Sylar Cloud
Since October 1, 2026, Meta charges for WhatsApp Business API service messages : the replies your agents and chatbot send when the customer wrote first. Each number gets 1,000 free messages per month and from the 1,001st on, every reply has a cost. If you want to see rates and examples, check how much WhatsApp Business API costs in Colombia.
The per-message rate is low, but in an operation with thousands of chats a day the bill grows fast. The good news: most of the savings don't depend on negotiating prices, but on sending fewer messages to resolve the same thing. These are the practices with the most impact.
1. Every message counts, not every conversation
Before, you paid per 24-hour conversation, no matter how many messages were sent within it. Now you pay per delivered message. That changes how you serve: a greeting, a "one moment please", the answer and an "anything else?" are four charged messages where before there was a single paid one.
2. Optimize the chatbot
The chatbot is usually the one sending the most messages, because it handles all the inbound volume. Small tweaks to its design show up on the bill:
- Merge the greeting and the menu into a single message, instead of sending welcome, privacy notice and options separately.
- Use lists and buttons instead of asking the customer to type a number and then confirm with another message.
- Ask for several data points at once when it makes sense: "Write your ID number and the reason for your inquiry" instead of two separate questions.
- Remove filler messages such as "Thanks for your reply", "We're validating your information" or "One moment".
- Review the loops: if a customer replies something unexpected and the bot repeats the menu three times, you're paying for three messages that resolve nothing. After one failed attempt, transfer to an agent.
- Resolve frequent questions in the bot (hours, order status, balances, requirements) so they don't reach an agent who will send several more messages.
A well-configured AI agent helps even more: it understands the question in natural language and answers with the complete information at once, without multi-step menus.
3. Train your agents to resolve in one message
The habit of chatting "like with a friend", with short messages one after another, is natural, but it now has a cost. A short training with these guidelines is worth it:
- Greeting and answer together. "Hi Ana, I'm Carlos from Customer Service. Your order 4587 ships today and arrives tomorrow before 6 p.m." instead of "Hi", "I'm Carlos", "Let me check", "Your order ships today".
- Read everything before replying. Many customers send their inquiry in several messages; if the agent answers the first without reading the rest, the conversation gets longer.
- Anticipate the next question. If the customer asks how much they owe, include the due date and how to pay in the same reply.
- Avoid "anything else?" as a separate message. Include it at the end of the main reply.
- Ask for everything needed at once. If moving forward requires an ID, email and a photo of the invoice, ask for them in a single message with a list.
- Use prepared, complete quick replies for frequent cases, with links and data included.
Measuring helps the habit stick: compare the average number of messages sent per conversation across agents and reward those who resolve with fewer messages without lowering satisfaction.
4. More tips to optimize costs
Use the right template category
A utility template (payment reminder, appointment confirmation) costs a fraction of a marketingone. Write transactional notifications without promotional language so Meta approves them as utility.
Segment your mass campaigns
Sending marketing to the whole database is the fastest way to spend. Filter by customers with real interest, exclude those who shouldn't receive messages and avoid sending the same campaign twice to the same person from different databases.
Take advantage of ads that open WhatsApp
Conversations that arrive from a Facebook or Instagram ad with a WhatsApp button have a free 72-hour window. If you run ads, that's your cheapest channel to serve.
Avoid reopening the window with templates
If the 24-hour window closes, you need a template to write again. Reply on time: a low first-response time not only improves the experience, it also avoids extra costs.
Move other topics to cheaper channels
Long documents, statements or supporting files can be sent by email, and complex cases may be better resolved with a call. Choose the channel by type of interaction, not out of habit.
Review usage every week
See how many messages the chatbot sends, how many the agents send and how many the campaigns send. What isn't measured can't be optimized.
Summary: checklist for your operation
| Area | Action | Impact |
|---|---|---|
| Chatbot | Merge greeting and menu, buttons, no filler messages, no loops | High: handles all inbound volume |
| Agents | Resolve in one message, read everything before replying, quick replies | High: each conversation goes from several messages to one or two |
| Templates | Utility category for transactional messages | High on mass sends |
| Campaigns | Segment and don't repeat the same person | Medium to high |
| Ads | Ads that open WhatsApp (72 h free) | Medium |
| Control | Measure messages per conversation and per agent | Sustains the savings over time |
WhatsApp in Sylar Cloud
In Sylar Cloud you get WhatsApp Business API with a multi-agent inbox, chatbot, AI agents with a knowledge base, quick replies, mass sends with filters and the one-person-at-a-time rule to avoid repeating messages. See the plans and pricing.