There’s an industrial revolution under way in businesses across the world, and it is all about automation. Businesses are embracing machine learning and artificial intelligence to make better decisions automatically. And the reason for this revolution is the comparative strengths of humans and computers.
Computers are strongest at repetitive tasks, mathematics, data manipulation and parallel processing. So long as a task can be defined as a procedure, a computer can do that task over and over again, without getting tired, giving the same results each time. Computers can manipulate numbers and data in volume much faster than any human.
Several years ago I went back to university to do a masters degree, and after a 25 year break from university I was out of practice at mathematics. Imagine my excitement and relief when I discovered that now there is software that will do algebra and calculus for me! And computers can do more than one thing at a time. Have you ever tried to rub your belly and tap your head at the same time? I can’t do both actions simultaneously. But modern computer networks are powerful, able to routinely do dozens of different processes at once.
This does not mean that humans are obsolete. What humans are much more skilled than machines at are communication and engagement, context and general knowledge, creativity and empathy. When I have a frustrating problem, I want to talk to a human. Someone who will understand my exasperation, listen to my experience and make me feel valued as a customer, whilst also solving my problem for me. Humans are much better at common sense than computers, instantly recognizing when a decision doesn’t make sense. And humans can be creative. I recently heard music composed by a computer, and I’m sure that song won’t make it into the Top 40!
Customer Service
Recently I had a conversation with the manager of a call centre that dealt with hundreds of customer service issues each day. In order to ensure the quality of the service and advice, the call centre operators were given scripts and were commanded to follow those scripts without changing a word. The problem was that both staff and customers became frustrated. Staff felt bored and unchallenged, and customers with non-standard problems felt like they weren’t being heard. Staff turnover increased, and customer satisfaction levels dropped.
Customer Satisfaction
The manager then tested using chatbots to answer simpler questions from customers, freeing up the human operators to deal with non-standard enquiries. This was a situation where computers had a comparative advantage over humans. The call center processes were fully defined, operating at scale, and the scripted answers were correct. The results spoke for themselves. Computers were much better at helping with the repetitive enquiries, and humans were better at dealing with the unusual enquiries. Staff engagement increased, as did customer satisfaction.
This has implications for human resources and process innovation. Processes that require humans to do repetitive, well defined tasks can be replaced by artificial intelligence. This frees up staff to do what humans are best at:
asking the right questions,
applying common sense,
creating new solutions,
evangelising new ideas, and
generating sales and profit.
Let your humans be human. Free them from repetitive tasks. Change job descriptions to focus on human strengths, and hire people who best embody the comparative advantages of humans. Look for human processes that are well defined and repetitive, and enhance the process by introducing artificial intelligence. Some ways company have started to incorporate artificial intelligence and machine learning into their processes include:
There are even some companies out there that have started automating the automation, like DataRobot. Instead of hiring and training up a data scientists, the arcane process of building predictive models, once the sole domain of data scientists, can all be automated. The system automatically builds predictive models based on your data, freeing up your humans to be human, to be better conversational AI specialists.
Based in Singapore, Colin is the Director, Customer Success and Lead Data Scientist, APAC for DataRobot, where he advises businesses on how to build business cases and successfully manage data science projects. Over his career, Colin has held a number of CEO and general management roles, where he has championed data science initiatives in financial services, healthcare, security, oil and gas, government and marketing. He frequently speaks at various global conferences. Colin is a firm believer in data-based decision making and applying AI. He is passionate about the science of healthcare and does pro-bono work to support cancer research.
The other day, my 73-year old father, was grumbling about something he read in the news about automation of processes at the local bank.
“People don’t talk anymore. In my day, customer service meant talking to someone, saying hello, asking how your day was. Now it’s a recorded voice or reading tiny print online. Customer service is dead.”
That got me thinking. The role of a traditional customer service representative has evolved over the years. Once the domain of primarily the service and hospitality staff, the role of customers and our relationships with them has seen several costume changes — the phone IVR, surveys, forms, smiling uniformed people, you name it. But even as the modes change, the role of customer services and engagement has only just increased. Today customer relationships have become a full-fledged industry. 70% of buying experiences are based on how the customer feels they are being treated. The more people feel they’re being listened to, the happier they are and the more money they’ll spend — or that’s the hope. An Adobe report even suggests that customer service can deliver a higher ROI than marketing. Customer service, once upon a time, used to be about happy people, lots of solicitous questions and a solution with a smile. While it’s certainly true that the human factor seems to have declined over the years, the key tenets of a human interaction customer service have remained — conversations, solutions and a smiling demeanor.
So can a bot — the latest entrant into the customer service role — actually deliver these admittedly-human qualities?
Chatbot as the perfect concierge
Businesses that recognize how much time consumers spend on messaging apps such as Facebook Messenger and Slack have developed automated messaging technology. In 2015 messaging apps surpassed social networking apps, and chat apps have higher retention and usage rates than most mobile apps. Today, brands are looking at bots to become the next concierge, to understand what the customer wants, which direction they’re headed on, to involve them in interesting content, spread brand awareness and indeed, carry on conversations with a smile. But is all of that realistically possible?
On paper, it’s the perfect solution. Bots are machines, easily duplicated and incapable of human drama. They can be taught to function perfectly with a specific set of rules or through machine learning. These capabilities, limited as they are, can be trained to emulate the perfect customer service person’s skills — kindness, patience and solution-oriented. A machine can be taught to never be sarcastic and to always have a listening ear. And because it doesn’t have human failings such as fatigue or just being an asshole, it’s becoming an increasingly widespread phenomenon.
Any company with a chatbot interacting in the marketplace has the opportunity to gain valuable customer information. This has benefits in several areas — more personalization, targeted marketing, sales strategies as well as manpower allocation.
