Notification prioritization based on user responses

A method comprising:

• determining, for a notification, a first score indicative of a first relevance of the notification to a first user at a first client device, the first score determined based on at least metadata characterizing the notification;
• prioritizing the notification for the first user based on at least the first score;
• presenting, based on at least the prioritization for the first user, the notification at the first client device;
• and
• determining, for the notification, a second score indicative of a second relevance of the notification to a second user at a second client device, the second score determined based on at least a response to the notification from the first client device.

A non-transitory computer-readable storage medium including program code, which when executed by at least one data processor results in operations comprising:

• determining, for a notification, a first score indicative of whether a first user at a first client device is likely to respond to the notification, the first score determined based on at least a machine learning model;
• enabling, in response to the first score exceeding a first threshold, the notification to be presented at the first client device associated with the first user;
• and
• training, in response to receiving a user response to the notification, the machine learning model.