May 19th, 2016 Do Chatbots Dream? By Fabio Bertone

Facebook, WhatsApp, Telegram have succeeded at making chatbots a trending topic again.
Why so? And, how soon will bots become part of our routine?

Why is kind of easy — Both Facebook and brands like the idea of bots.

Facebook
Facebook is succeeding at being the go-to place for a chat with a friend or for quick entertainment. We don’t Google our friend Chris: we WhatsApp him, and if he doesn’t answer, we peek on FB to see if he is posting pictures from Paris. We still go to news sites to get our coffee fix…but we keep tabs on the BBC, vogue, the economist through the day on our timeline.

Chatbots promise to bring the next step inside our chat tools: transactions. It will mean buying products, booking a table, playing a song. It will mean bringing chat tools to be a new platform…that will take its fees to allow brands to be there.

Brands
To brands, chatbots make two promises: that they will be able to connect at a personal level with consumers, and that doing so will be easy, supported by a new generation of AI algorithms.

A more personal connection is a brilliant call: instead of competing on price with Amazon and on Search budgets on Google, a brand could dial up on loyalty and invest on the value of their presence on the consumer mind.

Algorithms bring down a barrier that has been there since the invention of the “personal shop assistant”; you cannot have (if you are not playing in the luxury space) a human assistant for each of your clients…and training them is expensive and time consuming. Chatbots AI will break down the barrier: trained one time, they will scale to support all your clients on Black Friday.

But are we ready?
There are some considerations to be made: are brands ready? Are users ready? Is the experience ready?

Tom Ford

Brands
Maybe.

Big organisation have been able to provide software driven chats for a while. You.md launched a comprehensive healthcare app already in 2015.

Smaller orgs face a challenge. Skills to program or train bots are still rare on the job market. Facebook is aware of it, and this is why they have invested in wit.ai, a solution to train your own bot. But, even with skills and tools, there is something that need to be addresses first. Data. Knowledge. To train a bot on how to recommend handbags and presents for your girlfriend’s birthday birthday, you will need a training set. That nowadays sits in the mind of shopping assistants.

So, expect some months of work before your favourite companies have something interesting to say to you. And, pain one year later, when the tone of voice of the brand needs to change.

Users
No and Yes.

There is an underlying assumption that users will like chatting with brands as they like chatting with friends. Though one. We check WhatsApp notifications because we know there is no spam waiting for us there. What will happen when McDonald stalks us and lunchtime to offer a promotion?

But, self-service is a new value for younger generation. Getting to the point you need to call someone to understand how to move on a website is a signal of high frustration. Having a channel for consumer to solve problems by themselves will be a key differentiator.

Experience
Needs work.

This is probably the part of chatbots that interests us the most. While chatting seems a simple paradigm, it is based on a common understanding between us and other old-style humans. How will it work with bots?

There are some hurdles to overcome: discoverability, hand-over to real operators.

Discoverability
Telling people what a chatbot can do requires smart approaches. On a website or mobile app, you can add one more link or button, what do you do in chats when you add a new feature?

Hand-over
If you accept that algorithms are not simple to train, you immediately see a solution: passing the ball to a real word expert when things get too complex. How much should a user know about this switch? Are we transforming the world of customer service in a Turing test? Will people start typing “pass me to your manager” to get to premium support?

Conclusion
Chatbots are here to be used. But expect some work to go on behind the scenes before they take over.