Discover the insights from our latest white paper
In our latest research project, we challenged ourselves to create a series of strategies that can be been proven to increase the readership, and value, of push notifications for digital publishers.
The first step in this process was to identify a new metric to measure a real increase in readership. This goes beyond ‘vanity metrics’ in isolation, such as click-through rates, and creates a new measure of active, added readership created by push notifications. This added readership can then be distilled into tangible value through increased ARPU (Average Revenue Per User) for a publisher.
You can read more about how we defined Subscriber Lifetime Value here.
With our value metric established, we set about running a series of controlled experiments. These experiments compared the value of a strategy with a standard control group. This gave us a series of actions that can be applied to a publisher’s push notification strategy and are proven to add value.
From these experiments, we’ve compiled four best practices for publishers, aiming to increase their readership from message intensity.
Factor One: Test your message intensity
Every subscriber has a tipping point at which messages become too frequent and it drives them to unsubscribe. However, more messages will result in more clicks. The impact of seeing an increase in the unsubscribe rate has made publishers overly timid in regards to message frequency.
By measuring using SLTV, rather than just the number of clicks, we are able to isolate the exact frequency of messages that generated the highest increase in readership.
The experiments showed that the unsubscribe rate does increase inline with message frequency but that users were more resilient to receiving messages than most of our test cases first theorized.
Factor Two: There is an optimum delivery time for push notifications
We considered two questions in the next experiment. If a set time of day is more effective for all readers and; if readers more responsive to notifications sent based on their previous behavior?
We measured a series of set times for messages and then created a system in which messages were sent at the same time as a user read their previous message (Smart Delivery).
The results showed that notifications are the most valuable when people have the time to click the article at the time of the notification. For most people, this is after work, in free time. Smart Delivery then allows us to reach users that have other windows of available time.
Factor Three: Adding Images is mandatory
Any digital publisher running push notifications knows that adding an image to a push notification will improve their click-through rate. This led us to examine the exact value created by images and the impact of leaving images out .
Adding images, compared with messages with no image, results in a major increase in SLTV. Adding images is one of the most effective, single elements publishers can do to increase their readership and value from push notifications.
Factor Four: Personalize your content
Personalized content produces higher engagement. However, the volume of push notifications has meant this has been hard to implement beyond variables such as location.
In order to personalize based on reader behavior, our recommendation engine created a popularity score for all articles. The least engaged audience segments were sent the most popular articles. In addition, the engine would match the reader’s last-clicked article to a similar style of new content for their next push notification. Both factors enabled our publishers to increase SLTV.
These factors, plus the results of numerous other experiments demonstrate that an objective framework can be applied to optimize push notifications. The framework applied by MarfeelPush increases SLTV by up to 40%.
You can read the full report, including the data from over 5 million push notifications, in our latest white paper. It shows the way to measure absolute value from push notifications and implement a framework for optimized results.