When Deniz started out in digital, modems were the size of small cars and the population of 'online' was not far away from one of a sleepy outback town. After 20 years of working in key global digital agencies, building his very own and running some of the largest digital accounts in Australia, the only thing that has not changed is his fascination with this ever-evolving medium.
In the last 12 years, he has worked as the co-founder and managing director of Webling Interactive.
Deniz has won over 20 local and global awards across his career and is a frequent contributor to industry publications. Brands Deniz has worked with in his career include Microsoft, Google, Coca Cola, Coles, Telstra, Channel 7, NRMA, Mirvac, Acer, Australian Government, CommBank, Amex, and Arnott’s.
I’m sure that many of you out there have heard a lot about chat bots (aka messaging bots) recently, and the fact that they are here to stay is pretty evident by now.
You might also be wondering if this is something that your business should be considering, but might not be exactly sure how to go about it, or where to start.
I am hoping this article might help.
Firstly, you have probably heard these two terms in context of bots quite a bit; machine learning and artificial intelligence. A lot of the conversations around the ‘deep learning’ side of bots are focusing on an academic and theoretical view of these concepts, as well as the ethics and morality of artificial intelligence and its impact on the future of human race, including a few ‘Skynet theories’, so to speak.
These are fascinating conversation topics and an important sphere to stay on top of for the future, but very little practical use for your brand and business right now. Unless you are in the business of discussing philosophy with your customers, you can use bots right now commercially, without worrying about any of that.
How? Let’s start with the basics.
What is a chat bot really?
Messaging bots are automated programs that interpret and deliver messages pretty much like a human would. Bots can be programmed to carry out automated actions to respond to requests from users, or even engage them directly.
Their main difference from their predecessors –such as websites and apps – is that their user interaction layer is text, and in some cases voice-based conversation in natural language. This removes the need of traditional UIs such as menus, buttons, forms. Users just send and receive messages from the bot. No need to learn, understand and navigate disparate interfaces. Users interact with bots just as they interact with other humans. It’s the most natural way to communicate and transact.
Why is everyone excited about them?
They are always on and instant; they work through a medium called instant messaging after all. Furthermore, given that direct messaging is on the rise while other mass mediums including social and web are declining, it’s easy to see that bots are the future.They are endlessly scalable; with the right infrastructure they can engage millions of users simultaneously without breaking a sweat.
But most importantly, they transcend the decades old GUI-driven ‘human computer interaction’ paradigm into a ‘human to human-like interaction’, thus achieving warmer, personalised and more delightful experiences.
What can bots do well now?
Bots can very efficiently automate conversations, transactions or workflows, in most cases all these simultaneously. They can be trained on a subject matter and engage the user autonomously to answer their questions, give them advice, help them select the right product or service and if applicable help them complete transactions.
The most likely short to mid-term opportunities for any brand will be in the areas of:
- Customer Service: Bots can be used for automating contact centre processes and saving time and money, as well as offering instant 24/7 coverage and approximately 90 per cent of user issues. Think of live chat services many brands offer, without the cost of staff or worrying about time of day.
- Product selection and commerce: Bots can be trained to converse with consumers in natural language, understand their needs and goals by asking questions and provide recommendations for the right product to buy or service to subscribe to that is best for their circumstances. They can then seamlessly transition the user to existing ecommerce processes or channel them instore to complete the transaction, or even transact within the chat window. Bots offer the best value in areas where the consumer might need a lot of handholding for decision making and traditionally relied on “experts” for advice due to deep and wide options such as holidays and travel, health and beauty, employment or “connoisseur” style categories such as wine and fashion.
- Customer relationship management automation: Bots can continually and contextually engage your customers to drive decisions and actions out of them instantly. Think about those messages you receive from your telco letting you know that you might be exceeding your quota, and imagine if you were able to have a conversation with that message to find the best plan for you and upgrade right there and then, without leaving the chat window.
What about machine learning?
From an academic point of view, the self-learning ability of bots is the exciting bit. This is when a ‘deep learning’ bot learns from previous attempts and multiple independent data sources, and uses inference to solve new problems that it has not been directly trained on.
For the commercially feasible now bots we are talking about here, this is more like self-fine-tuning, where the bot stays within the boundaries of the direct training, but can figure out small details by itself and gets better and better in that narrow area.
Clear as mud? Let’s use an example. If we were to train a commercial bot to sell footwear, it will not be able to figure out how to sell handbags by itself, even though they are related subjects. However, it will be able to learn what type of users buy what type of footwear in time, and start getting better at providing recommendations that match the users’ taste, and increase conversion rates.
What do I need to get going?
Do you have a problem to solve or opportunity to capitalise on that matches the above? If yes, great, you are 80 per cent there. A quick checklist for other things that will come in handy are:
- Do you have data sources for the products or services that you want to provide? If yes, do they have metadata that help define and categorise them? Are they in feed / API format?
- Do you have an existing ‘selector’ style website or application that helps users with product / service recommendations, meaning that you have an already worked out business logic layer for this?
- Do you have existing customer inquiries / questions in any format, CRM database stored or as emails?
- Are there subject matter experts in your organisations that can help train the bot and teach them their deep experience about the business?
How do I take it to market?
Although bots can work on any instant messaging platform, Facebook’s Messenger is the key channel to reach users, as it has the highest penetration of users (pretty much everyone who has Facebook has Messenger). Facebook offers special paid ad units that link directly to a Messenger conversation to target and engage your audience as well.