Why should business deal with chatbots

Why there is no way around Facebook, Google and Amazon

The last wave of hype about chatbots is slowly fading, but the topic is currently taking off in the background. After the digital world was shaken up again, it seems that automated communication no longer predominantly encourages doubters to speak, but inspires doers.


Anyone who sets out to create individual chatbot software for their own online shop in the B2C context today is backing the wrong horse. Now is the time to deal with the bot engines of the major technology providers and to create interfaces - so as not to be suddenly overtaken in five years. Clicking through online shops will gradually become obsolete, and communication with personal assistants will increase. And then it's like search engine optimization - those who can't be found just don't exist in my world.

But automated communication is not as easy as it seems. Before companies start with well-intentioned activism and develop individual chatbots, they should ask themselves the question of whether a bot makes sense. Does the communication with the customer really have to take place via chat and also automatically? What exactly should automatic communication do? How big are the benefits for customers and companies?

In the end customer context (B2C) there are often additional questions regarding reach and personalization. If you are investing in a bot, it should also be able to address the maximum possible target audience. In the business context, questions arise that intervene deeply in service processes. How much self-service can a chatbot handle, or is a chatbot "only" suitable as a good lead qualification machine?
Digital Marketing Days

The importance of chatbots, influencer marketing, retail media, virtual reality and content marketing is also a key topic at the Digital Marketing Days, which HORIZONT is organizing on June 29 and 30, 2017 in Berlin. At the mandatory appointment for digital decision-makers, leading industry experts such as Vanessa Bouwman (We Are Social), Jérôme Cochet (Zalando), Martin Wild (Media Markt Saturn), Peter Frolund (HTC Vive) and Ulrike Hefter (Nestlé) discuss which The role that new disciplines and technologies play in marketing and where companies should urgently invest.Register here now!

In the case of B2C considerations, with a view to reach, technology and data requirements, in the medium and long term there is no way around the large platforms such as Facebook (wit.ai), Google (api.ai) and Amazon (lex). The opening of the platforms for external applications enables the relatively easy integration of a wide variety of chatbots into the platforms and thus the provision for a very wide audience. In addition to the access to a high range created in this way, the technologies of deep learning (DL) and natural language processing (NLP) are offered integratively with the platform. The aim of this is to collect and bind external content (and thus also to bind customers) to the respective platform.

Nevertheless, current chatbots are only a small and rather experimental interim solution on the way to autonomous virtual assistants such as Alexa, Allo or Siri. What is mostly typed into a chat today will be spoken into tomorrow. The technical logic of the chatbots behind it remains largely the same, only the medium of customer communication will change. The extension of the current assistants to include external content is known as a "skill" and is basically nothing more than a text chatbot. The wizard only takes over the transformation from speech into text and, when outputting the result, from text back into speech. Finding a suitable answer, however, works on the skill itself.

© Group M

more on the subject

Group M manager Christoph Duscynski

"The topic of chatbots is cooked hotter than it is eaten"

Pretty much everyone in the digital and marketing industries is talking about chatbots right now. But what relevance do the smart chat robots actually have for marketing? Christoph Dusscynski is skeptical.

Skills are currently called using a separate command. So you say to an Alexa "Alexa, start" and to a Google Home you say "OK Google, talk to" in order to be able to talk to a special application on the platform.

But honestly, does that make sense in the long term? No, because I don't necessarily know every skill that could solve my problem. After that, I don't really feel like starting a new application, a skill, each time by voice. The current behavior is therefore not user-friendly. Consequently, the next step at Google and Amazon will be to recognize the intent, i.e. the intention of the questioner, in order to then start the most suitable skill semi-automatically. Partly automated in that the user can still make a skill selection. But which two to three skill options are offered to the user? I would use an auction model that lists the three "most important" skills that would fit the intent - the mechanics are well known and have worked quite well with Google AdWords for years.

If you also look at the increasingly deeper integration of the assistance functions in the operating system at Google, it should be clear that sooner or later the chat or chatbot on a single page will no longer be decisive, but the person who answers a question to the smartphone can answer and assist in the best possible way.

Unfortunately, Amazon has not been able to implement its smartphone approach (Fire Phone), so an Amazon via smartphone does not get any closer to the end customer than via an Alexa app or an Alexa skill for Google. The native and mobile function of recognizing the need remains in Google's hands with Android users.

"The great advantage of Facebook could be that there is an extremely large amount of data relating to real human communication via WhatsApp and FB Messenger."

Alexander Käppler

We will also notice the further integration of external offers into the messenger service on Facebook. The great advantage of Facebook could be that there is an extremely large amount of data relating to real human communication via WhatsApp and FB Messenger. Based on this data, it should be much easier to actually create self-learning bots and AI systems that require a very short or no training phase. For Google and Amazon Skills, people still have to specify and learn a lot - this could mostly be omitted with a Facebook integration, with a simultaneous presence of a large range, one thinks in the B2C context.

Only one of the big tech players is missing here. Apple is very cautious in the field of automatic communication and assistance functions. Like Allo, Siri is (in the future) a deeply integrated part of the system, but unfortunately it is only too timidly open to external applications and extensions. The closed Apple system seems to be becoming a disadvantage once again.

Ultimately, it remains to be seen how the platforms can be remunerated for a skill selection and how seamlessly we can switch from chat to assistant - developing our own solution currently only makes sense, if at all, in a business environment. In the end customer business you shouldn't even start without a platform connection.