Alexa where do I start?


Unlike apps, conversational UX has little or no visual interface to give away an initial sense of functionality, completeness or finesse.

The democratisation of AI technology such as speech recognition and natural language processing means there is now a very low barrier to entry. The front-end, and some of the back-end tech is all nicely wrapped up for us in chatbot frameworks, Amazon’s Alexa Developer Portal and many other cloud-based API’s such as IBM’s Watson Conversation service.

Where we are now

alexa image 1.jpg

There is already talk of retention issues; a combination of poor findability, lack of notifications and the idea of a one-hit wonder.

That may be fine if you’re building something innovative for a simple campaign but it’s hardly the wider promise of AI.

It is easy to underestimate how hard it is to create killer conversational experiences which pass a variant of Larry Page’s toothbrush test. That is something that we use regularly and makes our lives better. In this case the beginning of a true ongoing dialogue between a brand and its customers.

Econsultancy recently called out the dangers of rushing into this.

Much like the evolution of the web some of this will only be solved through iteration. As we get better and better at creating the intelligence and back-end integration it will gradually evolve into highly functional one-to-one experience between customers and brands.

Developing your own conversational UX

First off, find something that will save your customers time.

They will love you for it.

In today’s world speed is always appreciated even if the task is simple. We remember great experiences if they are quick. We may even pay a bit more for it.

Which product, service or small customer service aspect can you wrap up in a way that reduces friction and can be made faster via conversational UX?

For example:

  • Remove the frustration from some commonly used aspect of your front-line operational service. E.g. I’m constantly amazed at how backwards many airlines still are in the digital space with basic stuff like checking-in.
  • Remove a heap of the clerical work - navigation, form filling etc. that is normally required to achieve a particular task with an app or the web. E.g quotations, risk assessment. Forms are a concept designed to help computers, not humans. Natural language processing can grab that information for you.
  • Provide highly precise answers to some particular (narrow) category of question for which you know you have examples of those questions and the answers. FAQ’s are the most obvious but static. What are people asking your call centre?
alexa image 2.jpg

You could use your analytics to identify some customer pain points. Are any particular kinds of interactions a good fit? Your contact centre, website and site-search analytics all hold clues.

But you probably already know this.

So which ones are then achievable today using conversational UX? 

Would they be more suited to a Chatbot, Alexa skill or both?

That is a much harder question to answer but now is the time to start mapping this stuff out.

How we can help

Even if some customer interactions are not yet ripe for today’s technology there are bound to be some great use cases that are.

New technology is emerging all the time which is being positioned as AI and not only is it hard to keep on top of, where does it all fit? It can be daunting.

We also need to be careful with expectation management. AI is not a magic wand and can require a great deal of hands-on and iteration to get to an acceptable level of accuracy for even the simplest use case.
alexa image 3.jpg

But all said and done we live in exciting times. I believe this next wave of chatbots and Alexa skills will gradually start to give us some real utility and user-adoption will increase as a result.

In the meantime, back to dinner.

"Alexa - Which cocktail would my guests like?"

Karl Hampson