A Comprehensive Guide to Chatbots in Ecommerce

robot

Chatbots are around for quite some time now

Even though it seems like chatbots are a new topic because they recently flood the news, they are around for quite some time now. The first chatbot was presented by Joseph Weizenbaum back in 1966. It was named ELIZA and was meant to show how computers and humans can communicate. It could simulate a dialogue with a psychotherapist, which it did by comparing the user input with its thesaurus, searching for a matching word and putting this in a preconstructed phrase. In this way it was already able to convince the users that it could understand their problems. However Weizenbaum did not claim that the system was in any way intelligent. Weizenbaums research goes back to the famous article by Alan Turing about “Computing Machinery and Intelligence“. But even if we don’t go back in time that far chatbots are not a phenomenon of 2016.

So what is a chatbot?

Is chatbot is a program that you can send messages to. It is able to interpret this messages and react on them. In the most simple form these interactions are scripted. If the user for example types “Hello”, the chatbot looks up his dictionary for “Hello”. If the phrase is available there will be an answer connected, like “Hi, how can I help you” that is send back to the user. If the phrase is not in the dictionary , usually you get a help message which shows you possible commands. In this way you can build already a fairly complex chatbot, but you need to foresee and think through the interaction with your users.
If we go one step further, there are chatbots that have machine learning and artificial intelligence running in the background. The machine learning part makes sure unknown phrases from the users get categorized and connected to known interactions, to convert them into known phrases. Therefore the vocabulary of the chatbot grows over time which leads to more conversational interaction. The user is not limited to type “Hello” to start, but could also use “Hi” or “òla”. The artificial intelligence part is able to get the context of what the user gives as input. Therefore it can interpret what the conversation is about (in contrast to a given scripted flow). So it can pull together information much more flexible and make the conversation even more “natural”.

Why are chatbots important now (for ecommerce) and what changed recently?

Chatbots are simply a good way to get into interaction with the user where he is a lot of his time – in messaging apps. This is in contrast to the usual website/app where you always try to get the user to come to you.
The barrier to use a bot inside the communication channel you already use is much lower than to download and app or browse to a website. Also its much more convenient and potentially faster, because you basically use text messages and don’t need to navigate or load content to start an interaction.
But why now? Because with the opening of the Facebooks platform there is access to the biggest social network on the planet and therefore to an huge audience which is likely to include a lot of your customers. Also messenger apps of all sorts are used more than social media by now (see here). So basically messaging apps are changing the way social media works (article).

Also on the technical side some things come together now in a way that makes using bots possible (in a nice and useful way). Let’s play some word bullshit bingo: cloud, big data, AI, broadband and flatrates all contribute to using chatbots.

Chris Messina (the guy who had the idea with these hashtags) proclaimed 2016 already as the year of the chatbots (article). In his opinion the whole ecommerce world will make a transformation to conversational commerce.

The Platforms

Active users on the biggest messaging platforms (from DMR):

  1. WhatsApp 1 billion
  2. Facebook 900 million
  3. WeChat 700 million
  4. Viber 608 million
  5. Tango 350 million
  6. Kik Messenger 275 million
  7. iMessages 250 million

Facebook

Facebook just opened up its messenger platform for chatbots and presented some examples. Right now you can find numerous bot for the platform. One nice  – and working in production – example is from airline KLM (video). But also Facebooks own “Hi Poncho” weather service.

Here is the video from the F8 conference where an example is given (at around 7 min):

Slack

Slack started out early last year to have little bot programs available that did lots of tasks for you (for example creating a google hangout, adding stuff to a to-do list etc.). You can find the available bots in the Slack App Directory.

Kik

Kik is very popular messenger platform that became famous with teenagers for its anonymity features. But also Kik made use of bots early. To get an overview you can go to the Bot Shop.

Telegram

Telegram introduced chatbots back in June 2015, with a full-blown developer program (see below) and has already gained some traction.

Microsoft

Chatbots are available for Skype since end of March (Skype blog). Also Microsoft was so brave to let one of its chatbots out in the wide open early on Twitter – Tay.ai. The experiment failed in the eyes of most commenters, because Tay started to produce very inappropriate comments (article). But was that a complete fail? I guess no, because Tay was trolled by a bunch of people who pushed her in that direction. And what happened was actually quite amazing: Tay was able to adjust the communication to its users.

What can you do with chatbots today?

To get a nice overview of available chatbots there is the botlist directory which list bots by platform.

A nice and simple to tryout example for a chatbot working quite well is Facebooks Poncho weather bot. Simply open up the Facebook Messenger App (or the the send message button the website), look for “Hi Poncho” and start a conversation. The bot will guide you through finding out where you are and what you like to know about the weather.

Some brands took a deep dive into bots very early. Earliest adopter and pioneer in chatbots is Uber which launched its Facebook Messenger integration back last year. But also Sephora for example launched a chatbot end of March on Kik (we also talk about it in our podcast). Airline KLM introduced a chatbot on Facebook Messenger that assists in booking flights and passes on all relevant information to you throughout the travel process (video). And also fashion brand H&M launched a chatbot on Kik for outfit inspirations.

What will the chatbot future bring?

Even though the chatbot technology is not really new the adoption in lots of areas is. Especially in ecommerce there are just the first attempts to find out, what could work or not.
The chatbots will stay, simply because they can make life easier and some tasks faster. Sure, every CMO wants to have an own app for the brand. But this is typically thinking from the company side, not the customer side. Therefore chatbots can offer a compelling way to serve the customer where he anyway is most of the time – in messaging apps.
One challenge for sure will be to find out which messenger platform is the right one for your company. The answer: “Facebook, because it’s the biggest“ will not be enough in this case. Every company has to find out where their respective customers are, and how a conversation with them can be structured best to create added value.
I doubt that the bots will eat all apps and websites. Simply because the complexity of a chat with a bot has certain limits (for now). Also a conversation is not always the best way to do transaction like in ecommerce. But if the chatbots develop like described, some apps and websites will get in trouble. If you take weather apps as an example: they provide lots of information, but most users for sure just look up the temperature and if they need an umbrella. You can easily do this with a bot – Facebook demonstrates it quite well with Poncho.
A big factor for chatbots is how the artificial intelligence will develop. It keeps getting better and this will improve conversations. Better AI will make sure that more facets of the human input can be interpreted the right way and the quality of the answers will rise. And therefore the ease and speed of use.
Still it will be interesting to see in which areas chatbots will develop the best and where they might fail. Ecommerce is a very interesting, but also complex field, where creativity will show the possibilities. The first steps of ecommerce companies at least are pretty promising.

Developer Program Pages – if you want to get your hands dirty with bots

Below you can find a list of some popular messaging platforms and their developer programs. If you want to try to build your own chatbot you should start there. All pages have very good tutorials that get you ramped up quickly to get the idea and architecture of a chatbot across. Have fun coding!

 

Also the following links may be interesting, as they provide services around building bots.

2 thoughts on “A Comprehensive Guide to Chatbots in Ecommerce

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s