What’s a good retention metric for bots?

At my company Dashbot, we’ve processed nearly 2 billion messages in the past 11 months. It took 10 months to process the first billion, and we’re on pace to process the second billion in about six weeks.

We took a look at the retention data across bots on our platform.

We used the same definition Localytics used when reporting on Mobile App Retention. For 30-day retention, we counted users who come back at any point within the 30 days after their initial visit. We followed the same approach for 60-day (users who come back between days 31 and 60) and 90-day (users who come back between days 61 and 90) retention.

Interestingly enough, chatbot 30-day retention is on par with mobile apps. Localytics showed a 36 percent 30-day retention rate for apps, whereas we saw an average 34 percent 30-day retention for chatbots across platforms.

Popular categories

We can dive deeper into categories on Facebook and see retention for specific use cases. The most popular categories, based on one-month retention rates, are Games, Social, Local, Health and Fitness, and Shopping.

Popular topics

Call it a social faux pas  —  we often hear you shouldn’t talk about sex, religion, or politics in social settings. Well, how about in chatbots? While politics has fairly low engagement, both religion and dating chatbots have decent 30-day retention, although they drop off significantly afterwards. Perhaps the users found what they were looking for.

Speaking of sex, what about drugs and rock and roll? While the overall number of drug-related or music chatbots are a bit low to be statistically significant, below are the early signals.

The lesson? Make sure you do a deep dive into bot analytics — it can help you capture and retain an audience.

Arte Merritt is the CEO and cofounder of Dashbot, a bot analytics platform for Google Home, Alexa, Facebook, Slack, Kik, and other conversational interfaces.