Chatbots are everywhere, and I’m a big fan of the constant improvements I see in UI, features, and more from the plethora of chatbot developers and platforms. But there’s just one thing that I think needs a little more attention with chatbots that could really help their utility and their potential to be used more often: a better memory.
In this case, I’m going to use the example of a pizza bot. Let’s say I regularly order pizzas online. It’s the same process each time, and I’m able to quickly order with specials, pick from a custom menu, and go through an easy and intuitive interface to reach my end goal.
What I wouldn’t like to do is take 16 steps and way too much back and forth with a bot every time I order a pizza. This is something that has plagued chatbots before, but there’s too much focus on interface, content, images, language, etc. The focus needs to be on time. No one will switch from an app to a chatbot if it only takes longer on a chatbot. While I know it’s still a bit of the wild west for chatbots, I feel like a chatbot with a memory and a way to really build and personalize to your users’ needs can really stand out from the rest.
Imagine if I could quickly select an order I made in the past and have the chatbot make relevant recommendations, just based on its memory of me as a user.
Let’s dig into the actual concept now. Memory is just a way of storing data — it’s what you do with that user data that counts. With a lot of the server-side development today, the opportunities for storing and recalling memory on command are nearly endless. Long-term memory allows you to easily create different situations where your bot can offer a more personalized experience for each user, which more than ever can help with retention.
Memory will make your chatbot smarter, more powerful, and more intuitive. While chatbots are still trying to prove themselves as worthy tools for daily life, no one wants to do long tasks over and over again just to find the end result to be the same as an app. This is all about the psychology of user engagement. Websites and apps follow this model, too. Imagine if you had to re-enter the same user info and other important info each time just to order something. Bot developers need to make sure they can keep up with what will make bots stand out.
The flow should be simple and focused on making the user’s experience the easiest possible, especially when they come back. Even conversationally, it’s already a really good sign if your bot says “Welcome back, Bob” after the user’s been gone for a while. A bot where you have to type in “start” or even say “hello” again only to start a long and frustrating conversation over probably won’t really see a good, consistent, and active user base. You need to focus on daily active users if you’re looking to build a bot that replaces something as big as an app. Long-term memory is the key.
In the end, chatbots still have a lot of potential to thrive technology-wise, but they need some work in personalization and the standards involved. The amount of work a user needs to do at times undermines the concept that bots can replace apps.
With apps, it’s a few clicks. Yet a chatbot can really “get to know you” with flexible and customizable user accounts. With the powerful potential of machine learning and AI on the horizon, this user experience has the potential to become exponentially better. A chatbot isn’t as extensive as an app or a website right now, but if you can successfully utilize both interface and memory together, you can dramatically improve your bot’s usability and overall potential for recurring users.