Instant messaging is hugely popular. Slack, Facebook Messenger, Telegram, WeeChat, and WhatsApp have billions of users. Paying for products and services inside chat apps is commonplace in many parts of the world. Users of Facebook Messenger can easily send money to other users. Apps like WeeChat provide a platform for successful eCommerce applications. Beyond instant chat, platforms like Amazon’s Echo are taking natural language processing for eCommerce to the next level.
A natural consequence of the popularity of chat applications and their bots, and of natural language processing advances, is the conversational user interface. Companies can build bots that are communicated with via messaging applications.
It’s a compelling idea. Imagine a shopper on her morning commute who remembers she needs to buy a new dress for the weekend. She picks up her phone and begins an instant message conversation with her favorite eCommerce retailer’s bot: “I’d like a simple black dress.” The bot responds with a selection of matching dresses in her size. The shopper makes a choice, and an hour later the dress is on its way to her house. She never has to open a traditional eCommerce application. Never has to wind her way through complex search and navigation. And never has to deal with a traditional eCommerce checkout.
It seems like this is the way forward, but conversational user interfaces are not without detractors. Is a conversational UI really the best interface for eCommerce? It works well for the simplest shopping sessions. “Hey Alexa, add shampoo to my shopping list” is a powerful convenience, but what if I want to buy a new router for my home network. In that case, I want to browse through lots of options, read the reviews, compare the specifications and prices, and so on. It’s by no means impossible that an artificially intelligent bot could do that for me in the future. A bot could learn my preferences and present me with exactly the product I would have chosen myself, but for most eCommerce scenarios, with current technology, bots just aren’t up to the job. “Traditional” eCommerce stores, with their faceted search, categories, suggested products, and reviews will offer a superior user interface until bots become smart enough to do it all for us — something that’s a long way off yet.
Simple chat bots can be useful for constrained tasks like adding appointments to a calendar, but any complexity or ambiguity almost certainly demands a fallback to a human operator. One of the major benefits of eCommerce is that we don’t have to involve a human in every transaction — automation is a money-saver. The cost of hiring and managing a support team capable of supplementing the limited capabilities of current AI platforms is not insignificant. As Ellen Huet discusses on Bloomberg, unless your aim to bamboozle a venture capitalist into throwing money at your startup, an “AI” chat bot that’s really an army of human workers is unlikely to prove financially viable.
I’ve no doubt conversational user interfaces will be big in the future — there a billions of instant chat users and that’s too big a market to ignore. However, for the time being, eCommerce within chat is likely to look a lot like a traditional eCommerce application.