Not there yet
While bots were also a hot topic at the recent Corporate Social Media Summit, the jury were admittedly slightly skeptical. Bots are still very much in their nascent stage. And there have been several failures. In the rush to develop the next Siri or Cortana for their businesses, what most companies have ended up with are simplistic, underdeveloped tech with limited capabilities and faulty data. Of course there are the filthy people of the internet. It took less than 24 hours for Microsoft’s Tay to turn into into a filthy Nazi racist troll and two weeks for the cute little hitchbot to become roadside shrapnel. Even if the world were a perfect place, everyone was sunshine and unicorns and keeping empathy and other qualities aside, the actual functions and solutions given to customers by these bots need to work. That requires very skilled developers — but even they aren’t free of error.
Having said that, the possibilities for a bot are immense. Even though the big tech companies haven’t quite cracked how to make it work. Our co-founder Chris has a strong vision on the problems with Conversational AI, and perhaps more importantly — he offers solutions. He will share his vision 26th August on Startup Friday (still some spots left) and will start to share his vision in our blog series about Conversational AI that’s coming up here on Medium.
I know I will be paying attention. My own life have tons of bots — from the local Asian store from online magazines to Facebook to even my fitness wearable. Chatbots might very well be the face of the future one day. Now if they only knew how to smile.
Machine learning. Yes, it’s one of the most popular buzzwords in tech today. And no, it does NOT involve robots replacing humans and sitting in classrooms. We’ve gathered 6 of the best machine learning courses for someone who has no idea what machine learning is.
According to TechTarget, “Machine learning is a type of artificial intelligence (AI) that provides computers with the ability to learn without being explicitly programmed.” Machine learning is behind the creep who can identify the faces and locations in your photos. It powers those smart thermostats that can learn about your daily routine and automatically adjust the temperature when you leave the house and come back. It’s the wizard behind the curtain for many speech recognition technologies. It’s the technology in Woveon that learns your behavior and helps you prioritize your customer conversations.
Request a demo on Conversational Technology and manage customer conversations with ease.
Sound fascinating but a little overwhelming? Want to learn more about machine learning but scared about not being able to understand? Don’t have a technical background? No worries! There are tons of people out there in your same exact position, and they’re scraping the Internet for resources that are easy to understand. We’ve got your back with these…
Price: free, $79 for certificate (financial aid available)
Length: 10 weeks
This machine learning course has received excellent reviews from students. It offers a structured combination of readings, videos, practice quizzes, and graded assignments (with flexible due dates if you work better with deadlines!). It is very suitable for beginners because it even reviews the mathematical background you’ll need for the course, such as linear regression with one variable and linear algebra. It even talks about cool applications of machine learning, such as smart robots, computer vision, medical informatics, etc. (‘Cause who doesn’t want to know more about smart robots that we can command to do our work for us.)
Professor Andrew Ng is the Co-Founder and Co-Chairman of Coursera as well as the Chief Scientist and Vice President at Baidu. His work focuses on machine learning, particularly deep learning. He was a leader in the creation of Stanford’s Massive Open Online Courses platform and the “Google Brain” project that developed deep learning algorithms.
Yaser S. Abu-Mostafa is a Professor of Electrical Engineering and Computer Science, Co-Founder of the Neural Information Processing Systems conference, recipient of several national and Caltech teaching awards, such as the Feynman Prize, and co-author of Learning From Data, a bestseller on Amazon. He has 9 years of experience as a technical consultant for Citibank.
Brian Charles Williams is a Professor of Aeronautics and Astronautics who built the first fully self-repairing, autonomous space explorer, Remote Agent, that was put into use in May 1999 onboard the NASA Deep Space One probe. In 2000, as part of the Tom Young Blue Ribbon Team, he evaluated future Mars missions. He is also part of NASA Jet Propulsion Laboratory’s Advisory Council.
Emilio Frazzoli is a Professor of Aeronautics and Astronautics and Director of the Aerospace Robotics and Embedded Systems (ARES) group. He received the NSF CAREER award and IEEE George S. Axelby award.
Price: 205 AUD (Bonus when they have promotions to get the course at $12.99AUD!)
Length: 40.5 hours
Kirill Eremenko holds a Bachelor of Science degree in Applied Physics and Mathematics from the Moscow Institute of Physics and Technology and a Master of Commerce, Applied Finance, and Professional Accounting from the University of Queensland. He was trained at Deloitte Australia and has 5+ years of experience as a Data Science management consultant in transport, finance, retail, etc. He teaches Data Science and Forex Training courses on Udemy.
Hadelin de Ponteves has degrees in Mathematics and Engineering and a Master of Research in Machine Learning. He was a Data Engineer at Google and has 4 years of experience in data science and consulting.
Price: $29/month or $299/year for entire Pluralsight platform, 10-day free trial
Length: 40 min.
David Chappell is Principal of Chappell & Associates, a law firm that advises technology firms on law and business. David has consulted for Stanford, Target, HP, Microsoft, and IBM, spoken at 100+ events, and led seminars in 45 countries for tens of thousands of attendees. He also teaches courses on IT Innovation and Cloud Computing on Pluralsight.
Includes 3,500+ videos of experts in the field delivering talks at conferences, workshops, summer schools, activities that promote science, and other events. Think of it as a database of machine learning courses.
Katie Malone graduated from Ohio State University with an undergraduate degree in Engineering Physics and from Stanford with a PhD in Physics. She works as a data scientist in the research and development department at Civis Analytics, a startup in consulting and data science software.
