Unlocking chatbots' full potential starts and ends with data, according to Lauren Kunze, CEO of Pandorabots, a platform for building and deploying chatbots. Analyzing chatbot data helps reveal shortcomings of the bot that humans need to go in and correct to make the bot smarter, she says. In other words, chatbot initiatives are all about continuous improvement that will ultimately boost the company's bottom line, Kunze stressed. In this video interview shot at the Chatbots & Virtual Assistants for the Enterprise event in San Francisco, she discusses how analyzing chatbot data and measuring specific KPIs can help elevate a chatbot's ability as well as its ROI.
How do organizations best leverage the data generated via chatbots?
Lauren Kunze: Just in terms of talking about KPIs and ROI, the most valuable thing that an organization can get today is the data. A lot of people have had a negative experience with a chatbot, feeling that it's very limited, kind of dumb and doesn't understand what you're saying. But you actually need that end-user input data in order to make the chatbot system better. It's an ongoing process. The data that you're collecting -- based on how people want to transact with your business and your services -- is critical for training future AI and building smarter systems.
How do people best leverage it? Basically using traditional analytics methods where you're looking at everything. Some platforms, like our Pandorabots platform, show you where the chatbot failed and where it didn't have the correct answer; they're very development-focused. A human can come in and write the answer and the bot can get smarter. They also can look at trends, funnels and retention. There are a lot of great platforms that are focused exclusively on analytics, and I think that's a market unto itself in leveraging chatbot data.
How do you measure the success of a chatbot initiative?
Kunze: At the end of the day, the bottom line is: Can the chatbot application save you money or make you money? What are the real business results? In terms of KPIs that we're measuring today, we're looking at relatives across a number of different channels -- your existing funnels, how the bot is performing on web versus mobile, the timelines and development costs for building mobile applications, and the engagement rates there.
Again, on the service side it's usually about saving money; have we led to actual reductions in terms of the load on human agents and actual savings for that organization? On the marketing, brand engagement and brand awareness side, KPIs are usually about user retention. [It's about] driving direct sales, click-through rates, all the standard metrics that you would expect in other channels.