26 March 2021

How BBC World Service fights the coronavirus infodemic with chat bots

Jan Bruck
Digital Strategist, BBC World Service Digital Development

When the world stopped in its tracks about a year ago we were faced with a significant challenge: How can we improve the impact of our content in the middle of a pandemic?

We needed to find new ways of bringing our journalism to our audiences, aligning with their changing news habits. The pandemic also completely changed how most of us work. Many journalists had to adapt rapidly to being based at home while at the same time keeping our daily news reporting up and running.

On top of this, the fear of the new virus and the many unknowns gave rise to a plethora of disinformation – which in turn made our audiences sceptical of coronavirus information. In an international study by PR firm Edelman at the time, over a third of respondents said they didn’t trust coronavirus information if the only place they saw it was social media. 74% said they were concerned about fake news.

As we now know, the challenges of fake news and mistrust have only got worse in the pandemic, with the WHO branding it an "infodemic", and offering its own chat bots on WhatsApp.

Screenshot of the Vietnamese coronavirus chat bot on a Messenger

The Vietnamese bot's welcome message with three choices: top stories, explainer videos and uplifting stories.

At the World Service central social media team our job is to provide strategic advice and support to our 42 language teams around the globe. For many of them, social media is the most important gateway to their online content. We offer digital solutions to journalists who are extremely busy with daily news operations, while keeping in mind their local market, the type of content they have to offer and the resources they have at their disposal.

The pandemic meant an even bigger strain on resources than in normal times. So how could we execute our goal to offer a lifeline service with essential, reliable coronavirus information in local languages? Facebook Messenger chat bots seemed like the perfect answer for several reasons:

  • Journalists would be able to update them from home without too much time required
  • BBC News Labs had bot solutions ready to use
  • According to Facebook, activity on their Messenger was surging at the beginning of the pandemic

Why Messenger?

With so many chat apps in use, Messenger was still the obvious choice. All of our teams already have Facebook news pages which come with an inbox by default. It is also the second biggest chat app worldwide with 1.3 billion users, compared to WhatsApp’s 2 billion (We Are Social). Market penetration varies widely by region. We wanted to make sure to hit markets where Messenger is the leader:

  • Vietnam: 52 million monthly active users which equals a market penetration of 76%
  • Bangladesh: 25 million users which equals a market penetration of 53% (We Are Social)

Screenshot of the Bangla coronavirus chat bot on a Messenger screen

Embedded video explainers from BBC World Service in Bangla helped increase engagement with the bot in the fight against the infodemic.

The bots wouldn’t only help us beat disinformation with minimal resources but also achieve a number of strategic social media goals:

  1. Make existing Facebook followers of the BBC Vietnamese and BBC Bangla pages more loyal to BBC content
  2. Improve the user experience when people message these pages
  3. Make them engage with the bot multiple times
  4. Encourage users to subscribe to automatic updates from the bot

In Vietnam, we face the added challenge of tough restrictions on internet freedom and freedom of expression. Finding alternative distribution methods is essential in such markets. Facebook Messenger offered a way to circumvent these restrictions to some degree.

A new perspective on a long running story

Apart from the newest information on infection numbers and lockdown measures, we wanted to deliver added value and content that people might have missed on their Facebook feed or the BBC website. The bots offered explainer videos (on anything from coronavirus symptoms to the effect of the pandemic on tourism) and uplifting stories about doctors, nurses or everyday people fighting the virus.

In two weeks, we were able to build out the conversation path of the bot, using a third-party tool called Dexter along with News Labs’ expertise and dedicated journalists from World Service teams. The BBC Vietnamese and BBC Bangla teams were then set up to update the bot daily with fresh content at peak traffic times.

For BBC Bangla, we first launched a beta version of the bot to see how users would interact with it. After three weeks, we added a subscription feature to allow automatic push notifications. A third of users who interacted with the BBC Bangla bot chose to get those regular notifications.

Overall, the two bots sent and received nearly one million messages in Bangladesh and Vietnam. Engagement levels were higher than we had anticipated with about three to four queries per user on average. Users didn’t bounce off but came back for more after making their first choices.

What did we learn from our bot experiments?

My three key takeaways:

  1. There is an untapped potential when it comes to people wanting to subscribe to automatic updates from news publishers on Facebook for topics that matter to them (Facebook only supports a certain number of languages for these subscriptions, so do check if it is available in your language)
  2. Our inboxes on Facebook are underutilised spaces that are ripe for smart solutions like conversational bots. A lot of social teams only use this for automated replies redirecting users to email at the moment - a real missed opportunity for meaningful engagement
  3. We learned that discovery is a challenge with Messenger bots: How are users going to find out that this service is available to them? The answer is promotion, promotion, promotion - and smart calls-to-action, preferably on your Facebook feeds and Facebook Stories.


We are examining where to take this project next, taking into consideration local markets, what content we have to offer and whether a bot is the right fit for a particular team. In autumn of last year, BBC Mundo (our service in Spanish for North and South America) and BBC News Brasil launched US election bots using the coronavirus blueprints.

Most importantly, we’ve seen that there is an appetite for conversational information: Useful public service journalism, delivered in a fast and light-touch way, relevant to local markets and to people’s behaviours and choices during this pandemic.

Jan Bruck is a Digital Strategist in the BBC World Service Digital Development team which supports 42 Language teams with digital strategy, growth and best practice.


Love data and code?

We're looking for talented developers to join our team.