WhatsApp is where Indian customers actually reply, which is why almost every consumer product here eventually builds on the WhatsApp Business Platform. The economics changed materially on 1 July 2025, when Meta moved from conversation-based pricing to per-message billing. Anything designed or costed against the old model needs revisiting, because the levers that reduce spend are now completely different ones.
Templates fall into three categories and each is charged differently. Marketing templates are charged for every message. Utility templates are free when sent inside an open customer service window and charged outside it, with volume tiers unlocking lower rates as monthly volume rises. Authentication templates follow utility pricing rules with the same volume tiers, and some countries carry a separate authentication-international rate. India sits at the higher end for marketing and authentication-international, which makes the tiering worth modelling rather than assuming.
The customer service window is the mechanism that decides most of your bill. It opens when a user messages your business and lasts twenty-four hours, closing if they do not write again. Inside that window all non-template messages are free, and utility templates are free too. Service conversations have been free entirely since November 2024. In other words, replying to customers costs nothing; initiating contact costs money. Every optimisation follows from that asymmetry.
There is a second free path worth designing around. When a user starts the conversation through a Click to WhatsApp advertisement or a Page call-to-action button, your responses within twenty-four hours are free and a seventy-two hour free entry point window opens during which messaging is unrestricted and unbilled. For products with a paid acquisition motion, routing new conversations through those entry points rather than through outbound templates is a straightforward structural saving.
The design consequence is that a support product and a notification product have opposite cost profiles, and building them on one undifferentiated messaging service hides that. Instrument by template category from the first day, keep a live view of what proportion of your utility messages land inside an open window, and treat that percentage as a product metric. Teams that measure it usually find they can move it substantially by changing when they send rather than what they send.
Two operational cautions. Template approval is a real gate with real latency, so anything time-sensitive needs its templates approved well in advance rather than on the day. And quality ratings matter — a business messaging pattern that generates blocks or reports degrades your sending limits, which is a failure mode with no engineering fix. Getting the frequency and relevance right is a product decision that shows up later as a technical constraint.
