We chat with Banking Sector Magazine about whether your chatbot is helping or hindering your customers. We also discuss how AI can help train and automate your chatbot automatically for a better customer experience.
User experiences and expectations are changing with the growth in chatbots and there is an element of self-empowerment to solving problems as well as not wanting to interact with someone in a call centre. A study by Salesforce found that 72% of Millennials did not believe a phone call was the best way to resolve their issue.
But before we answer the question, ‘Is your chatbot helping your customers?’, we need to set a measure of desired outcome.
Analyse customer frustrations and the failure within each channel e.g. containment and the root cause of switch. That’s going to give you the right insight to reduce friction.
Identify the insights into what customers are trying to achieve in the channel and help build that roadmap to change. So if customers are telling you they’re trying to do certain transactions in a certain way, capitalize on that to get the best bang for your buck from the digital roadmap.
Prioritize changes that deliver both an improved CX as well as reducing operational costs. You’ll get a quicker payback to fund more change.
Work to reduce the channel silos and reduce customer friction. A typical target operating model to achieve this is between one and three years so this will take time but start none the less and you will realize benefits and reduce channel friction as you go through that journey.
Aim for an omni-channel experience in which every customer knows what channel to use for what transaction and every interaction is handled correctly first time, Again this will take time and may not by 100pc achievable but let’s make inroads to try to move to that space.
Be prepared to be overwhelmed for a short period of time – by the data and by the possibilities. Work with a partner who can help to make a quick difference, because otherwise PRCs will just become science experiments and a reason to not do things. Making a small thing successful to start with is always better than anything else.
In the 4th of our Omni-channel Series of blogs we look at why your omni-channel experience might be failing. It’s likely that you will be receiving customer feedback through a growing number of different channels such as surveys, customer conversations from interactions, or insights from customer actions – or how Garter describe it: Direct, Indirect and Inferred customer feedback. But as you try and provide a customer experience that operates consistently and efficiently across all channels there are some common barriers that might be getting in the way of success.
No empowered CX leader
Only 30% of organisations have an executive on the board representing CX. We think this is very low as we believe not having a single owner has a high risk of creating channel silos.
For example, we quite often see web self-service being owned by somebody different to telephony and customer service, with different agendas and different objectives. And that can lead to some friction.
Paradoxically, 9 out of 10 organisations see customer experience as a competitive differentiator. But if only 30% have an executive responsible for CX on the board, how are they going to deliver great customer experience with a silo channel approach? There’s certainly a high risk of failure there.
Furthermore, probably less than half of the 30% may not be empowered CX leaders. In other words, even if there’s someone responsible for CX, actually having the capability to affect cross functional change is easier said than done – some boards are just more receptive to empowering individuals. Consider whether your CX leader is as empowered as they could be.
Channel failure driving interactions
We found that 40-70% of interactions were driven from some sort of failure in the originating channel. In particular, 25-40% was driven from web self-service. This means customers have tried a cheap service channel in the first instance and then ended up switching to a more expensive channel to get their query resolved…that’s a massive amount of unnecessary demand and cost.
Although many organisations acting on the drivers of failure demand have delivered operational savings of around 11-20% this is still way below expectations and there is plenty of opportunity to improve, particularly if up to 70% of contact is waste!
Channels are not connected
Only 8.4% of organisations have every channel connected. For the other 92%, this is inevitably going to lead to friction because they’re not aligned. That runs the risk of some customer dissatisfaction and so we’re seeing an increasing number of organisations who are actively working towards a truly connected channel strategy.
You are likely to have a target to increase NPS or some similar metric for satisfaction and potentially customer loyalty within your business – this is a common use case for us and we have helped many clients increase satisfaction by 20% or more by helping to find drivers and root causes of customer dissatisfaction and setting an action plan for improvement – in many cases this is linked to a poor channel experience so getting this right can pay dividends.
For many clients we focus on understanding the drivers of dissatisfaction typically driven from channel shift and increased effort from customers. Understanding those insights leads to a pretty significant improvement in MPs. We would typically see between 10 and 20 point improvement by acting on those drivers of dissatisfaction.