Udacity aims “to bring accessible, affordable, engaging, and highly effective higher education to the world” and provides Nanodegree credentials and programs that are catered to aspiring data analysts, mobile and web developers, etc.
Did you find these machine learning courses helpful? Comment below the most interesting thing that you learned about machine learning!
Looking to apply your newly acquired knowledge on machine learning to Internet marketing? Make sure you make a plan first! We have for you 8 Marketing Plan Templates to Blow Your Competitors Out of the Water.
Want to use a software platform that employs Machine Learning? Check out Woveon today!
A must-read guide for enterprises with billions of conversations and millions of customers.
Modern call centre with millions of customers.
Enterprises are much more overwhelmed with conversations than ever before. Not only do they have to actively respond to customers over a myriad of channels like email, phone, social and livechat, they’re expected to give personal, relevant and fast responses. To tackle this problem, many organizations are looking at new technology to help them meet customer expectations. Some of the most notable are AI chatbots, self-service knowledge bases and good old Interactive Voice Response (IVR) systems. The problem? These all aim to lessen the time customers spend with agents.
While people do like self service for speed and convenience, majority still want to be able to talk to a person in times of need, or at important turning points in their life. Curiously, while we’re moving more towards a more digital and self-service world, most consumers still want the ‘human touch’ in their service communications.
Human touch is more important than just automation.
The challenge is to provide highly personalized and relevant offerings to meet both customer and business goals, all the while delivering the experience through the customer’s natural mediums of interaction. Counterintuitively, the likeliest solution to bring the human element back into customer conversations is though technology and big data. So, what should you look for in a technology that will give you both customer satisfaction and maximize revenue?
Multichannel Conversations
At the basics, an organization’s communication channels should be in one view. That means a business should be able to see and reply to customers by email, phone, livechat, social media, forums and wherever they could be talking to you, or about you, on one platform. Why? Convenience and transparency.
Convenient Conversations
A single platform for the entire range of conversation channels is much more efficient for customer-facing agents. Often, they have to switch between multiple channels to check for new customer interactions, and unfortunately, miss some communications here and there. With one view for conversations, they save on time, and reduces the chance they will miss communications from less monitored channels.
The convenience isn’t just for agents. Customers want to interact with brands through their medium of choice. 51% of U.S. consumers are loyal to brands that interact with them through their preferred channels of communication. Younger consumers especially, want to interact with large organizations via instant messaging channels where they can use natural language. Having all channels on one platform allows agents to have visibility across all channels, instead of doing well on a few and lagging on others.
Transparent Conversations
In so many organizations, a different team handles a different channel. They are responsible for that channel, and that channel only. But the customer is dynamic. They might reach out on one channel, and upon finding that it isn’t fast enough or substantial enough to resolve their problems, they will switch channels.
The ‘different team, different channel’ approach doesn’t account for the customer’s flexibility, resulting in multiple replies or inconsistent replies from two different people, both creating bad customer experiences. With multiple channels on one view, conversations are transparent. Conversations from the same customer are stitched together, and the same person can handle issues without making the customer’s journey difficult.
Holistic Customer View
In an enterprise with multiple departments, systems and channels, it’s necessary to have a collective view of the customer. A single customer view (or a 360 degree view) is a complete profile of a customer, created from aggregated data points within an organization’s systems and channels. It collates data from multichannel communications and customer data platforms (like CRMs, analytics, marketing and legacy systems).
Customers often complain about the lack of continuity in their conversations and having to repeat themselves. Problems like this arise because agents have no visibility on what customers have said on a separate channel, or what customer information exists on a separate system. As such, interactions are treated as a completely new “ticket”, and in the worst cases, existing customers are seen as a new customer. With a single customer view, an agent can see a given customer’s conversational, transactional and behavioral data in one place. This not only improves time-to-answer by 20% – 80%, it also ensures customer information flow is consistent and continuous, reducing awkward moments like the ones above.
The use of a single customer view can even go beyond customer care activities. Integrated systems mean that there could be a seamless blend of sales, marketing and service activities through conversation. Having this feature marks the start of being able to use critical sources of data collectively. The key however, lies in how the customer intelligence is used. The following presents ways customer intelligence can be used to take control of conversations in providing exceptional customer experience and maximize revenue.
AI-assisted agents
Use of artificial intelligence (AI) in enterprises is not new. For decades they have been used to automate heavily manual processes to increase efficiency, accuracy and decrease costs. What is new, is the use of AI beyond processes to interactions. Use of AI opens up the potential to deliver personalized interactions and hyper-relevant offerings that are scalable.
Whether it’s the AI itself doing the talking, or an algorithm providing assistance to a human representative, online, or face-to-face, AI holds incredible potential to re-establish the human-to-human connection in an increasingly digital world. Check out some examples below.
Deliver relevant content and information with AI
Many organizations have invested heavily into user experience, self-service and knowledge management tools. Yet, it is still difficult and time-consuming for customers to find the right information when they need it. Companies like Zendesk have developed AI-powered virtual assistants that help customers self-serve. By processing natural language, the technology suggests articles in the knowledge base to help them resolve their problems on their own. Research has found that most people are open to using self-serve AI technology like this, and see it as faster and more convenient.
Other organizations like Woveon have built AI-powered response assistants to help agents have more productive conversations in real-time. As agents talk with customers, the response assistance helps guide conversations so better results can be achieved for both the customer and the business. It would suggest opportunities like ‘other customers like her also bought’, or ‘he mentioned credit cards, link to these articles from our blog to help him decide’.
