Engagement, revenue, loyalty. These are concepts not normally associated with chatbots. For now, we tend to see Facebook Messenger bots as early in the development cycle, still trying to find a niche with users, and offering weather bots that tell us to wear a rain jacket.

The Msg.ai platform seeks to change that. Presenting at MobileBeat 2016 today, The CEO and founder Puneet Mehta revealed how the A.I. engine works within Messenger, other chat programs, and within apps. It’s used by marketers for larger brands as a way to engage with customers and, ultimately, lead to more revenue for that brand.

Now, the company has announced a major improvement to their core toolset. Known as multivariate testing, the new addition provides more than just an A/B test for a new bot. Developers can monitor user flows within the bot, test for bot tone (is it better to have a light and friendly tone or one that is more buttoned-up?), and experiment with rich media.

That last addition is critical for the success of bots. It lets developers test out videos, images, coupons and deals, GIFs, and other media inserted into the chat to see if those features in the bot lead to more customer loyalty and engagement. In terms of conversational commerce, it means the staid text chat with a bot becomes livelier and more interesting.

Because Msg.ai is more than just a platform for creating bots and has a backend dashboard for monitoring all user interactions, the multivariate testing means developers can find out more about how a specific user segment interacted with a specific variant. They can see how factors like the time of day, the day of the week, and even the weather factor into engagement with a variant. And the control center can provide specific suggestions.

It’s no secret that chatbots need this kind of backend, since there are now 11,000 of them in Facebook Messenger (although many are simple chatters for entertainment or getting sports scores).

“Conversational commerce is a groundbreaking opportunity for brand marketers, providing the most intimate and immediate channel for 1:1 consumer interaction. Being able to test various parts of the conversation, and access real-time insights into the performance into how different variants are resonating with consumers, will empower brands to fully take advantage of the potential of B2C chat,” said Mehta in the company press release.

Stay abreast of the latest news on bots, messaging, and A.I. from MobileBeat 2016. Read our coverage here.