Knowing the amount of channel failure
It’s really important to understand the amount of failure that you’re handling to help inform your channel strategy. Using analytics you can do this by:
Isolating contact where the customer mentioned that they tried to complete something in a different channel – frighteningly we have seen this number to be nearly half of the context we analyse;
Identifying the processes behind the root cause drivers of failure. Prioritise a set of actions that the organisation should take to help reduce that channel failure.
By carrying out these two actions you could easily achieve between a 5 and 7% reduction in headcount which could equate to around 150 FTE. A huge business benefit from the insights and if there’s a very high failure rate, there’s much more to go after too.
Understanding the types of demand you are handling
To understand the different types of demands being handled within a contact centre, you first determine whether a transaction is a value to the customer or an irritant to the customer.
And then do the same from your organization’s view i.e. is the transaction a value to you or not.
You can then plot these on a matrix. In this example (figure 1), the organisation was handling more than half (54%) of transactions that were of value to the customer but an irritant to the organisation. So these are prime for what we call automated drive self-service and that creates a massive opportunity for operational efficiency by maintaining a great customer experience, but freeing up that expensive resource in the contact centre.
Previously only 3% of the cases agents promoted self service to the customer but once they understood the value irritants and coached agents to promote self service, the number rose to over 60% in just 6 weeks.
Best of all, there was a massive impact in the contact volume with a 30% reduction in the contact centre
More accurate classification
When agents are manually classifying all of the data after every call, a lot of these classified labels are not actually very helpful. We call them bucket categories, because they are literally another category, which can be up to about 20%.
And there are also pseudo bucket categories, which is when two smart agents are doing what they think is right, but actually they’re classifying things differently. The class becomes confused and becomes unactionable. The organisations ability to develop their customer experience is limited because the analytics is just not granular or accurate enough.
By automating the classification or labelling, these bucket categories are removed and you start to see things that you may have missed from an early warning point of view that you wouldn’t have otherwise spotted. We call it a back cast.
And the nice thing is, when you’re automating all of that human activity, you’re taking away sometimes 10% of the work from those human agents as well as providing better insight.
Whatever methods that you, or your analytics partner choose, always look at how the customer is feeling in the beginning and then look at what the agents do during the interaction to make them feel happier. To achieve this it can be very powerful to look at the topics or emotional intent within the conversation. With the right analytics this can be done automatically and precisely – there’s no guessing what the customers are telling you or how they’re telling you. As Bill Gates said: “If you want to learn about the business, you’ll learn the most from your unhappiest customers.”
Now we’re not suggesting that you deliberately make your customers unhappy in order to retrieve information. But it is ironic that organisations tend to send follow up surveys people get very fatigued about, just to understand how a conversation went.
A lot of the time the customer is clearly telling you how they are feeling or what their intent is – you just need the right analytics to pick these sentiments up and the need for the survey is removed.
For example, if a customer is actually telling you they’ve switched channel you can look at the meta data of the topics to get a much more accurate and actionable qualitative view about that issue. Or conversely, if they have come over to a different channel you might not have picked it up in your initial FCR analysis.
Being able to isolate those comments where your customers are really telling you these things is really powerful. Yes, it’s a motive. Yes, no one likes to hit bad news. But it’s all there. And if you let it speak to you, you can follow these pathways, see which are the big ones and tackle things in the priority order to be able to drive the quickest improvements.
There’s always a lot of low hanging fruit when you can effectively and accurately identify multiple intents across an omni-channel experience. Because when you know what you didn’t know before, then you can go after it quickly.
The AWS (Amazon Web Services) Partner Network (APN) is the global partner program for AWS. It is focused on helping APN Partners build successful AWS-based businesses or solutions by providing business, technical, marketing, and go-to-market support.
APN Partners are focused on your success, helping customers take full advantage of all the business benefits that AWS has to offer. With their deep expertise on AWS, APN Partners are uniquely positioned to help your company at any stage of your Cloud Adoption Journey and to help you achieve your business objectives.
Dan Somers, CEO of Warwick Analytics says: “We are delighted to become an official AWS Partner, providing software solutions that are either hosted on, or integrated with, the AWS platform.”