Speed up resolution times
On average, a customer care specialist spends 20% of their time looking for information and context to resolve a customer’s problem. That’s one whole day in a work week! AI can help organize information so that it’s easily digestible and relevant to a customer’s enquiry. Woveon’s Intelligent Response framework for example, will change the information it displays to assist agents based on the flow of conversation. If a customer talks about their personal loan, their loan details pop up. If the conversation shifts to their lost credit card, their shipping details will surface and agents are prompted to cancel the lost card.
Instead of wasting time looking for information, AI assistance leave agents more time to build a relationship and take up on untapped customer opportunities. Customers also love a quick and productive interaction. 69% attributed their good customer service experience to quick resolution of their problem.
Reduce repetitive admin tasks to open doors for higher value interactions
Administrative tasks like After-call work (ACW) have been a constant headache for employees in customer-facing roles. Though they are necessary, it’s tedious, repetitive and and takes up too much time. Technology can help to reduce time spent on these menial tasks, leaving agents more time to build customer relationships and, in the process, make their jobs more productive and meaningful.
For example, Avaya has a natural language summarization tool to help agents process customer information post-call. Talkdesk automates call routing, where the customer is automatically paired with an agent with the best ability to solve their problem. Woveon can prioritize conversations real-time, based on customer importance, value, urgency, or a mixture of all factors.
Freeing up employee time away from menial tasks allow them to participate in higher-value activities.
Intelligent Analytics
There’s no doubt that data analytics is incredibly beneficial for customer conversations. The trick is knowing what data to use, how, and when.
Whatdata is being used matters because not all data is created equal. For example, rather than looking at metrics at a point in time (customer rated the agent 4 out of 5 for resolution), it’s much more important to look at the larger picture (that it took 3 calls and an hour on hold to get there).
Howdata is used is arguably more critical to conversational success. The key lies in knowing what datapoints to tie together, and what analysis to draw from it. A mesh of marketing and service data can show how a recent marketing campaign has affected conversation volume and NPS. A cluster analysis of related keywords in customer conversations can lead to discovery of a huge logistics flaw.
Whento use what data is of particular importance to customer-facing agents. 74% of Millennial banking customers for example, want their financial institutions to send them information about services exactly when they need to see it. This could be information about personal loans when they’re starting to look for a house, or travel insurance before they intend to travel.
Companies these days have a wealth of data on their customers. In theory, organizations should have the ability to know who they are, what they need and what makes them defect to another company. However, lack of visibility on the holistic customer journey and customer intelligence tools stunt their ability to provide such excellence.
The following section will delve into three types of analytics particularly useful for managing customer conversations — predictive, clustering and revenue-generating.
Predictive Analytics
Predictive analytics provide foresight into potential customer problems and opportunities. Extracted from existing historical conversational, transactional and behavioral data, it can help agents better prepare for customer outcomes and trends.
A pretty common example is prediction of when influxes of customer conversations come in. For eCommerce businesses, holiday seasons generally see a spike in customer conversations and steadily reduces till the next holiday season. In a more complex scenario, predictive analytics can find that customers with a particular occupation, a certain concern and at a similar stage in their lives is actually a niche the organization hasn’t capitalized on.
Cluster Analysis
Now this one isn’t as common in a conversational technology, but is definitely worth mentioning. Cluster analysis involves conversations and customer information to be tagged, then for similar or related tags to be clustered together to draw insights.
Cluster analysis can draw out how topics in conversations can be relevant, or how particular customer segments can be feel about a product. This customer intelligence can then feed into other parts of the business. It could be used to help create a new automated customer workflow for upsells, or contribute to a new marketing campaign for a newly discovered customer segment.
Revenue-generating analytics
As repetitive and menial conversations are moving towards being solved by self-service solutions, agents must also move from a traditional support role to a hybrid service-to-sales model. This category of analysis is as the name suggests, analysis that serves to generate revenue for the business within conversations.
For example, Woveon’s Intelligent Response Framework suggests ways customer specialist representatives in banks can sell more products to their customers. A customer who fits the profile of ‘customers who typically get a black American express card’ will prompt a suggestion for the agent to talk the customer into an upgrade from their current card. A customer who is at a stage in their life where ‘customers like him are looking at buying a property’ will prompt a suggestion to link some home loan webpages, or a free session with a financial planner.
In the best possible scenario, this analysis is also delivered at the right time for an agent to capitalize on the opportunity, like in an intelligent response framework.
Be a data geek, not creep
Of course, it’s important to know that use of data should be “cool”, not “creepy”. There’s a fine line between the two that should never be crossed. Also, everyone’s fine line is drawn differently, so what one customer may think is cool, can be perceived as creepy by someone else.
Enterprises should have enough data about their customers to track and understand individual preferences, and see how customers respond to different use of their information at different points in the customer journey. Conversational intelligence and analysis tools can help create better relationships without overstepping the customer’s boundaries.
On a whole, customers don’t mind companies using their data for personalizing their experience and suggesting products and services that benefit them.
—–
While human contact is diminishing in volume, the quality and importance of each interaction increases. Forward-thinking organizations should be balancing quantity with quality to maintain a competitive advantage in customer experience. Technology can be a great booster to that end.
Have more ways you think businesses can improve on their customer conversations? Reach out to us to add to the article. We love chatting to like-minded people!
Artificial Intelligence and Machine Learning may seem like buzzwords, but they are potent technologies whose capabilities businesses have yet to fully understand. We can safely say that these technologies will end up revolutionising not just technology but the whole world, and this is not an overstatement. These technologies promise to be so destructive that businesses all over the world, and especially in Australia, are starting to realise that they will have to reinvent their organisations to succeed. When you look at countries spending money on automation and AI, Australia spends the 2nd highest amount of money on this field – second only to the United States – yet it lags behind in several key areas. If you run a business in Australia, you need to understand how machine learning and artificial intelligence will fundamentally change the way you do business. We know you keep running into advice and warnings about how we all need to start preparing for an AI world, and the best place to start would be understanding the basics of these technologies.
