WhatsApp is your shopfront

What a message costs

Everything in this section is the Platform's pricing, from Meta's own documentation as read on 11 November 2026. [3]

Per-message pricing replaced the older conversation-based pricing on 1 July 2025. The rules, in plain words:

The customer service window is the period that opens when a customer messages you. Check its current length in Meta's documentation, and ask your provider which category each of your templates falls in.

For India there are three dates to know. [3]

1. 1 January 2026. India's marketing rate went up, and billing in rupees began.
2. 31 December 2026. Eligible businesses must have moved all their WhatsApp Business Accounts to rupee billing.
3. 1 January 2027. Meta stops delivering messages from non-rupee accounts of eligible customers.

Ask your provider whether your account counts as eligible. If you are reading this in 2027 and your Platform messages have stopped going out, check the billing currency before anything else.

Now the rupee amounts, and a caution. Meta's page links to rate cards but, in the version read for this book, did not print the figures. The amounts below come from a messaging vendor's blog (Monty Mobile, 28 September 2026): marketing messages ₹0.8631 each, up from roughly ₹0.7846, utility and authentication about ₹0.115 each, with 18% GST extra. [5] That is a secondary source with something to sell. Confirm against Meta's rate card or your provider's invoice before you budget.

Example (fictional): A hardware wholesaler in Rajkot uses the Platform and wants to announce a festive price list to 2,000 retailers. At the marketing rate quoted above, 2,000 messages cost ₹1,726.20, and about ₹2,037 with 18% GST. He also asks his provider what it charges on top. Then 140 retailers reply with questions. The answers his team types back while each customer's window is open are free. The numbers are illustrative.

What stays true:

• The cheapest message is a reply to a customer who wrote first. Build the shop so the customer starts: your number and a QR code on every bill, box, counter and web page, and a reason to message ("Send HI for the new catalogue").
• Messages you start cost money and attention. Send fewer, to people who asked.
• If you run Click-to-WhatsApp ads, the free period after the click is when to answer, quote and follow up. A lead answered two days late has used up most of it.
• Rates change. Give one person the job of reading Meta's pricing page and your provider's invoice every quarter.

Tidy the counter before you automate anything

Before AI goes near your WhatsApp, fix what is already there. Your saved quick replies were probably written by five different people over several years. Make them sound like one business.

1. Copy your most-used replies into one note: price list request, delivery time, store timings, returns, thank you. Leave out bank details and account numbers.
2. Run the prompt below.
3. Check every fact yourself: prices, timings, policies.
4. Update the saved replies so the whole team uses the same set.

Below are the saved replies our team sends to customers on WhatsApp. They contain no customer details, bank details or account numbers.

[paste your most-used replies]

Task: rewrite them in one consistent tone: warm, clear and short.
- Keep every fact exactly the same: prices, timings, policies, names.
- Do not add offers, promises or urgency.
- Keep each reply under [number, e.g. 60] words.
- Give each reply a short name our staff can search for, such as "timings".
- At the end, list anything that looks unclear, out of date or contradictory. Do not fix those yourself.

Check: read each rewritten reply against the original. AI should not change a fact, but you confirm.

Three more pieces of housekeeping:

• Labels. Mark every chat with its stage: New, Quoted, Follow-up, Won, Lost. Your follow-up list from Chapter 8 then sits inside the app.
• Away message. Say when a person will reply. Promise nothing else.
• Catalogue. Make it match your current price list and product facts. Customers read it, and so does any assistant you later switch on.

What stays true: same brand, same voice, whoever is holding the phone.

Broadcasts: consent first, then usefulness

A broadcast list of retailers and repeat customers is a direct line to buyers. It is also an easy way to get blocked. Treat it as a channel with house rules.

1. Only people who agreed. Add three columns to your customer sheet: agreed to receive messages (yes or no), date, and how they agreed. A visiting card collected at an exhibition is not agreement to receive your offers.
2. An exit in every message. "Reply STOP to stop these messages." Remove the person the same day.
3. One useful thing per message. A new range, a restock, an order cut-off before the festive rush. Chapter 8 makes the same point about follow-ups.
4. A written limit. Decide how many broadcasts a month each list receives, and hold to it.
5. Separate lists. Retailers, end customers and lapsed buyers need different messages.

The GoKwik report offers one comparison worth knowing. Across its brands, automated customer journeys averaged an 11.1% click-through rate, against 2.6% for broadcast campaigns. [2] It is vendor data from D2C brands, so do not expect the same figures. The likely reading is simple: a message tied to something the customer just did is more welcome than a message sent to everyone.

A phone number is personal data. A law-firm summary from August 2026 places the main duties of the Digital Personal Data Protection Act, including notice and consent, in May 2027, and other sources disagree on some dates. [6] Chapter 21 sets that out. Ask your lawyer what your consent record must contain.

[Paste your business note]

Audience: [e.g. retailers who ordered in the last 12 months and agreed to receive updates]
Purpose: [e.g. announce the festive range / share a restock / remind about the order cut-off]
Facts: [paste the product facts, and any dates or offers I have confirmed]

Task: write one WhatsApp broadcast under 80 words.
- Open with what is useful to the reader, not with our name.
- Include one clear action, e.g. "Reply YES for the catalogue PDF".
- End with: "Reply STOP to stop these messages."
- No false urgency such as "only today" or "last chance" unless it is in my facts.
- Use only the facts I have given. Do not add prices, discounts, dates or claims.