Warwick Analytics has launched a disruptive new approach to text analytics. ‘MaaS’ (Models-as-a-Service) is a quick, easy and low-cost way to apply AI to text to get powerful automated insight.
MaaS provides a shortcut to getting started with text analytics AI. Existing models are provided for specific industries and use cases which can be fine-tuned or used standalone.
Until now ‘AI-based Text Analytics’, the use of machine learning to classify text, has been cited as an expensive solution requiring data scientists to craft bespoke models for datasets.
Warwick Analytics is best known for its AI text platform PrediCX that can generate accurate machine learning models without the need for a data scientist. It is this ‘human-in-the-loop’ technology that has enabled them to create MaaS, a disruptively low-cost AI solution. For as little as a few hundred dollars per month, a model can generate predictive insight that other analytics costing ten times the price can’t deliver.
Models are available across multiple industries such as restaurants, hotels, leisure, banking, insurance, retail, ecommerce, CPG, transportation, manufacturing, utilities and healthcare. Use cases for each industry range from identifying root causes of churn and loyalty, predicting sales of new products, predicting marketing effectiveness, finding root causes of brand equity, as well as automation use cases for CRM and helpdesks. The datasets can vary too from social media posts and reviews to surveys and CRM notes and contacts.
Warwick Analytics will continue to expand the range of models and is happy to ‘build for free’ for new customers who have new challenges or datasets.
To demonstrate the effectiveness of MaaS, Warwick applied it to publicly available data in different industries to identify key insight and savings. In one example MaaS found one leading UK telco (O2) could automate 45% of the Tweets, chats and direct messages into its contact center, as well as identifying easily-addressable root causes of churn and customer purchase difficulties worth millions of pounds per year. In another example MaaS looked at addressable root causes of churn for Expedia and the savings were estimated at over $5m per year. When looking at TripAdvisor reviews for TGI Fridays churn and loyalty root causes, not previously found by one of the other leading text analytics provider, were identified. The number of churners identified was 10 times higher with a much higher precision rate of 88% versus 54%.
Dan Somers, the CEO of Warwick Analytics comments: “We are delighted to launch MaaS. Not only is it exciting to bring tangible financial benefits to customers with actionable analytics and automation, but it is a privilege to help disrupt the market in a positive way, removing some of the ivory towers of data science and smashing some of the myths and claims. Democratizing data science means that data scientists still do the ultra-cool stuff and get the credit they deserve, whilst for everyone else in both large enterprise and smaller companies, they get transparent, powerful, cost-effective tools to make their businesses customers happier and their businesses more profitable.”
A lot has been made of digital transformation, and how many businesses are using self-serve web-based applications to engage with their customers, employees and other stakeholders to be able to enhance and in some cases reinvent the customer experience often with both a stickier customer journey and lower service costs. Uber and AirBnB are often held up as the poster-boys but there are many businesses who are not ‘digital native’ companies emulating in their own way.
As with so many buzzphrases, there is usually a less-sexy way of saying the same thing which has been around for a long-time. In the field of customer interaction, most people will think of digital transformation as the growth in chatbots and social media-enabled communication. However I would argue that the main bastion of change has to be directed at FAQs.
Sexy or not, FAQs used to be the only way to read about self-help and avoid calling a contact center. They are frequently cited as inherently flawed as these blogs from the UK Government and eloquently in this technical writers blog. Yet if you stop and think, a well-structured FAQ if it is searchable with natural language is a critical asset as it is really the same thing as a chatbot, but perhaps without the charm or manners.
In a more measured manner, really FAQs are part of a spectrum of communication channels (one-way and two) where customers can solve problems. They sit alongside forums, social media, chatbots, chat, phone and email (see diagram below).
Also surveys and reviews can trigger interactions depending on the content. The current state of most organisations is that all these elements are separate silos, and whilst the customer experience team are trying hard to break these silos down, there are few who would see FAQs as on the same spectrum as chatbots and forums. Also people’s expectations of FAQs are to see a laundry list of requests which is not how they desire to interact. Imagine though if you could write your query in any way you wanted into a search bar and it would retrieve the correct response. Imagine also that the search was entirely consistent across all channels. Is this just a fantasy?