Understanding Artificial Intelligence
Our computers are much better at many tasks than us humans can ever hope to be, which is why we use them so much. When it comes to creating a database or doing a calculation, no human being can even come close to the efficiency, accuracy, and the blazing speed of a computer system. Yet, there are many problems which normal computers cannot even hope to solving, and many situations in which they are useless, because of the way our systems are designed.
Computers are machines which we can program. We can tell software and hardware to do x in case y happens, or do z if y does not happen, and so on. Our computer systems are smart, but they are not intelligent. Intelligence here is defined as the ability to create a solution when you face a new problem you haven’t solved before.
Computers are much more efficient than people as long as you can tell them exactly what to do. You have to program the computer for different situations. As long as the situation is something the computer is programmed for, the computer will be able to accomplish it without any problems. If the problem includes something the computer has never seen before, it will not be able to do anything. This is important because it is the main reason computers are still so limited, and why we need people to do all the work.
Tasks that Require Intelligence
In order for a computer system to be artificially intelligent, it will need the ability to understand things, instead of just learning them. Consider this task: you run a company that makes boxes. You want to quality test them using a computer. You can do this easily – you simply have to tell your computer what a perfect box looks like, and how much deviation from perfection is acceptable. It will not be hard for a computer system to compare a box made by you to the model of a perfect box in its memory and determine whether the product is acceptable, or if the shape is too deformed and the product should be rejected. Computers will be much better at detecting these imperfections than people – exponentially so.
Now consider another task – you are running an event for pets, and you take a lot of pictures. Half of your customers brought their dogs to the event, while the other half brought their cats. You want to be able to categorise photos based on which animal was in them – you want all the cat photos in a separate folder than all the dog photos. This problem will not be a problem for a human being at all, who can tell at a glance whether the picture includes a dog or a cat and categorise accordingly.
A traditional computer system, on the other hand, will struggle at this task. How does it detect ‘dogs’ or ‘cats’? Do you store every possible body type of a cat and a dog so that the computer can compare the animal in the picture to the models? This would be impossible, simply because of the variety present in animals. Size, shape, colour, any disability – any of these things can prevent a computer system from accomplishing this task.
Neural Computing
The main reason that humans are better at some tasks is the way our brain works. We don’t store information in absolute terms, we store it in relative terms. We have a general idea of what a dog looks like, and what a cat looks like, and we compare what we see to this general idea. Neural computing emulates this way of thinking. Instead of knowing exactly what a dog or a cat looks like, a neural computer has a general idea of what these animals look like, which allows it to make the right decision like a human.
The difference between Machine Learning and Artificial Intelligence
The terms machine learning and artificial intelligence are sometimes used interchangeably, but they mean different things. A true artificial intelligence system will be general purpose – it will be able to solve any type of problem you ask it to, the same way that a human being can. Except, it will be able to do what takes a human, several years, in the span of seconds, simply because computing power available to it and the memory in it will be vastly more than what a human brain has.
Machine learning is a limited application of artificial intelligence. It means creating a system which can learn through feedback – imagine a car going through an obstacle course, and every time it crashes, it realises it shouldn’t do what it did the next time. Run it through the course enough times and it will be perfectly ‘trained’ using experience/learning.
Conversational Artificial intelligence is still years away, while machine learning has been a reality for years. Machine learning allows robots to accomplish tasks such as managing a warehouse, and putting products in shelves in a supermarket almost perfectly, which will result in a lot of job losses for people, and an increase in efficiency for business owners. Self-driving cars are also based on machine learning. Our businesses are already fairly automated – every company uses a database and digital communications in some capacity – but machine learning will allow us to automate much more, and thus replace many more people with a few machines.
Artificial Intelligence (AI) and machine learning (ML) is all the buzz right now, and rightfully so with the significant contributions it has made to redefining many aspects of business. However, many people are still skeptical about the application of AI and ML to enhancing customer experience.
Some would argue that machines cannot possibly take over customer service, something that has a heavy focus on human interaction. Machines lack the empathy and emotional intelligence core to providing a great customer experience. On the other hand, many also see the benefit of applying AI and ML to automate repetitive tasks, allowing humans to dedicate more time to, well, being human.
We reached out to some experts from Oovvuu, Canva and The Minerva Collective to pick their brains about the issue.
What is the current state of customer experience, and how do you see it evolve with AI & ML technology?
Present customer experience is “all over the place, with wildly varying results. Two customers using the same service can have completely different impressions of their experience, and in many cases the service is clunky and poorly structured” says Anthony Tockar, Data Scientist and Co-founder of The Minerva Collective. The unfortunate reality is that 78% of consumers have bailed on a transaction or not made an intended purchase because of poor service experience. In fact, companies only hear from 4% of its dissatisfied customers. With so much choice available to consumers, it’s much easier to find another company with similar offerings than spending time complaining or calling about a problem. Which is why there is a very real need to focus on customer experience, a factor that is becoming increasingly important to retain the modern customer.
Paul Tune, Machine Learning Engineer at Canva, believes “there are two trends in improving customer experience:
A trend towards tailoring for the individual, as more data is gathered about each customer at a large scale, and;
A trend towards providing a smooth experience for customers across multiple touchpoints by anticipating their needs. “
To demonstrate how customer experience has evolved, Paul continues with an example. “Early recommendation systems, such as the recommendation engines developed at Amazon and NetFlix in the early 2000s, provided recommendations at a much coarser level, chiefly for specific groups of customers. The granularity of recommendations in the near future is going to be much finer. For instance, an engineer from NetFlix I spoke to recently, mentioned that a subscriber’s favourite character for a TV series would appear in the menu when the TV series is selected. This means having to learn more about each customer and predicting their habits. We also see this in the form of smart personal assistants, such as Alexa and Siri” he says.
