RCS Agent Marked as Spam by Google RBM – Best Practices

Overview

Google RBM (Rich Business Messaging) continuously monitors RCS agents for spam signals and policy violations. If an agent accumulates negative signals, it may face reduced reputation, message throttling, or in severe cases, suspension. This document explains why an RCS agent may be marked as spam, how reputation works, and what measures can be taken to improve and maintain agent health, in simple, non-technical terms.


How Google RBM Detects Spam

Google RBM does not provide real-time reputation scores or dashboards to solution providers or customers. Instead, reputation is inferred indirectly based on user behaviour and content patterns over time.

The key indicators RBM uses include:

1. User Opt-out Signals (Most Critical)

RBM closely tracks how users react to messages.

Negative signals include:

  • Users clicking "Unsubscribe"

  • Users replying with "STOP" or similar keywords

These events indicate that recipients find the messages unwanted or spammy.

Important: While real-time reputation is not available, unsubscribe and STOP events pushed to your webhook are the closest indicators of agent health.

A rising number of these events significantly increases the probability of an agent being penalized.


2. Content Relevance & Use Case Compliance

RBM validates whether the message content aligns with:

  • The agent’s registered business identity

  • The approved use case (Promotional, Transactional etc.)

High-risk scenarios include:

  • Sending promotional content using a transactional agent

  • Content unrelated to the business owning the agent

  • Misleading, deceptive, or overly aggressive marketing language

Such violations rapidly degrade reputation and may lead to suspension.


3. Message Volume vs Engagement Quality

RBM evaluates quality over quantity.

Sending a large volume of messages is acceptable only if:

  • Opt-out rate remains extremely low

  • Content stays relevant and expected by users

High volume combined with poor engagement accelerates spam classification.


Reputation Levels – Practical Understanding

From operational experience:

  • Agents with high reputation typically send up to 8 messages per user in a rolling 28-day window

  • Such agents maintain an average spam rate of ~0.01% annually

What does 0.01% spam rate mean?

  • Out of 10,000 RCS messages delivered

  • Only 1 user unsubscribes or replies STOP

This level of message quality is what RBM expects before considering reputation improvement.


Common Reasons for Agent Being Marked as Spam

  1. High number of STOP / Unsubscribe events

  2. Irrelevant or misleading content

  3. Breach of approved use case (Promo vs Transactional)

  4. Excessive message frequency

  5. Messaging users without clear opt-in

  6. Sudden spike in traffic from a new agent


How to Improve & Maintain RCS Agent Reputation

1. Monitor Opt-out Events Closely

  • Actively track STOP and Unsubscribe callbacks from web-hook

  • Identify agents with rising opt-out trends early

  • Pause or reduce traffic before penalties occur


2. Send Only Relevant, Expected Content

  • Ensure all messages align with the agent’s business purpose

  • Avoid generic marketing blasts

  • Personalize messages wherever possible


3. Respect Use Case Boundaries

  • Transactional agents → Only transactional content

  • Promotional agents → Only promotional content

  • Do not mix use cases under the same agent


4. Control Message Frequency

  • Avoid excessive messaging within short durations

  • Follow the recommended limits (e.g., ≤ 8 messages per user in 28 days)


5. Maintain Clean Opt-in Practices

  • Message only users who have explicitly opted in

  • Clearly communicate why users are receiving messages

  • Provide easy and visible opt-out options


6. Gradual Traffic Ramp-up for New Agents

  • Start with low volumes

  • Increase traffic gradually as engagement stabilizes

  • Avoid sudden spikes in message submissions.


Is Real-Time Reputation Tracking Available?

❌ No.

Google RBM does not provide real-time reputation scores or APIs to track agent reputation.

✅ What is available:

  • STOP / Unsubscribe webhook events

  • Delivery and engagement trends over time

These should be treated as early warning signals.


Key Takeaways

  • Reputation is built slowly and lost quickly

  • User opt-outs are the strongest negative signal

  • Relevant content and low spam rate are the only proven ways to improve reputation

  • Prevention is far easier than recovery


Need Help?

If your RCS agent shows signs of degradation or suspension:

  • Share agent/bot name and recent traffic details

  • Provide timestamps of failures or penalties

  • MSG91 Support can help with analysis and escalation where applicable


Prepared by: MSG91 Support Team