What I learned broadcasting a Democratic debate via text message
Everyday more media outlets are using bots and messaging apps to distribute their content and interact with their readers. Quartz launched in March an app that emulates a messaging interface and composes haikus about Wall Street; The New York Times addressed questions from readers about the presidential elections through Slack and The Guardian used WhatsApp to cover a republican debate.
At Univision News, we have also started experimenting with platforms and messaging environments with the purpose of interacting with our communities through mobile devices. We did our first test a little over a month ago with Purple, a startup that is covering the presidential campaign through text messages.
For that purpose, we decided to use the tool to cover an event that had its beginning and its end: the Democratic debate between Hillary Clinton and Bernie Sanders that Univision hosted in Miami.
This is how it worked: a user subscribed to the service through our website and started receiving text messages related to the debate. Each message contained one or several keywords written in uppercase that allowed users to navigate and choose how much they wanted to know about a tidbit of information. If an specific story wasn’t interesting, they could just pass and they wouldn’t receive more updates.
The founders of Purple, Rebecca Harris and David Heimann, helped us design and optimize an on boarding process that was attractive and made sense to our audience. The reader’s response was fascinating. In little bit more than a day and without barely promotion in social media, 250 people subscribed to the service. Each user sent us an average of 15 messages on the day of the debate.
Now, as the second phase of our project, we have decided to extend this test to one of our main bets for the electoral campaign: our political fact-checking platform The Lies Detector.
But first, let’s break down what we learned:
Predisposition to an informal tone, in both ends. The messaging environment invites users to express themselves in a more playful, close and informal way.
The conversation is in real time. In the days following the debate, we had very interesting exchanges with some of our users, with whom we could learn more about their doubts and interests about the electoral process. For example: Who can be a superdelegate? Is Obama superdelegate? Users expect an answer to their questions in real time, as it would happen when you’re asking something to a friend (in fact, we noticed a tone of impatience from some users.) This immediacy in the service can be difficult to scale, a problem that The New York Times also experienced with its Slack channel.
Rhythm. As a distribution channel, messaging has an unique intimacy component. With every message you send to your users, you are getting into the deepest inbox that exists in a phone, between a shopping list that you send to your partner and the last argument you had with your father over text. Even covering an event as agile as a debate, we prefer to space our messages as much as possible and send them in seven-minute intervals to avoid becoming spam.
Lack of references. Conversational information is a novelty for some of our users, who still do not have an expectation or clear guidelines on how they should interact with us or with the system. One of the main discoveries is that it is essential to guide users through the process, signaling what they’re expected to do at every step of the interaction.
New narratives. Telling a story through text messages is a new creative process. It is not about breaking an article into micro units of content, but about building an architecture with ramifications and levels of depth that allows users to choose several routes, as in a Choose your own adventure book. This process requires time and attention, resources that are not always abundant in the hectic dynamics of a digital newsroom.
What metrics should we use? It doesn’t matter to reach thousands of subscribers in a service like this if their behavior is passive and they don’t interact with your publication. I believe that the metrics of success in these experiments has more to do with the users retention capacity and with their frequency of interaction, rather than with the number of subscribers. Therefore, the metrics that we must use are much more similar to those used in the video game industry than the traditional ones in the media.
The promise of ‘narrowcasting’. As users choose the depth level they want to reach within a certain story, Purple’s content manager generates groups of people based on the different options they have chosen: a group for users who want to receive live updates, another for those who prefer to receive a summary at the end, etc. I think this gives us an interesting opportunity: to use this model to adapt the service to the preferences of the users and their niches of interest, instead of to a specific news event (a group for people who want to receive news by the tomorrow, another for those who prefer them in the afternoon, another for those interested in economics, another for the environment, etc.)
It’s been a long time since our smartphones stopped being just phones: they have become some kind of physical extension of our brain. The most active users interact with them more than 100 times a day and our dominant behavior is to send messages to our friends and contacts.
From our perspective, we will continue to exploit the potential and the narratives associated with this new medium. Soon, more about phase two of the experiment.