Ricky Sutton, Founder and CEO of Oovvuu, adds on that whilst AI and ML “certainly has an element to play [in customer experience], it also lacks a key element…empathy. So my thought is that it will evolve. The more AI is used, the more it learns and the better it gets, but human-level empathy remains a pipe dream for now.”
What is the biggest lesson you have learned from applying smart technology to customer experience?
For Anthony, the lesson has been the need for people using smart technology to properly understand it – “My experience is that people often don’t trust what they don’t understand. The latest technologies have been great for grabbing headlines, but only the most forward-thinking businesses are serious about applying them to derive value. This isn’t necessarily a bad thing – domain knowledge is essential for good data science, and blindly relying on new approaches has many inherent risks. There is a lot that has been learned about customer experience over time and there is a need to explain smart technology to business people using the right language to allow them to fully realise its value.”
To Paul, what matters most, is the customer’s end-to-end experience. Meaning that all the touchpoints with the customer should be seamless. For him, “the challenge with integrating smart technology to improve user experience is similar to managing any other complex system: with more moving parts, there is a higher chance of failure in the system. Naively applying machine learning to improve customer experience is misguided. Machine learning works best if it is complementary to the customer experience, serving to enhance the experience of a great product.”
“At Canva, our goal is simple: we want to give the customer the best experience in empowering them to create and design. To that end, there are two aspects that we focus on. Firstly, how do we make the content that they need for their designs easily accessible. Secondly, how do we anticipate what resources might be helpful for them in the future. We achieve these goals by improving our search and recommendation services to enhance customer experience.”
The biggest lesson for Ricky is that “AI turns humans into super-humans, but only for certain tasks.” – “When we started Oovvuu, we hired editors to read articles and find relevant videos, and they were able to read one publication each and find 40 relevant videos per day. That same person using the AI tools that we created, can now read 100,000 publishers, and 300,000 stories a day, covering 26 million topics and find relevant videos from more than 40 global broadcasters. AI is mind-blowingly powerful for automating manual human tasks, but humans remain better at all the things that, well, make us human.”
What are some challenges for businesses who try to integrate AI & ML technology and customer experience?
Anthony, Paul and Ricky all agreed that a huge challenge for businesses is not having a solid data infrastructure, or a deep understanding of what exactly should be measured to achieve business goals and customer satisfaction.
“Many companies approach us seeking to use conversational AI as a ready-made silver bullet for a business problem. Others come to ask to play with AI, so they can find a business opportunity. Neither really works.” Ricky said. “For us, the solution was to know what business problem we were trying to solve: namely, to put a relevant video into every article being published worldwide. We then used AI to solve it, but what we started with was very basic and not up to the job. We have had a team nurturing the teaching for almost 1,000 days to get it where it is.”
Anthony went on to add that “there is no silver bullet – good data scientists are required to translate these algorithms into business value. Having a solid data science strategy is essential, and through good leadership, increased data literacy and an understanding of how to build a high-performance data science team, businesses can harness these technologies to forge a competitive advantage.”
Paul concludes with another common challenge many businesses face when adopting AI & ML into their processes – the volume of data. “Present machine learning techniques rely on a relatively large amount of data to provide good predictions” he says. “While there is fundamental research being carried out presently to (hopefully) reduce the amount of data required to train these machine learning models, the current main technological limitation of requiring a huge amount of data is here to stay for the foreseeable future.” But “fortunately, this effect can be mitigated if the data collected is of sufficiently high quality.”
Are you implementing AI and ML technology in your business? Share your story with us in the comments below!
About the Contributors
Anthony Tockar
Anthony is a leader in the data science space, and has worked on problems across insurance, loyalty, technology, telecommunications, the social sector and even neuroscience. A formally-trained actuary, Anthony completed an MS in Analytics at the prestigious Northwestern University. After hitting the headlines with his posts on data privacy at Neustar, he returned to Sydney to practice as a data scientist while co-founding the Minerva Collective and the Data Science Breakfast Meetup. He also helps organise several other meetups and programs for data scientists, in line with his mission to extend the reach and impact of data to help people.
Paul Tune
Paul Tune is a Machine Learning Engineer at Canva, responsible for developing solutions for tailoring and personalising content for Canva’s customers. He has several publications in prestigious computer science conferences and journals, including the ACM SIGCOMM conference in 2015. His interests include deep learning, statistics and information theory.
Ricky Sutton
Ricky is founder and CEO of Oovvuu, an IBM and Amazon-backed start up that uses artificial intelligence to match videos from global broadcasters with publishers worldwide. It’s mission is to use AI to insert a relevant short form and long form video in every article. In doing so, it aims to tell the news in a new and more compelling way, end fake news, and in doing so, repatriate billions from Facebook and Google back to the journalists and broadcasters who make the content.
Recently I wrote an article for LinkedIn titled “Can we maintain the human touch with customer service?” I couldn’t help think about how fast we are moving with Artificial Intelligence that the question still remains, I am not worried about 5 years from now or what new customer interactions will be digital, but how will businesses maintain the reality check with their customers? Surely digital chat bots and automated ticketing systems will ask random customers surveys about what they thought about their service response and the level of happiness to refer another customer. To implement is very easy but to keep the human connection with your customers will be the challenge.
