17th Sept - Newsletter (ICP Q - Non India)

Hi There,

We’re all familiar with Duolingo’s green owl mascot, Duo, and its signature sass. Over the years, Duolingo, a popular language-learning app, has perfected its brand voice - witty, sarcastic, and always ready to guilt-trip you into practicing your Spanish. And they have achieved this by making their push notifications unignorable.

“These reminders don’t seem to be working. We’ll stop sending them for now.”

Push notifications can boost app engagement by 88%, given they are clicked in the first place. With an average mobile user receiving 46 notifications per day, it’s easy for your notification to slip under the radar. Yet, that’s not the case for Duolingo because they got people talking.

And they did it by making their push notifications unique to each user.

The AI algorithm behind the meme

Duolingo utilizes Bandit Algorithm that repeatedly chooses between the same pre-written content sets, and gradually learns from the user’s past decisions. The adaptive algorithm learns from user behavior - language being learned, app usage frequency, notifications interaction rates, etc.

To train their algorithm they collected and analyzed data by sending 200 Million notifications within 34 days. The findings are as follows:

• Notifications aligned with the language a user is learning perform better.

• Brand new notifications are more convincing.

• But once the novelty effect wears off, users would need a new reminder.

• To solve this, they demoted reminders that were seen recently to limit the algorithm from resending the best-performing notifications often.

The domino effect of engaging notifications

To set themselves apart from the pools of daily notifications, many brands are switching to a humorous approach. Zomato, Swiggy, Netflix, and Tinder, are some of the examples that are known to send notifications with higher engagement.

Apart from being quirky, their strategy includes a combination of practices to ensure no notification goes unnoticed.

1. Segmentation-based Personalization: Instead of sending a common notification broadcast to all users, segmenting them based on their behavior, needs, and preferences allows for a more personalized experience.

2. Engagement-based Personalization: LinkedIn sends targeted notifications for new content from creators or connections based on users’ past engagement. Similarly, personalizing notifications for each user becomes critical as you scale.

3. Clear Objective: Every notification should have a purpose. Be it conversion, engagement, or retention, make sure your notification aligns with the objective for better impact.

4. Smart Copy: It’s a no-brainer that push notifications have to be highly creative and concise. Leveraging proven marketing tactics like creating scarcity and urgency can increase CTRs.

5. In-app Notification Feed: Pushing every notification to users leads to brand fatigue. A smarter idea is to create an aggregated notification feed to list all the updates sequentially and push only the priority ones. 

We understand that building a reliable notification system is hard. The entire process requires significant dev efforts and iterations with potential scalable issues. This is why our Chat and Feed SDKs come loaded with all the necessary advanced features, like personalization, notification feed, and more, to reduce your work truly.

About LikeMinds

We enable businesses to deploy feature-rich Chat and Feed into their app within minutes. Our low-code SDK and pre-built UI truly reduce the dev effort and let your team focus on what really matters - building a superior experience for your customers. 

Until Next Discussion,

Nipun Goyal,
Founder & CEO, LikeMinds