Using Chatbots in Customer Communication

Few topics have made as many headlines in recent months as chatbots. Facebook, Google, Microsoft – the big names have discovered the subject and are publicly pushing it. If the major players have their way, then there will soon be no getting away from bots in customer communication.

Prominently showcased bots such as the Poncho weather bot show how clever machines can already be in speaking to humans. But there is also disillusionment.

As Eike Kühl of Zeit Online explains in his article “Chatbots – Hey, du Mensch” (“Chatbots – hey, human”), a genuine conversation is not currently possible as artificial intelligence (AI) is simply not yet smart enough. Even industry giants like Microsoft have had their noses bloodied when it comes to AI, as seen from the example of Tay on Twitter – in just one day, the self-teaching bot had turned into a racist Hitler-bot under the deliberately negative influence of numerous users.

What is already possible?

If we view the concept of bots and human-machine interaction from another angle, however, there are already many opportunities and use cases emerging where real value is added. Bild.de has long been delivering news via Facebook Messenger, and in the USA it is possible to book an UBER ride by bot. H&M has launched a shopping bot for kik Messenger – an exciting and thoroughly successful example of shopping directly through a messenger.

This is an approach with which WeChat in particular has established itself as a bot and e-commerce pioneer in the Asian region. Retail giants such as jd.com are already making the entire customer journey possible within the WeChat platform, allowing prospective customers to become actual customers without visiting the website a single time.

Where are chatbots currently useful?

In these cases, the messenger replaces the browser and the bot takes over the role of the website. So it is not a matter of conducting a conversation with a human using artificial intelligence, or even establishing an AI seller. What is needed here are bots which have a look and feel that is perfectly tailored to the messenger medium, and which can help users with day-to-day problems. Some examples:

• Do I have to send my energy supplier my meter reading for my annual statement?
• Does my insurer need the treatment cost plan from my dentist?
• Can I book a hire car for tomorrow at Hamburg central train station?
• Where can I order two concert tickets for Saturday?
• How can I quickly send a bouquet for Granny’s birthday tomorrow?

All of these questions can already be answered using a messenger-and-bot combination. And if the bot gets stuck, then there are interfaces that transfer users to a human salesperson without interrupting the conversation, so that the transaction can be completed via live chat. Customer contact takes place where the audience spends its time – and they are increasingly spending their time within the messenger microcosm.

Where is the road leading?

Only the use of bots in conjunction with messenger apps currently makes sense, as in the example of H&M and kik. However, once artificial intelligence has become a little more intelligent (and the tech giants are frantically working on just that) then there will be many more opportunities for dialogue between humans and machines. But a little more patience is needed, unless you want to annoy your customers – the low acceptance of voice-recognition systems on telephone hotlines being a cautionary example. It will be interesting to see what happens.

What is optimise-it’s position?

We are certain that bots can provide good answers to highly standardised questions. But there are a few things to bear in mind – for example, a bot should never pretend to be a human. We also think that a human contact still has clear advantages when it comes to emotively encouraging a purchase decision. With regards to the shopping experience or conveying appreciation, a human chat agent and their capacity for empathy is still irreplaceable (for the time being at least).

Comments are closed.