Deliver Smarter Customer Service Solutions
At Woveon, we watch and analyse through thousands of conversations all uniquely handled by diligent customer service agents who, assisted with technology, work tirelessly around the clock to acknowledge, understand, listen to and resolve the incoming customer conversation. Clearly customer service has the human touch here! Even with today’s conversational AI technology surpassing standards in reliability, accuracy and now business intelligence the human touch in AI must not be far away? This is an important consideration looking at the technology landscape today, companies are working on delivering smarter customer service solutions, from chat bots that understand your sentiment and can adapt to your tone and writing style to automated enquiry systems that can help recommend products while you shop online. Yet still, customer service and particularly conversation management is still a human “touch”, something that is defined intrinsically in the term “customer experience”.
Let’s take the example of creating an outstanding customer service experience. Data tells us that outstanding customer service increases brand loyalty. Examples include begin a conversation with a podcast, send personal messages, create a lifestyle and get back to your customers. We’re not talking about getting back to your customers via a bot or automated reply email, but rather using an actual person who understands your customers and can understand the fine details and semantics of human feelings. Remember, customer service is all about listening to your customers and putting yourself in their shoes. Great customer service professionals can quickly adapt and understand the customer’s frustrations and calm their emotions. Being present and responding quickly in human is very different to doing this via a scripted automated response. However, in the enterprise world, a study by Oracle put it at 8 out of every 10 businesses who are already implementing or about to implement AI as a customer service solution by 2020. Nearly 40% of all enterprises are already using some form of AI technology with Forrester predicting a 300% increase in AI investments, the disruptive power of AI will impact every part of the business from customer service to sales and support. So are businesses going ahead at this the wrong way?
Having interviewed several CTOs and CMOs working with the technology, there is no rushing into the game looking for the holy grail. For most, the best step moving forward is in assistive and adaptive technology or to assist with data collection and analysis. AI technology is encapsulating more and more human qualities as technology advances. Bots are often deployed to collect data based off human input and use it to optimise the customer’s experience. This is particularly applicable to personalisation. Human teams then need to help filter, sift through and make sense of all the personalisations so the system can make better judgements in the future. Artificial intelligence predicts and prioritises the user’s interests according to their searches and similar inputs given by other users. This, when compared to the pros of human service, has similar benefits to empathy and experience. For example the human touch can continue on more serious, complicated customer challenges whereas standard, mundane everyday enquiries can be handled by AI bots. An example is AI assistance to lessen waiting periods for customer inquiries. KLM, the flag carrier airline of the Netherlands, used DigitalGenius’ AI system to answer customer’s questions faster. The AI units interpreted the questions and answered them with a quick edit of the preformed answer to relate directly to the language used by the customer. It was also able to adapt to the platform for the inquiries, pumping out longer responses to emails but limiting Twitter responses to 140 characters. Digital customer service seems to be directed towards matching human interaction but with the removal of prominent flaws.
So can we maintain the human touch in customer service? Having been a product manager and worked in technology since the first dot com (no I am not that old, I was just young when I first got into technology), we can expect to see customer service significantly enhanced with AI bringing the human touch to a new level. The amount of data that AI and ML will help sift through to help “advise” and “suggest” to a customer service team will break new boundaries. Customer service teams can then be deployed to work on escalated or prioritised items that result in a big sale or help close the deal. Customer service, intuitively is tied closely with the human touch, a computer cannot learn years of successful customer interactions without first being taught and guided by humans. This is a realistic fact.
The hype around Artificial Intelligence technologies is at its peak. According to the 2017 Gartner Hype Cycle, emerging technologies such as deep learning, machine learning and virtual assistants are at the “peak of inflated expectation”. Cognitive expert advisors have passed this peak and are now descending towards the “trough of disillusionment”. This occurs when interest wanes as experiments and implementations fail to deliver.
The benefits of AI for customer experience management are potentially game changing. AI has the capability to analyse vast amounts of data in real time from various sources, including human behaviours and emotions. Expectations are high because this capability can then be used to create seamless and personalised customer experiences that are optimised to the device and channel of choice.
Pragmatists and battle hardened cynics will recall that when automation was first introduced into customer service channels, the results were often spectacularly underwhelming. So, is the application of AI to customer experiences destined to fall into the trough of disillusionment before climbing the slope of enlightenment? Or is there a path to follow to avoid the pitfalls of unmet expectations?
Learn how Conversational Technology works. Request a FREE demo today!
Intelligently using Artificial Intelligence for Customer Experience
In order to find out whether the application of AI to your business’ customer experience will take a downturn, it is necessary to first ask yourself: What is driving your organisation’s AI strategy? Is it because:
AI is all the rage in your industry and your organisation is fearful of being left behind?
If you take the lead in implementing AI, it will make you look smarter/cooler than your colleagues?
It sounds like a cool and fun toy to experiment with?
Your organisation needs to catch up with your competitors who have been early adopters of AI?
AI is a great opportunity to reduce the cost to serve our customers?
If the answer to any of the above is Yes, then the trough of disillusionment beckons.
Alternatively, if you are deploying or considering AI because…
AI can enable your people and optimise your processes to operate more intelligently and efficiently, in order to provide individualised and predictive experiences for your customers at scale
…..then a brighter future awaits.
For these technologies to have any chance of success you should have a clear sense of purpose of how to you intend to deploy AI to drive CX in your business. Here are three ways you can use AI in a purposeful way to create meaningful customer experiences.
1. Use AI to Enhance your Knowledge of the Customer
An example would be using data analytics to anticipate the needs of individual customers at each moment of truth and key stage of their journey. Some specific examples oh how businesses are using AI to enhance customer knowledge:
2. Use AI to create stronger emotional connections with your customers
Using AI to recognise a customer’s emotional state helps agents better respond to the customer during an interaction, thereby creating stronger emotional connections.
