"Is paneer available?" is the question a grocery bot is most tempted to answer and least able to. Unless your inventory is connected and updated as items sell, the bot does not know. The right reply routes the question: "Let me check with the store — someone will reply in a few minutes." If fresh items sell out every afternoon, add that to the knowledge base as a fact the bot can repeat.

• Complaints, refunds and anything where the customer is upset.
• Anything involving money the bot cannot see — a disputed bill, a custom quote.
• Health, safety or legal questions.
• Any customer who asks for a human. Do not make them ask twice.

For grocery stores specifically: missing or damaged items, substitutions the customer rejected and payment issues need a person.

Handover on Reputoo moves the chat into the shared inbox with the conversation so far, so staff never ask the customer to repeat themselves.

1. Connect your WhatsApp number to Reputoo (official API; takes minutes).
2. Add your website and documents to the knowledge base so the AI answers in your grocery store's own terms.
3. Build keyword flows for the top tasks above. What to connect: A billing POS and sometimes a delivery app; the order history is what makes reordering work.
4. Choose who gets handed-over chats, and during which hours.
5. Test from a personal phone, then share the number or a QR code with customers.

• Send a typed list, a photo, and a voice note from a personal phone; each should reach the picking queue.
• Order at 11 pm and check the reply sets the right expectation for the morning.
• Ask for an item in stock and one that is not; both answers should be honest.
• Complain about a late delivery and check it reaches a person with the order attached.
• Write in each language your customers use, including mixed.

Every bot meets messages it does not understand. For a grocery store the fallback should be short and useful: "I've passed this to the store — someone will reply within a few minutes during opening hours." Two details matter. Say when a person will reply, especially at night. And never loop the customer back to the main menu more than once; if the second attempt fails, hand over.

Three things tell you whether the chatbot is helping or getting in the way:

• Orders that reach the picking queue without a person touching the chat first. This number should grow week by week as the flows settle.
• Handovers, and why. Read them. A recurring reason — a delivery-area question the bot cannot answer, a word it does not recognise — is a flow to fix, not a staffing problem.
• Chats with no reply. Any customer message left unanswered is a lost order. After hours, the bot should be the reason this number is zero.

Review these weekly for the first month, then monthly. Most of the gains come from fixing the handful of messages the bot keeps getting wrong.