As a marketer, your goal is to drive engagement and conversions. Google's new spam filter for Chrome on Android directly impacts your ability to do that by hiding notifications it deems spammy. This guide describes how to protect your campaign performance and ensure your messages get seen.
How the Spam Filter Impacts Your Campaigns
When Chrome flags one of your notifications, your audience doesn't see your carefully crafted message. Instead, they see a warning prompt.
Step 1: The Warning Prompt
The user sees a generic warning with two options:
- Unsubscribe: A one-tap exit from your subscriber list.
- See Notification: The user must take an extra step to see your content.
Step 2: View the Notification
If the user chooses to see the notification, your message is displayed with two new options:
- Always allow: The user can whitelist your site, ensuring your future notifications are delivered normally without warnings.
- Unsubscribe: The option to opt out is still available.
This process gives users more control and makes it critical for marketers to send high-quality, relevant content.
Key Factors That Trigger the Filter
Our analysis shows the filter is primarily sensitive to three areas:
- Notification Content: The filter scrutinizes keywords, phrasing, and emojis. Clickbait, high-urgency words, and vague language are major red flags.
- Individual User History: If a user frequently dismisses or ignores notifications from your domain, future messages are more likely to be flagged for that specific user.
- Message Repetition: Repeatedly sending the same notification to a user is a strong signal to the filter to hide your content.
Best Practices to Avoid Being Flagged
To adapt, your strategy must focus on content quality and continuous testing.
-
Write Clear, Honest Content
- Avoid sensationalist language: Your goal is to provide value, not to trick users into clicking.
- Avoid Spammy Keywords: Avoid words like "FREE," "WIN," "URGENT," and "ACT NOW."
- Be Specific: A vague title like "Klean Beauty to Cart Delivery!" was consistently blocked in our tests, while "Your Klean Beauty order now qualifies for free delivery" worked perfectly.
-
Don't Impersonate System Alerts
- Never make your notifications look like official system or OS messages (for example, "Your system may be at risk"). Chrome is specifically trained to detect and penalize this.
-
Test Rigorously and Continuously
- Message Volatility: A message that works once is not guaranteed to work again.
- Test on Your Own Device: Before any campaign, send the notification to your Android test device multiple times. If it gets hidden even once, revise the content.
- Isolate Problematic Words: Our tests showed that removing a single word like "FREE" could get a notification unblocked. This highlights the filter's sensitivity.
-
Personalize and Vary Your Messaging
- Irrelevant Messaging: Generic blasts are irrelevant to a large portion of your users, which is why they become a liability.
- Avoid Repetition: Never send the exact same content to the same users repeatedly.
- Segment and Personalize: Use MoEngage to send relevant messages based on user behavior and preferences.
-
Monitor Your Campaign Analytics
- Observe your MoEngage UI closely. A sudden spike in unsubscribe rates or a drop in CTR can indicate that a campaign is being flagged.
The Key Takeaway: Evolve Your Strategy
This feature represents a fundamental shift in web push notifications. Remember that the machine learning model will constantly evolve. A strategy that works today might not work tomorrow.
Focus on quality, test continuously, and respect your users. By delivering genuine value, you can build trust and ensure your messages continue to land.
FAQs
Currently, this filter is active on Google Chrome for Android.
No. The filter is on-device, so no signal is sent back to the MoEngage SDK or your website.
Yes. An impression is counted on delivery, even if the notification is hidden behind the warning prompt.
The event is not tracked in real time. MoEngage can only identify the user as unsubscribed on their next visit to your website.
The model learns from individual user history with your domain. If many users unsubscribe or ignore your messages, it will become harder to reach your audience effectively.