Not only can AI empower agents with emotional intelligence to reply appropriately to customers, it can be used as a tool to connect service agents with the right information in the organisation’s knowledge base in real time. Examples of why this can be powerful to a business:
As a result, nearly two thirds of CX leaders say their organisation’s revenue growth outperforms their industry counterparts, compared with only a quarter of CX laggards. The proof is therefore clearly in the pudding: when applied in a purposeful and meaningful way, AI technology can enable organisations to increase agility and overcome competitive threats and leverage this advantage to drive acquisition.
Written by Dr Steve Nuttall – Head Of CX Research, Fifth Quadrant. Steve has worked in various leadership roles as a market research insights professional for over twenty years in Europe, Asia and Australia. He leads Fifth Quadrant’s program of CX strategy research and is an international speaker and presenter on best practice customer experience. Steve assists organisations to deliver their customer-centric strategies and business performance goals including designing and implementing programs to help optimise the customer experience.
Smart technology is rapidly changing the world we live in today. From smart houses to high-tech artificially intelligent robots, the world we live in is being saturated with advanced technology that simplifies our lives today. One industry that is fully embracing artificial intelligence is customer service. Companies across many different industries are jumping on board and using artificial intelligence and machine learning to create chatbots, remember customer problems, and create suggested responses. These are just some of the few incredible capabilities that Artificial Intelligence and Machine Learning and other smart technologies are bringing to the field of customer service.
Learn how intelligent customer service works. Sign up for a customer Conversational Software demo now!
How Artificial Intelligence is Leading the Smart Technology Revolution?
Smart technology is transforming the customer experience, but there’s also a more self-serving business benefit from its use. Those teams enabled and empowered through the likes of AI and predictive intelligence have higher rates of employee engagement. This is because agents feel more empowered and experience first-hand the positive impact this personalized style of service is having on the customer. There are two main ways that businesses are augmenting their customer care units with AI: “front-end AI-powered bots” and “AI-assisted human agents.” A front-end AI-powered bot is a conversational computer program that interacts directly with a customer without human intervention. On the other hand, an AI-assisted human agent is a human customer service representative who is supported by AI technology. Both these models are being used in service departments across industries.
Here are the 4 Smart Technologies for Smarter Customer Service
1. Chatbots
Of all the fields in the chatbot-crazed world, customer service is one of the prime targets for automation. Virtual customer agents (customer service-focused bots, or VCAs) are intelligent systems able to understand what users ask via chat and provide them adequate answers.
In 2015, the number of people using messaging apps overtook the number using social media. Beyond communicating with friends and work colleagues, individuals are increasingly using messaging apps to interact with brands. Messaging services are a brand new space for organizations to connect with existing and future customers. Businesses now have the opportunity to create new revenue streams using real-time, customized customer service bots within messaging applications. More than 34,000 chatbots on Facebook Messenger alone are a reflection of this opportunity. The airline, clothing and tourism industries are already leveraging this space. Consumers are connecting with brands through messaging apps to purchase airline tickets, book hotel accommodations and receive fashion advice. It’s only a matter of time before the other industries catch up.
The WeChat Messenger bot deployed by China Merchant Bank, one of the largest credit card issuers in China, is another example of a front-end bot. According to the AI technology provider Xiaoi, the China Merchant Bank’s front-end bot handles 1.5 to 2 million customer conversations per day, an inquiry volume that would typically require thousands of additional employees to answer. As most questions relate to card balances and payments, automation via a bot interface presents a relatively easy and cost efficient solution.
These above examples of chatbots already in use give a great introduction of the possibilities that AI brings to the field of customer service. AI makes it possible to create these large indexes so that the bot can respond back with pre-understood phrases or suggest responses to the human controlling it.
2. Suggested Responses
AI and ML are frequently misunderstood in the business world. They are not taking over jobs, but further enhancing current human positions. With suggested responses, companies can get through all of the requests at a much faster pace. AI bots can be used to read through customer requests and find the problem they are having. If you have repeat customers, customer conversation history is extremely important tool for your AI and ML. It can learn from past experiences and more accurately detect the problem the customer is having.
3. Conversation History
Woveon, a startup that provides intelligent customer service, uses AI programs to enhance its customer conversation history. Conversation history is simply saving past conversations between the customer and the business. Customer service representatives can then go back and use this data to help future questions the customer may have. Woveon is using this to essentially make a 3D diagram to further help the representative. The program will be able to go into the conversation history and track down the problems. It will then make a web with the biggest points being recurring problems. Branching off of the bigger problems will be other related problems that customers often have. This program, though still in development, will change the way customer service is able to work with customers. Woveon’s software, as mentioned above, is an AI-Assisted Human Agent and proves the importance of human interaction as it simplifies and speeds up these interactions.
4. Smart Tech Integration to the Business World
“As humans, we advance, that’s what we do. And the rise of AI in the customer service field is just another step in our advancement and should be looked at as such. There might be some growing pains during the process, but we shouldn’t let that stop us from growing and extending our knowledge. When we look at the benefits these chatbots, suggested responses, and conversation history can provide to the consumer and the business, it becomes clear that we are moving in the right direction.”
Based in Singapore, Colin is the Director, Customer Success and Lead Data Scientist, APAC for DataRobot, where he advises businesses on how to build business cases and successfully manage data science projects. Over his career, Colin has held a number of CEO and general management roles, where he has championed data science initiatives in financial services, healthcare, security, oil and gas, government and marketing. He frequently speaks at various global conferences. Colin is a firm believer in data-based decision making and applying AI. He is passionate about the science of healthcare and does pro-bono work to support cancer research.
Written by 