Key learnings from managing bot babies
Chatbots, voice assistants, home devices, machine learning. It was a whole new world having come from managing your classic website / mobile applications, and an important part of my learning curve as a product manager.
A common use case for chatbots is to provide customer service support or triaging, but I personally can’t align with that thinking. It’s efficient for easy to answer customer queries, but the investment to setup correctly is significant. Which brings me to my first learning, you can’t skimp out on virtual assistant technology. Unlike a mobile application, you can’t leave the bot stagnant while you focus on launching a new product. Failing to monitor its responses could leave your business in a compromised position. It’s an extension of your workforce and as such, requires on going training. That means, you either need a team to manually input these learnings and correct the bot on an ongoing basis, or you have to invest in multiple bots to train the external facing bot through machine learning. The preference being the latter, while both options don’t come cheap.
Since it’s not a cheap exercise, is it really the best use case to try to replicate human support when customers need it most? My personal view is, it isn’t. But that’s not to say there isn’t use cases where bots can add value and a return on investment.
Going back to the basics, consider where can bots add customer value. If you look at implementation of Alexa or Google Home devices, they add value by reducing the effort you, as a customer, need to undertake to get what you need done. Now that’s efficient. Your website is a matrix of information that’s difficult to find? Yep, a bot can definitely help out there. Booking your whole trip for you and keeping the itinerary on hand? For sure, I wish a bot existed to do this today.
The second biggest learning for me was the lack of easily accessible data. You need to invest in a data scientist or have the right tools established to leverage a data warehouse, so you can actually see insights and learnings. As a PM, I pride myself on making decisions based on some supporting evidence or experimental testing. Having hundreds or thousands of raw transcripts coming through without being able to aggregate at a higher level, was literally like flying blind. Sure you can sample size, but there’s key parts of the picture you’re missing. The second biggest learning ties back to the first you see.. it requires more investment to yield results.
The final learning is my favourite. Save the best ’til last! Bots and machine learning will drastically change how we interact with technology in future. Asking customers to complete forms on a website just won’t do, waiting around to make an appointment on the phone won’t cut it, there will be bots to do all these things and more. If you aren’t investing in this type of technology in your business then you’ll be left behind.
The best learning for me personally was how much it opened my mindset to the ways machines can help customers in big ways through automation. Reminding me to think of the bigger picture when there’s a job to be done that isn’t limited to re-engineering an experience based on how it‘s done today. Can’t wait to apply these learnings to my new products! ;)
