This report was created by a team at The Associated Press, with support from the 2019-2020 Reynolds Journalism Institute institutional fellowship. The AP team created this guide as they sought to improve their own push alert strategy, documenting what they learned along the way to share with others in the news industry. The report draws from analysis of their own push alerts, research studies, surveys they conducted and a consumer focus group they commissioned.
The intention of this report is to provide some answers to the two main questions of push alert strategy for news publishers: 1) What do users want from push alerts? 2) What is the value of push alerts to publishers?
A huge explosion had rocked Beirut. Information was scant, but it was clear that a major story was developing. The cause of the blast was unknown. A military or a terrorist attack couldn’t be ruled out. Surely, many people had died.
At 11:30 a.m. ET on Aug. 4, 2020, AP sent a push notification to our mobile app users with what little we knew at the time:
<aside> 🚨 A massive explosion has shaken Lebanon's capital Beirut, wounding a number of people and causing widespread damage.
</aside>
For a news organization whose bread and butter is breaking news, the decision to send this push was a no-brainer. But the discussions regarding subsequent push alerts on the same story brought up many of the issues that news outlets face when delivering news via lock screen.
Nearly two hours after the first push alert, AP sent an update to our wire service customers, quoting a Lebanese Red Cross official who said there were hundreds of casualties, including dead and wounded. Should we send another alert to our app users?
Some staffers believed that the first official word confirming casualties was a significant enough development to warrant a second push alert to app users. Others thought we should wait until we had a preliminary count of dead and wounded which, they argued, could happen within minutes. We still had no information about the cause of the explosion — or even what had exploded. By this time, video of a mushroom cloud rising over Beirut was all over TV and social media. Was it worth alerting something expected — that people had died — before we had more specific updates?
Ultimately, we decided to send a second push confirming casualties at 1:20 p.m. ET:
<aside> 🚨 Lebanese Red Cross official says hundreds of casualties in Beirut explosion, including dead and wounded.
</aside>
Those that argued that casualty numbers would come quickly were wrong. It was another hour and a half before that word came. When it did, we sent our third and final push alert about the Beirut explosion that day at 2:50 pm ET:
<aside> 🚨 Lebanese health minister says more than 25 people dead, over 2,500 injured in Beirut explosion.
</aside>
Of the three Beirut push alerts, the first alert was clicked on the most by app users. The second, hotly debated alert got 30% fewer clicks than the first. And the third one, with the actual casualty figures, got the fewest clicks — over 50% lower than the original alert.