Recently I studied IM Bots, but unfortunately every time that I experienced them on a Facebook Messenger I felt unsatisfied. Bots and AI are for sure the personal assistants of the future, but we must wait for their evolution and for our language adaptation (read my posts about Bots: Chatbots are contents, not conversations and Facebook Messenger’s Bots are direct, customizable and automated communication channels, not personal assistants).
Intead of Bots, in these months I used many times a lot of Chat Customer Services on some companies web sites and on Messenger. As all the studies say, communicating with a company through our favourite instant messaging app is smarter than downloading any branded app or using the old-fashioned email. My experience was great and these companies increased loyalty and my admiration.
Using Whatsapp, Messenger, Telgram or WeChat for companies is a great challenge for many technical and communicational factors:
- Technical, because CRMs should access to IM platforms for identifying users and managing the requests trafic.
- Communicational, because some contents should be always and easily available for customers instead of lose in the chat’s flow.
As a Product Manager I focused on the second problem and, starting from a Whatsapp-like layout, I designed the “Featured contents” function. The scope of this function is to enrich the discussion between the customer and the company saving the requested contents in a reserved area of the app.
Watch the “Featured contents” gif animation for understanding how it function in the direct relationship between a Hotel and its customer.
After my first post about Facebook Messenger Bots, I continued my research because I understand the importance of Chatbots and Instant Messaging apps. Following a transcription of the presentation that I published on Slideshare.
The assumptions from where I started my reasearch are:
- Users don’t want conversations. Users want pertinent and timely contents within the app that they use most.
- Chatbots have the reason to exist because users don’t like to download lot of apps and because mobile sites are slow or difficult to navigate.
- Chatbots are a communication channel with an interaction pattern in a sort of way similar to the natural language. They aren’t virtual sales agents.
- Chatbots have the difficult mission to bring together contents and services within messaging apps.
- The best chatbots performances aren’t based on conversations. Interacting with them requires new functions and a standardized command language.
So I can say that Chatbots are an important technology because:
- they represent a way for engaging users within their favorite apps
- they can replace apps and websites for simple and recurrent tasks
- they are the only direct marketing channel comparable with the email
- they revolutionize the smartphone’s push communication marketing
- they are the entrance point for advanced data building programs
- users interest in downloading branded apps is decreasing
- mobile navigation sometimes is frustrating
- users are accustomed in making Google searches in a conversational way
But this importance bring with it some threats:
- chatbots can’t really understand natural language
- chatbots can’t replace the all the other apps functions
- chatbots could decrease the users curiosity and research capacity
- chatbots will struggle for visibility
- chatbots can’t wrong a lot of answers and they can’t ask too much questions
- chatbots must care a lot about language, style, frequency and relevancy of their push contents
- chatbots aren’t a branded channel
Chatbots are the future of Customer Relationship Management (CRM) and Direct Marketing for the following reasons:
- because they deliver profiled offers and contents, receiving immediate feedbacks
- because they are an effective support for the human-based customer care
- because they will build accurate customers profiles analyzing the interactions and asking for information, ratings etc
Thinking about all these incredible opportunities, I examined the standard instant messaging apps user experience and I realized that Chatbots should have a dedicated set of functions that designed as following.
At this point I tried to go practical matching my Chatbots functions and experience with some generalistic companies.
Reading on the web the new Facebook Messenger’s Bots reviews I confirmed my idea: human language is still too complex to understand by any kind of Artificial Intelligence (AI). Read “I Tried Shopping on Facebook Messenger. It Didn’t Go Well” by Lukas Thoms and “Facebook’s grand plan to simplify your life is off to a rough start” by Alex Heath.
So please, stop dreaming about a J.A.R.V.I.S.-like Bot. AI will never be like a personal assistant that knows everything about you, that understands the environment, your feelings and your needs. AI assistant will be for ever a digital system that gives complex and nice outputs just because someone coded all kind of linguistic inputs that a human can produce; this kind of assistant will never really understand what’s happening. The most advanced AI possible is the one that has the biggest relational and semantic database tested (manually!) by real operators (read “The Humans Hiding Behind the Chatbots” by Ellen Huet).
Natural language isn’t the key
Machines that understand some plain language commands and that can anticipate some users needs are possible, but computers that are able to understand all kind of phrases that a human pronounces, sorry, but aren’t near to come.
Like everybody us today can understand icons on expensive glass-plates called smartphone, in the same way we must create a simplified language for communicating and using Bots.
For me nobody wants to lose his time talking with a Bot even if companies would love the idea that millions of virtual and assertive sales people talk 24h/7 with customers. Instead, the most amazing feature of the Bots AI isn’t their humanity, but the fact that users can treat them without any courtesy, that they will memorize users tastes and credentials, that they will anticipate users needs thanks to some “natural language” commands and some Facebook profile analysis.
All this doesn’t mean that companies shouldn’t care about language per se, but that they should drive users to use a simplified language for the following reasons:
- a simple language is easier to explain in a sort of tutorial during the first chats
- a simple language is faster and more efficient than the natural one. If the number of taps for receiving an information on a chat is a way more than searching it on a website, the chatbot is going to fail
- creating a sort of standard simplified language for all the Bots will ease exponentially their usage.
The users fruition model will be like the one that today drives sites like Yahoo Answers, Quora or the common FAQs pages where contents are organized and required using the “How to…” and “What is…” format.