Table of Contents

Meta Pricing Changes & How to reduce service message costs

Raquel Updated by Raquel

Starting October 1, 2026, outbound service messages, which are free today, will be charged. This is a meaningful shift for anyone running support, onboarding, or engagement journeys on WhatsApp, and it's worth taking a fresh look at how your messaging is structured.

How much will this cost?

To determine how much this could cost per month:

  • Determine how many service messages you send per month. You can estimate this by going to the Insights page, scrolling down to the Insights Per Month graph, select Last month and hover your mouse over the Insights Per Month to see the total number of service messages for last month.
  • Determine how much utility template messages cost in your region. Go to Meta’s message rates calculator, and:
    • Select your market (message rates are based on the user’s location, not yours — to make calculations easier, select where the majority of your users are)
    • Select USD as the currency
    • Select Utility as the category
  • Estimate your cost. Finally, multiply that number with the total number of service messages you got in step 1.

What is Turn.io doing?

We’re focused on minimising the impact of this pricing change, specifically:

  • Clarifying waiver impact. We’re engaging with Meta about the waiver and how it interacts with the service messages , and we’ll notify you as soon as we have a definitive answer.
  • Building a coalition. We are bringing together affected organisations and their funders to negotiate better rates with Meta. 
  • Product Features. We are working on features to help you reduce the amount of messages you send, such as making Flows easier to work with. We’d love to hear from you if you have specific features that you think will help.

Below is a practical breakdown of where the biggest savings usually live, starting with the highest-impact lever and working down into more specific tactics.

Things you can do to reduce your costs

The single most effective thing you can do is shorten the distance between "contact opens a chat" and "contact gets what they came for."

1. Onboarding, registration, and data collection

If you're collecting information one question at a time — "What's your name?" → reply → "What's your email?" → reply → "What's your address?" → reply: you will be paying for every single round trip.

Switch to WhatsApp Flows. Flows behave like a form embedded inside WhatsApp: name, email, address, date of birth, consent checkboxes, dropdowns, whatever you need, all on one screen. The user fills it out and submits once.

That entire exchange counts as a single message, regardless of how many fields you're collecting.

Good candidates for conversion:

  • New patient registration and KYC-style data collection
  • Appointment booking (date, time, location, notes)
  • Post-consultation surveys and feedback forms

If you're currently running any journey with more than two or three sequential questions, that's a strong signal it belongs in a Flow.

2. AI Agents: prompt for resolution, not conversation

Conversational AI agents can accidentally become expensive by being too good at conversing. An agent that asks a clarifying follow-up, then another, then offers a friendly wrap-up message, then confirms once more before closing — that's four or five messages where one or two might have done the job.

Practical changes to make:

  • Prompt explicitly against chattiness. Add instructions like "resolve the request in as few turns as possible" and "do not add unnecessary confirmations or pleasantries once the answer has been given."
  • Define clear end conditions. Give the agent explicit criteria for when a conversation is resolved, and instruct it to close politely rather than keep the thread open "just in case."
  • Batch information requests. If the agent needs three pieces of information from the user, prompt it to ask for all three in one message rather than one at a time — or better, hand that collection off to a Flow.
  • Avoid redundant confirmations. If the agent already has enough information to act, it doesn't need a message that just restates the plan before doing it.
  • Review your transcripts. Pull a sample of recent AI agent conversations and count messages per resolution. If you're consistently seeing 6+ messages for requests that could be handled in 2-3, that's your prompt to fix.
3. Helpdesk and human support: design for fewer, denser replies

Human agents fall into the same trap AI agents do — it's natural to respond in small increments as you think through a problem. That habit is now a cost driver.

Things worth discussing with your support team:

  • Train agents to front-load information. Instead of "Let me check that for you" followed by "Okay, I found it" followed by the actual answer, aim for a single reply that includes the finding and the next step.
  • Use canned responses and templates for common issues, edited for the specific case rather than built from scratch message by message.
  • Set an internal norm around message count per resolved ticket, and review it periodically the same way you'd review response time or CSAT.
  • Watch for repeat-question patterns. If the same clarifying question comes up over and over before an agent can act, that's a sign the initial intake message (or Flow) should be asking for it upfront.
Other places costs quietly add up

Beyond the three areas above, a few other habits are worth reviewing:

  • Menu-driven flows using free text. If your bot asks "What can I help you with?" and waits for a typed answer, you're relying on the user's phrasing and often needing a clarifying follow-up. Interactive list messages and buttons let the user select an option directly, which tends to resolve intent in one exchange instead of two or three.
  • FAQ and knowledge base deflection. For genuinely common, simple questions (hours, location...), consider a lightweight self-serve option: a quick-reply menu or a link to a help article, so the answer doesn't require a full back-and-forth with an agent or AI.
A few things to check on your end
  • Review your journey and AI agent analytics for average message count per completed interaction. This is now a cost metric, not just a UX one, so it's worth tracking alongside resolution time and satisfaction scores.
  • Audit your most-used journeys and flows (onboarding, triage, common FAQs) first. That's where volume is highest and small improvements compound the most.

None of these changes require sacrificing quality of service. If anything, cutting unnecessary back-and-forth tends to make interactions feel faster and more respectful of your user's time.

If you'd like help, reach out to turn.io support and we can walk through it together.

Where can I learn more about these pricing changes?

You can read more about this in Meta’s announcement.

We understand this has an impact on your operations, and are here to help you minimise this impact. Please reach out to us if you have any questions, or need any support or advice on how to optimise your service.

Was this article helpful?

Meta Fee Waiver FAQs

Meta Billing FAQ

Contact