Live Segment Notifications

Alerting BBC audiences to live content about topics they're interested in


Can we combine one of the BBC’s greatest assets, world leading live output from linear broadcasting, with a modern digital approach focusing on personalised experiences and mobile devices?

Audience viewing habits for the BBC and others, are moving away from televisions and increasingly towards phones and laptops. This has led to an increase in the popularity of quick, customised, easily consumable content.

We wanted to investigate if we could alert our audiences to such output, live as it is broadcast as well as highlights and moments from previous output. This would mean we could bring relevant content directly to our audience rather than relying on them having to discover it for themselves. The hope is that we could make the most of the BBC's ability to deliver high quality live broadcasting but in a more contemporary way.

What are Live Segment Notifications?

In BBC News, TV & radio, programmes are planned and organised using running orders. These running orders are broken down into stories that correlate to sections of a given programme. This defined structure is particularly clear in programmes such as Newsnight or The Andrew Marr Show. For example, an interview with the health secretary about the NHS.

Live Segment Notifications would involve a member of the production team marking a story as high interest and therefore suitable for a notification when creating a running order. These stories would be picked up by the prototype and a notification to the audience could be scheduled and delivered. A further development would allow individuals to subscribe to topics they have an interest in and would want to be notified about.

How would notifications be delivered?

There are a range of delivery methods available to send notifications to an audience. For the initial prototype they were sent to a BBC Twitter account.

Screenshot of a tweet from the live segment notification Twitter account. Shows a programme description and a Twitter card that links to the programme segment.

Serving notifications via Twitter would allow users to click through and watch the specific story mentioned in the tweet.

The notifications could contain information about the piece of content and a link to view the story. The point in time the user clicks the link would determine what they are shown. If clicked before the story is broadcast, the user would be taken to a live stream of the programme and could watch the story from there. However, if they click the notification after the story has aired, they will be shown a catch-up player starting from the story in question.

Where are we up to?

Currently we have an early prototype to test this concept. Running orders are compiled and stored, notifications can be managed and sent to Twitter automatically or through an internal dashboard, and users can be taken to a webpage where they will be shown the story content in a media player.

However, there is still plenty that we would like to do:

  • Improving timing data: Running orders are complex and can be updated at any stage, which can affect the timing of stories within a programme. The calculation of story in and out cues is something that is still in progress.
  • Personalisation: Topic extraction and tagging of stories would allow audiences to tailor the notifications they were receiving to their interests. It would also allow editorial to get their content in front of interested people more easily.
  • Notification delivery methods: Alternative delivery methods such as SNS push notifications or Twitter direct messages would allow us to bring information directly to users. This is something we know is possible and want to develop further.
  • Notification content: Currently the content of notifications is generated from data in the running order. As there have been no editorial workflow changes at this point in the project, the programme titles and descriptions can be unsuitable or unreadable in some cases. This is something we'd therefore like to improve, either by developing an improved workflow with editorial or finding an alternative way to get those details.


  • We have successfully shown how we can deliver a notification to a user and take them to the relevant content.
  • We still have plenty of ideas around how we could improve the experience and make it more personalised to users.


Love data and code?

We'd like to hear from you.