Last update: June 2021

6 mins to read - 2021/04/09

OpenBack DeviceDecisions Signal Results in +39% CTR for Notifications

In our Newsroom, we are currently releasing a series of posts mapping out the new OpenBack dashboards, metrics, and ever-growing signals capabilities. As we build our on-site archive of resources for customers to learn about these different features and their use cases – and examples of ways to leverage them towards maximizing your return on your push notifications – we have also been collecting data on the results these unique tools have on click-through rate. We have always been confident that using signals to determine the perfect moment of delivery for sending notifications will boost click-through rate (CTR) in a big way. However, following anonymized data analysis from billions of data points, we are now sharing results that show this is absolutely the case.

Download our Mobile Marketing Playbook to perfect your user engagement game!

other apps signal

What Are Data Signals?

OpenBack offers the potential of 52 different signals – some are called “triggers” by other vendors, if they offer them – and they give apps real-time contextual factors to use to improve the impact and user experience of push notifications and in-app messaging. These are effectively data flags that alert the system of when a certain event or circumstance has come to pass on the device-side. This could be environmental – such as the strength of a Wi-Fi signal, or the noise level in the surrounding area. Or it could be connected to the device – such as its battery level, whether its screen is locked, or whether its headphone jack is engaged. 

The DeviceDecisions Machine-Learning Signal in particular combines insights on a user’s daily and weekly schedule (e.g. through their daily habitual phone usage, etc.) with moment-by-moment user behavior, and interprets them through an algorithm to predict in real-time when a user is likely to have the attention span to interact with a push notification. It’s just as important to determine the wrong moment, so you can avoid it, as it is to determine the right moment to send a notification. 

Otherwise, there is also a set of Real-Time User Segment type signals. These focus more on mapping out a profile of your user, their use of your app, and their behaviour in relation to your app. This helps developers determine what notifications should and shouldn’t be sent to app users. 

For more information about OpenBack’s custom data signals, visit our Documentation page.

What Do We Mean by “Moment of Delivery”?

We use the term “Moment of Delivery” a lot, especially when talking about signals. By this, we mean the moment when it’s most convenient for the user to receive a push notification. (Or, put another way, NOT in a moment which ISN’T INCONVENIENT for the user.) When they’re not distracted by anything else, and most likely to click on a message – and perhaps interact further with your app. If you hit the nail on the head with Moment of Delivery, this can be one of your biggest drivers of successful push notification campaigns.

The perfect moment can differ from user to user. Different apps and app categories can get different benefits from various content and signals. This is why it’s crucial to leverage user data on a device by device basis. For example, for users who work 9 to 5 jobs, certain hours can be designated “no-send” periods for push notifications. If there are long periods of device non-activity during the day – when your user is at work, commuting, or asleep – these are times when a user will see a notification as obtrusive, and they will be less likely to engage with it.

On the other hand, if there’s a time in the evening – say, 7 to 9 o’clock – when your user is relaxing, maybe on the couch in front of the TV, and scrolling through their device, that might be the opportune moment. If you send a notification at this moment, you can be fairly certain it will grab their attention. 

signal ctr moment of delivery

How Do You Use Signals?

Although not every signal needs to be used for every scenario, combining two, three or more signals can give you a more complete understanding of the context on the user’s side. Sort of like data analysis echolocation for determining the perfect moment of delivery. 

For example, if a user’s device is unlocked, that tells you they’re actively using their phone. If their headphone jack is engaged, this may be a good moment to send them a notification with rich media, such as video content. However, if their device has low battery power, the user may not appreciate receiving a 5-minute video that will further drain their device. Such a thoughtless notification might drive them closer to opting out of push notifications… Or even deleting your app. Ultimately, all notifications should have the user’s benefit in mind. And the more you employ this ethos to your push campaigns, the better your relationship with users will be. 

That’s the theory, anyway. Now for the real-world results…

Signal Insights

We selected 17 of the key signals OpenBack offers, to really hone in on the effect they had on interaction rates (meaning when a user either clicked or dismissed a message), and then on both clicks or dismissals individually. For the purpose of these analytics, we are taking dismissals to be positive. Even though the user isn’t entering your app, they are likely still reading and gaining value from your message.

This analysis doesn’t take into account what sort of mobile app, app category, user segments, message content or language, or other means of personalization. This is solely a measurement of the changes in rates according to different signals being used – or not used – to determine the Moment of Delivery.

Conventional Push Delivery (Now Signal)

This signal resulted in immediate delivery of the message, like every other platform offers. As we would have expected, this resulted in far lower interaction rates, with a 63% drop in overall interaction and a 111% drop in click-through rate (CTR).

Time Signal

Interestingly, the signal that lays out a window of time during the day for notification delivery also resulted in lower interaction rates: 18% lower interaction rates, and 54% lower CTR. It seems that this signal isn’t very useful for predicting a good moment for users to engage with your app. Which stands to reason, as time is just one component of a good moment for a user… And it’s impossible to accurately predict when a server is sending out a conventional push notification.

Unlock Signal

This signal sends a notification when the user unlocks their device screen. It saw strong gains in interaction, with 6% boost in overall interaction rates and 12% jump in CTR. It seems that, if the user already has the spare time to go on their device, and they have it in front of their face, they are more likely to click on a notification that pops up on their screen.

