Sorry, but chatbots are not the new apps

If you even remotely follow tech news, it is very likely that you have read outlandish claims like “chatbots are the new apps.” But the fact is, there are way too many misconceptions around conversational products.

Facebook Messenger already has over 11,000 bots. Let’s add Kik, Telegram, Line, and Slack bots on top of that. Now that Apple has joined the race as well, expect the bot explosion to get even more intense.

I’m building a simple bot myself. (No, I won’t start a bot company. And, no, it won’t replace an app.) I can guarantee you, from both personal experience and research, that “normal” people do not care about issues being faced by app developers. If you are not a developer, investor, or journalist, you probably do not spend a minute of your day thinking about dwindling app store revenues, worsening app discovery problems, or the fact that an average person downloads no new apps every month. So, if the motivation behind building a bot feels like something your non-techie friends don’t care about, it is safe to drop the bot product altogether.

The narrative of “Bots are the new apps” or “Bots will replace apps” leads to the following:

• “If you can get people to take your price advantage for granted, you’ve won” wrote Sarah Tavel in one of the most insightful pieces I have ever read on consumer behavior and products. Most consumers do not like paying for apps, due to the sheer abundance of options. We can imagine the same happening with bots. So, let’s assume the “price” in this context is the cognitive load of downloading, storing, and maintaining an app. A large part of the narrative around the bot hype is that people will not have to pay this “price,” since using a bot will be as simple as opening a chat thread to talk to a friend. But, as Tavel argues, price is a feature. Ultimately, it is about the 10x product. From a customer’s perspective, it only makes sense to adopt a new solution if it is significantly better than the alternatives — it’s not enough that it’s a trendy thing to do. Again, most people do not care what theme all the journalists are writing about these days.
• It pollutes our understanding of what a 10x product will look like. It will be one of two things: 1. A 10x bot product will create an experience that apps or websites cannot offer. A truly A.I.-powered assistant that can handle and deliver on highly complex search queries would be a good example here. Imagine being able to tell a bot (via your voice or typed words) to find you a 2-bedroom apartment on the ground floor of a building that is in a 1-mile radius of NYU and has food spots within a 5-minute walking distance. Now, that’s something. 2. A 10x bot product will enable people to do something significantly more conveniently than they could with existing solutions. A powerful example of this category would be bot solutions in countries like India, where an average smartphone owner probably owns a cheap model. Besides a set of core apps that such users already have, they will either not have enough data connection or enough local storage space to download a new app. Furthermore, the mobile web is slow, and very few websites are optimized for mobile. Bot solutions in countries like India provide the best way of accessing the internet. Contrast this with smartphone usage in the U.S. Most people have unlimited or huge data plans that are spent on countless hours on Facebook, Snapchat, emails, etc. Not wanting to download new apps out of “app fatigue” is significantly different from not being able to use new apps due to financial and technical issues. A good proxy for identifying a 10x better product in the U.S. would be to see if a large group of people choose a bot for a specific use case over a well-designed app that serves the same purpose. Data plans and regular smartphone upgrades enable people in the U.S. to use whatever apps they want. The switching cost for opting for a bot over a website or an app is very high.
• It instills a wrong mindset. Conversational products will not and should not replace apps. Apps stand for visual experiences. Bots, whether voice or textual, stand for conversational experiences. When we use a “bots will be the new apps” mindset, we blindly try our best to convert visual workflows into conversational workflows to create new kinds of apps (chatty apps, if you will). This will change. We just need to wait it out. For a lot of use cases, conversational products are simply not feasible. At least not yet. Furthermore, regardless of how far we get with ML and NLP, some experiences are best suited to visual mediums. Textual and voice bots seem like good solutions for high-intent tasks like search and work-related problems. Visual experiences are best suited for discovery-based, low-intent experiences.
• It tends to ignore the reality that the successful apps of our time weren’t built thinking “apps are the new websites.” They were built by unlocking the tremendous value of certain capabilities of smartphones. Think Uber and GPS. Think Snapchat and the camera. Similarly, successful bots won’t be built by thinking “bots are the new apps.” They will be be built with a “bot-first” mindset by developers who truly unlock the value of deep learning and natural language processing.
• It leads us to believe that building a bot company is much more important than building an experienced-based, problem-solving company. Apps, websites, bots, etc are only distribution channels. If it’s bots today, it may very well be VR tomorrow. Building a product that happens to deliver its value and experience on a platform is different from building a product whose value and experience are designed after picking a platform. It needs to make sense to have a messaging app as the platform.
• It instills a mindset that makes people want to approach bot-building the way we approached app-building. Besides social apps and apps with a social component, most apps are used on a one-to-one basis. We interact with an app whenever we want to do something specific with it. Sure, when you have a lot of people using the app independently and individually, the app developers get data effects to leverage. But the experience is still in isolation. The “bots are the new apps” mindset will lead us to craft bots that we use in isolation. Interacting with a bot one-to-one is not a pleasant experience in most use cases. A lot of great experiences can be crafted by building on top of an existing experience or workflow and summoning different bots for different use cases at varied times. This will require organic discovery via leveraging social graphs and platform assistance.
• It overlooks the fact that most people are visual learners. Playing around with a visual interface feels a lot simpler than figuring out conversational commands and, more importantly, remembering them the next time we use a product. With well-designed visual cues, a user does not have to remember a lot of things. That being said, “visual bots” or “micro-apps” with a lot of visual elements will be helpful. Both Apple and WeChat are headed in this direction.
• It motivates us to optimize for a misdirected goal. If there is anything that successful bots will likely replace, it will be humans for specific tasks. Not apps. Bots, in the true sense of the word, fundamentally, stand for automation. Per this definition, Apple and WeChat’s services within messaging context do not count.
• It is confused with the narrative of “providing services where people spend most of their time on smartphones.” This narrative is a meaningful and justifiable one. But having a messaging/social context as an environment is different from having it as a medium. WeChat, as Connie Chan and Dan Grover have explained, is the best example of this outside of U.S. Apple is on the path to becoming an example of this in U.S.

I think changing our line of thought from “bots replacing apps” to “bots creating something new” would really help in making the bot trend about consumers and not developers. Ultimately, selfish people will be using these products.

And, no, chatbots won’t fix your customer service.

P.S .:  The bot I’m working on will probably replace a newsletter. Not an app. So that’s different :)

A version of this post appeared on Medium.