We've always known two things about Wordhop.io. The first is that AI is really hard, and it will be many years before robots can communicate with humans 100% like friends. After all, robots don't resonate like humans. Second, Facebook consumers don't message a business because they want to shop or are interested in the weather. These experiences are predicated on having chatbots, and they also require industry data that has expanded from messaging to social networking.
What Facebook users really want is a quick response from merchants when they have questions.
The main reason consumers send messages to businesses on Facebook is to get service. Everything else is a cloud. Such interactive pre-sales may start with a customer's question, and timely responses can advance the transaction. If a customer contacts sales customer service, it is usually because they really have a question to ask customer service. Some answers can be answered by robots, while more complex questions can be solved by humans, who can respond as quickly as robots when given the right tools.
Yet instead of focusing on the most in-demand aspects, many bot developers seek to create new user experiences or reinvent and upgrade old concepts about chatbots. If bot developers feel engagement is low, it may not be because AI is failing, but because they are directing consumers down a path they did not want to take in the first place. Without leveraging existing patterns of consumer behavior or training robots to respond to customer needs, the role of AI is simply to enable the creative experience imagined by robot developers and guide consumers to interact through that experience.