Check: the owner confirms every date and offer. Send it to yourself first and read it on a phone.

What stays true: permission, usefulness and restraint. No pricing change or new feature will make an unwanted message welcome.

The AI assistant inside WhatsApp

Here is what is known, as of the announcement on 14 May 2026. [4] Business AI was announced for eligible small and medium businesses using the WhatsApp Business app in India. It works in Indian languages. It is trained on the business's own profile and catalogue. The owner can take over a chat or switch the assistant off. In-chat UPI payments are planned, with no date given. The announcement states no price.

Meta's announcement includes two small Indian businesses. The Purple Sunset, a personalised gifting business, handled 60 to 70 queries a day before the tool. It says it now closes 6 to 7 orders a day through AI-handled queries and reports a 40% increase in sales. Soil Concept, a plant-based personal care brand, reports an 80 to 90% conversion rate after adding round-the-clock support. [4]

Be clear about what those are. They are the businesses' own figures, published by the company that makes the tool. Nobody independent has checked them. Treat them as marketing claims.

What is missing is an independent test. The research for this book found none, so nobody can yet tell you how the assistant behaves when a customer asks something your profile and catalogue do not cover.

Now set that against the rule this book runs on: a person reads every message before it reaches a customer. An assistant that writes and sends its own replies breaks that rule by design. Chapter 9 gave the reasons to wait, including a company held to what its chatbot said. So the recommendation here is plain. For now, do not hand customer conversations to it.

Fixed messages are different. An away message, a quick reply or an order update template was written and approved by a person once. Automating those keeps the rule, because nothing new is being composed.

If you still decide to test the assistant, that is the owner's decision, made under the team rules in Chapter 24. Make these true first:

1. Your FAQ sheet from Chapter 9 is finished, and your profile, catalogue and prices are accurate, because that is what it learns from.
2. You have tested it as a customer from another phone with 20 real questions, including five it should not answer. Note every fact it invents.
3. A named person reads its chats every day for the first month and takes over any chat about price, custom work, a complaint or credit.
4. If it states one wrong price or date, it is switched off until you know why.

Two safety points apply whichever tools you use. Do not export chats with customer details into an AI tool. Copy only the words you need, with names and numbers removed. And never act on a payment instruction that arrives only on WhatsApp, even from a familiar name and photo. Chapter 23 explains the fraud behind that rule.

What stays true: automate fixed words a person has approved. Keep a person between any newly written sentence and your customer.

Mistakes to avoid

• Budgeting from a blog. Take per-message rates from Meta's rate card or your provider's invoice, with the date.
• Missing the rupee billing deadline. After 1 January 2027, messages from non-rupee accounts of eligible businesses are not delivered.
• Broadcasting to every number you have ever collected. No record of agreement, no message.
• Letting AI rewrite quick replies and saving them unread. A changed timing or price becomes your promise.
• Switching on an AI assistant because the announcement quotes a sales increase. Those figures were published by the company that makes it.
• Using WhatsApp as the order book, the complaint register and the accounts ledger. It is the shopfront. Records belong in a sheet or a system.

Do this week

1. Find out which WhatsApp you are on. If it is the Platform, ask your provider three things: is our billing in rupees, what are the current per-message rates by category with GST, and how many messages of each category did we send last month.
2. Clean your quick replies with the prompt in this chapter, check every fact, and update the saved set.
3. Add the three consent columns to your customer sheet and write your broadcast limit at the top of it.

Sources

1. afaqs, report on WhatsApp's Mumbai business summit (payments, AI support, ads in Status; Kantar 2025 figure cited by Meta), 16 September 2025. https://www.afaqs.com/news/media/whatsapp-sharpens-its-business-play-at-mumbai-summit-with-payments-ai-support-and-ads-in-status-10470432
2. Outlook Business, report on the GoKwik WhatsApp Commerce Intelligence Report 2026 (vendor data), 18 June 2026. https://www.outlookbusiness.com/corporate/d2c-brands-lean-on-whatsapp-to-find-new-customers
3. Meta developer documentation, "Pricing on the WhatsApp Business Platform", read 11 November 2026. https://developers.facebook.com/documentation/business-messaging/whatsapp/pricing.md
4. Meta Newsroom, announcement of Business AI on WhatsApp for small businesses in India (vendor-published), 14 May 2026. https://about.fb.com/news/2026/05/introducing-business-ai-on-whatsapp-for-small-businesses-in-india/
5. Monty Mobile (messaging vendor's blog; secondary source for rupee rates), 28 September 2026. https://montymobile.com/blogs/whatsapp-business-api-pricing-in-india-inr-rates-gst-and-the-2026-currency-migration-deadline
6. King Stubb & Kasiva via Mondaq, summary of DPDP Act and Rules compliance milestones, 14 August 2026. https://www.mondaq.com/dpdp-act-and-rules-2025-the-2026-compliance-milestones-businesses-cant-afford-to-miss/1830402