Chatbots are fine for some queries and some occasions. However they do not entirely replace the need to speak to real people – both on an emotional level and from the point of view of resolving more complex issues. Ultimately, in my opinion, retailers should always ensure customers have an option to speak with a human representative!

Chatbots are not ready for prime time, but they will get there within two or three years.

The pluses are related to availability. Many questions are simple and can be handled by chatbots – stock, order status, even questions about product attributes. Outside of these boxes, chatbots fall down. That’s OK.

I’m confident a lot of these advances will continue to escalate. Ten years ago, who would have thought we would have had something like Alexa? The future is bright for this technology in my view.

My personal experience with chatbots has left me a bit frustrated, as the technology needs to mature and strengthen significantly. Many times we get caught in logic loops as it is impossible to canvass all combinations of queries. The good news is that data science continues to evolve in application and will eventually improve service and communications. Be patient as data science moves out of academia into pragmatic application and use.

There are some queries that are straightforward and are not worth the wait for a customer service agent to get on the line. Retailers and brands have an opportunity to anticipate customers’ common queries. Until they are able to improve the satisfaction scores, chatbots should be on a training mode where humans intervene whenever it takes more than a few attempts to understand the customer question or senses frustration.

Chatbots are like auto-attend on the phone: Retailers must give customers an immediate way to get to a human if they want one. With chatbots, it’s essential to map the customer experience so that it aligns with the customer moment. This is not easy, but it’s essential. A huge array of potential customer questions and comments must be mapped to what they really want, and not too much what the retailer wants to push them toward. Test, test, and retest again and again, with real customers, to get it right before even thinking about deployment. The dislike of chatbots runs deep just as the love of them does. I believe that retailers should leverage their telecommunications infrastructure to provide a real associate from the stores to anyone who wants to speak to a person. With retailers spread across all our time zones there is sure to be someone available to help a customer and get a commission or credit for a sale.

As ubiquitous as search is and as dominant as Google is, user search is an antiquated technology. With search — even powerful search like Google offers — the user has the burden of sorting through results to narrow what they’re looking for. Filters help, but after price range, user rating, and a couple of other basic filters, most are poorly designed.

An intelligent assistant powered by machine learning algorithms that constantly train and improve and when given permission by the user, can learn the user’s personal preferences and access that personalized data, is the new paradigm that is on the near horizon. The experience will be magnitudes better than what e-commerce has been and yes while there will be a role for humans, it will in many ways exceed human capabilities. As pie in the sky as that sounds, the groundwork has been laid. Retailers that don’t again want to play catch-up, need to be planning/building out in that direction right now.

My short answer is, please let me talk with a human being. I find chatbots today need to be asked a specific question in order to get the answer to what is being asked. Use the same words in the wrong order and they get confused. One of the first things I try to do is find a phone number for the retailer or other service and call it.

There is still a good amount of novelty with chat bots, which explain consumer interest and engagement. Consumers still want to figure it out despite the frustrations that come when you realize you’re talking to a bot. Consumer interests in chat bots is a good indicator of where we are headed with online shopping. Soon, chat bots will be table stakes in ecommerce.

It’s also a perfect solution for consumers who truly do not want a live shopping experience but do want answers to questions. To that end, advances in natural language processing will be key to driving long-term use and to curbing frustrations from consumers when they accept chat bot limitations and reluctantly pick up the phone.