Artificial Intelligence; What, Who, Where, Why, and how do we use it?

Emiel Tindemans
Whoever wants to keep up-to-date in the area of online cannot escape from it: Artificial Intelligence (AI). But what exactly is it and how can we deploy it in the area of online marketing? Read more about it in this article.

Firstly, a few more facts & figures that stress the importance of AI and our market:

  • According to research firm Markets and Markets, the value of the AI branch will increase to 5.5 billion dollars in 2020, whereas in 2014 it was at 420 million.
  • From 2018 on, over 6 billion devices will apply for support for AI (Consultancy firm Gartner).
  • Gartner also concluded that in 2020, 85% of the interaction between companies and their customers will no longer include human interference.
  • From 2020 on, 50% of all search queries will be spoken.

What is Artificial Intelligence (AI)?
Artificial Intelligence in itself is not new. Artificial Intelligence is about self-learning devices which, based on data, impulses, and their environment, can learn to take autonomous decisions, where ‘normally’ human intelligence is required. These are, for instance, voice recognition, visual observation and taking human-like decisions. As such, a bot was able to beat the world champion in the most difficult mind game worldwide. Simply by playing thousands of games and learning from the opponent’s, as well as its own decisions.

Who makes use of Artificial Intelligence?
It’s impossible to imagine today’s society without Artificial Intelligence. It is used in self-driving cars by Google and Tesla, the songs recommended to you by Spotify and of course in Google’s search results. For a number of years, Google has been showing different SERP’s (Search Engine Ranking Page) for similar search terms, based on your history and the history of comparable persons, but also based on the behaviour on these SERP’s. For us, online marketeers, AI has become a lot closer by the use of chatbots on websites and the emergence of ‘voice assistance’ by Google Home, Amazon Alexa and Microsoft Cortana.

Where do we use it?
The ‘where’ questions was in fact answered in the previous paragraph, but as we have come closer to our own operating area, we can address it further, using examples from the (Dutch) online world. Firstly, there are various companies that make use of chatbots already, as a (primary) means for customer contact. Insurance company ASR recently made the transition to an AI chatbot on the homepage, which replaces the traditional menu.

Furthermore, KLM recently introduced BlueBot which, as they name it, is a new member of the service staff. This bot is built in Facebook Messenger and provides chat function users with the opportunity to book a plane ticket via a chat message.

Why do we deploy Artificial Intelligence?
The two examples of ASR and KLM in fact show the reason why we wish to deploy Artificial Intelligence: we have found a new means of contacting our customers! A website may answer many of the questions, but still it is a static product that simply doesn’t enable us store all the information. In addition, in the current market, visitors are no longer required to sort everything out themselves, certainly not whilst competitors offer easier solutions.
Perhaps the most important reason is that, according to the current prospects, in 2020, 50% of the search queries will be spoken search queries. On the one hand this is a trend in the area of search behaviour, but on the other hand, it is a trend in the entire online world that we focus on. Someone who performs a spoken search query will also want to continue this by use of speech. That is what makes Chatbots the ideal solution. In a while you will be able to ask BlueBot to book flight tickets for a weekend break, whilst you are cooking a meal.

How do we deploy Artificial Intelligence?
Currently, Artificial Intelligence and chatbots are the most logical combination in taking a new step within our operating area. Different standard tools are available to set up a chatbot, which is never ‘just plug and play’. The fact that your bot will make mistakes in ‘small talk’, typo’s, smileys, as well as the simple fact that users may become annoyed, makes the set-up of a bot an enormous challenge, specifically when this bot is required to learn from its own mistakes. Messenger is a good tool to build a bot within. Do you wish to know more about chatbots and the associated code? MOZ recently wrote a blog about it. Finally, we would like to emphasise that even if only 1% of your visitors is going to use your bot, you will be forming a special bond with that 1%!

Some best practices & tips to take account of during the process:

  • Make it clear that one is dealing with a bot, as a result of which failures may occur. This will reduce the irritation factor;
  • Before setting up a chatbot, view the present issues and categorise them;
  • Do give users of the bot the option to contact a person, for instance by providing ‘speak to a person’ as a code word. Do not forget to indicate that a bot’s reply will be faster than that of a human being.
  • Add UTM tagging as soon as a URL is communicated within the chat;
  • Make use of a soft launch; to begin with, allow only part of the visitors to make use of the chatbot.

In Jonathan Seal’s words (of the British bureau Mando): BOTS are User interface (UI), not AI, but when combining UI and AI, magic appears!
Do you wish to know more about the future of Artificial Intelligence in the online marketing world? Then contact one of the RIFF specialists.