We launched the FactorDaily news-bot exactly a month from now. It was our first beta – a tentative first step into the brave new world of bots and conversational news. In this short post, I am sharing what we learnt in the first month of the bot’s interactions.
Users are very curious about newsbots
We averaged around 200 sessions over this one month. A session is a single interaction with the bot, where a user asks for news, gets the news and then closes the conversation. While this is a low number, in light of the facts that we do not use any push notifications, our conversations are limited to a small span of stories published from FactorDaily, and our technology is still in an early beta stage – we think this is a good indicator of the curiosity and readiness for users to consume news over chat.
Users want news, stories…conversations
When we launched the bot, we assumed that people would use the bot to search for keywords, trends and we built tech around that. We were wrong. Based on our analysis of how users interacted with the bot, the intent in most cases was to engage the bot in a conversation about latest news or stories. This is very powerful, because if we can engage users into deeper conversations around news, we can get a better understanding of what their likes and dislikes are – which in turn will help us fetch more personalised news and stories for them. We are now working towards that.
Users choose to type rather than click
In our first version of the bot, we had buttons for every choice that the user needed to make. Based on analytics, we quickly realised that users are choosing to type the choices instead of clicking option buttons. This is an interesting insight into the expectations of users interacting with messenger bots – where majority of the interaction happens on mobile.
The next version of our bot is coming soon. We used this month to learn a lot about how users interact with the bot, what they want and how we can serve them better. We have also invested a lot of energy in building a framework that can be used across various chat platforms, sharing the same news infrastructure, personalising a user’s experience and maintaining their context, even when they move from one chat platform to another.
The biggest learning for us in this one month, has been that bots by themselves cannot make much difference. They have to be backed by two things –
- A conversation experience, which doesn’t need to be necessarily powered by AI, but definitely needs to be thought through and designed in such a way that it gives a good chat experience to the user.
- An infrastructure that is tailored towards getting the content or the service for which the bot is developed. In our case, this is new and stories which need to be converted into a experience stream that chatbots can serve to users.
A lot of this will become clearer as we move from our 1.0 beta to our next release.