App Open Signal

This signal schedules notifications to deliver only when the user is in your app already. It resulted in a 24% boost in CTR, although a 15% drop in overall interaction rate and 26% drop in dismissal rate. This might imply that, once a user is already devoting their attention to your app, they’re more likely to engage with content that doesn’t require them to switch tracks. As such, this signal is a great way to engage, educate and tell your users about something relevant.

Source: Anonymized legacy data sample where logs polled back to OBE (OpenBack Engine, backend platform)

Battery Signal

This signal checks up on the device’s battery life. Meaning developers can avoid sending notifications when a user’s device is nearly dead. This saw positive results across the board: 25% interaction rate, 26% dismissal rate, and 22% CTR.

Headphones Signal

The headphones signal likewise saw good results, with a 44% increase in CTR. (Although a 58% decrease in dismissals, and a 27% decrease in overall interaction.) It seems users that are already watching a video or listening to music on their phone might be more inclined to click a notification. However, if they don’t hear the beep because they’re listening to something else, that might account for lower interaction and dismissal rates. 

Auto Remove Setting

This setting assigns notifications an “expiry date.” So if they go for a set period of time without the user interacting with them, they are removed. This resulted in a 48% drop in dismissals. This equates to less users being potentially interrupted, unless notification content provides value to the user. That said, there was a big drop in CTR (279%). But this is explained by the fact that many campaigns use Auto Remove in churn prevention and re-engagement campaigns. These scenarios often also include recurring delivery settings. So users will end up getting the same message many times.

DeviceDecisions Signal

The DeviceDecisions signal combines user data with an intelligent algorithm to take real-time user behaviors and changes into account. Use of this signal saw a 39% jump in CTR, and a 76% drop in dismissals. This is likely due to the signal’s effectiveness at gauging the best Moment of Delivery for users. Even higher rates of CTR can be achieved when this signal is combined with personalization tactics, engaging content, and rich media.

Source: Anonymized legacy data sample where logs polled back to OBE (OpenBack Engine, backend platform)

Deep Dive Into Location

We paid particular attention to location-based signals, which resulted in increases in interaction rates, dismissals, and CTR across the board. While these weren’t necessarily as influential in determining the Moment of Delivery, they showed an important aspect of personalization that developers should take into account when designing push campaigns.

Language & Locale

Customizing notification language and content according to a user’s region may help boost interaction rates. Engaging this signal saw a 28% jump in CTR and 15% boost in overall interaction rate.


Different from the Locale signal, Location focuses on either:

1) specific geolocations: a user’s country and geofencing

2) a machine learning (ML) algorithm – Places. This algorithm learns for that user, on the device. Using what it learns, it makes a prediction whether the user is likely to be home or at work at that moment. The algorithm requires no geolocation data. Although if the data is already available, it does improve the ML model and prediction. This signal saw a 13% increase in CTR, and a 14% increase in overall interaction rate. We further explored the results for this signal, on a rolling 7-day basis.

The Places ML for home/work signal saw a 171% increase in CTR and 46% decrease in dismissals when notifications were sent at times when users were predicted to be at home. When the ML predicted the user to be at work, it blocked those hours out as a “no send” time. As a result, CTR jumped by 215%, and dismissals fell by 41%. (Of course, this analysis was carried out during the COVID-19 pandemic. Many users may have been working from home, which somewhat blends the results… This suggests it’s not a matter of where the user is, but a matter of what they’re trying to focus on doing.)

Country Signal

This signal uses geolocation/GPS data, IP address, and device locale settings to determine what country the device is located in. This can then inform user segmentation based on notification language and content. This saw an increase in over 17% across the board for CTR, interactions, and dismissals. 

Interestingly, using the Exact Location signal to time a notification to deliver according to a user’s exact coordinates saw a 23% decrease in CTR. It also saw a 9% decrease in overall interaction rate. This may be due to the fact that user opt-ins for allowing an app access to their exact GPS location is typically low. This may skew the data somewhat. 

Source: Anonymized legacy data sample where logs polled back to OBE (OpenBack Engine, backend platform)


Ultimately, only a handful of signals will be really useful for maximizing your app’s CTR. However, once you are certain of which signals achieve the best results for your app, there will be many different combinations you can apply for optimal user engagement. Our customer success team is always on hand to advise you on which signals will best complement your push campaign.

Certain very basic signals (such as Conventional Push Delivery) are available with many push platforms. (OneSignal does have more than 1 signal, despite their name.) But many of OpenBack’s 52 signals are unique to OpenBack. They are only possible given OpenBack’s patented device-side decisions engine that operates and controls notification delivery from the device itself. These include:

  • DeviceDecisions Machine Learning Signal
  • Headphones
  • Battery Life
  • And many more

OpenBack’s range of signals offer an enormous new potential for truly personalized user engagement. New data and results are still being aggregated, so keep an eye on our Edge Blog and Newsroom for new insights. This is a very exciting time, and we foresee our findings taking the mobile engagement industry in a completely new direction.

For more information about how to use OpenBack’s signals to boost your CTR and return on push, reach out to one of our experts with any question you may have.

Leave a Reply

Your email address will not be published.

5 × three =


Translate »