Machine learning for text is capable to classify interactions to be able to automate responses to natural language. However as we see with chatbot fails, it is hard to get this right due to the complexity and variability of human dialogue and chatbot containment rates are still below where their proprietors would want them to be. Further complexity is that human dialogue varies immensely across the channels: People don’t write in emails how they use chat which is different again to forums, nor how they speak or write a complaint, or even fill in a survey. By way of an example, a study at an airline found that the average topic in a chat was just over one whereas in a call it was nearer to two and in a complaint it was two and a half. People use different channels for different things, and also use different channels for the same thing in different ways. If the company is trying to classify aka tag or ‘label’ each interaction, then it will very easily fall into the trap of having different categories or tags for different channels, not by design but because it is hard to normalise them whatever technology you’re using. This phenomenon doesn’t really have a formal name but it is rife and disruptive. The ideal is some kind of ‘homogenization’ of the tags i.e. so that “late shipping” can be the same concept whatever channel. This then allows the guardians of the customer journey to understand what’s going wrong (and right), get a global view, and also understand if customers are calling back about the same thing on a different channel because they didn’t get it resolved. This also means that the customer journey and knowledge base can be fixed once for each breach, in the knowledge that it is fixing things across the board.
Machine learning can help this harmonization process although it is fraught with challenges, not least because the models for each tag need to be built especially for each channel, for example the “late shipping” tag for chat will need to be a different model to the “late shipping” tag for email or complaint. What data scientists know is that the process of building the machine learning models is intense: New York Times estimated that up to 80% of a data scientist’s time is spent “data wrangling”. CrowdFlower estimates “data preparation” at 80%. Further, assumptions and errors are an inevitable part of the process where human judgement and skill is required. More than this, 76% of data scientists view data preparation as the least enjoyable part of their work. Furthermore someone needs to build a training set for the models and that typically involves a human somewhere labelling the various interactions into topics and a topology that can drive the correct response. This is laborious in a linear fashion.
There are a number of different approaches to this problem. One company addressing this problem in a novel way is Warwick Analytics which is a spin-out from The University of Warwick. It has developed a proprietary technology called ‘Optimized Learning’ which puts a ‘human-in-the-loop’ in a very effective way: What this means is that the technology classifies the customer interactions in a meaningful way but when its certainty is low, it asks for assistance for a human to classify or ‘label’ the interactions which provide the most information back to training the models. Therefore it is theoretically and practically guaranteed to involve the minimum human interaction to maximise the performance of the models and hence the accuracy. The human trainer can be offline, as well as involving the customer in certain circumstances. The company has worked with many enterprises to improve chatbots, automate contact centers, complaints handling and improve the quality of self-service and FAQs.
So in conclusion, FAQs are an old-fashioned and much discredited digital experience, yet in the new world of digital transformation and harmonization, they can come back to center stage thanks to some clever technology and the human-in-the-loop.
Warwick Analytics is one of the five start-ups selected to join the BMW Group Financial Services Innovation Lab, the first financial technology business ‘incubator’ in the automotive sector.
The five finalists present a range of innovations that could revolutionise how consumers own and insure cars in the future, from opening up entirely new types of leases to consumers, through to tackling the barriers young drivers face.
Mike Dennett, chief executive BMW Group Financial Services, said: “Despite being vital for many drivers to access vehicles, the auto-finance sector has failed to innovate in the same way that most other areas of the automotive industry have in recent years.
“Our industry is ripe for disruption and each start-up we’ve selected to enter the Innovation Lab offers the potential to change how thousands of consumers gain access to and use vehicles, and I’m confident that our support will help us both work towards this goal.”
Warwick Analytics was selected by a panel of BMW Group Financial Services representatives following an intense pitch day this month.
Warwick Analytics will receive access to BMW Group Financial Services, in order to support innovation and fast track the growth of their business.
· Dedicated office space at the BMW Group Financial Services head office in Farnborough
· Mentorship from members of the business’s senior leadership
· Product validation and testing opportunities
· Access to investors at corporate innovation specialist L Marks
· First class education programme based on the ‘disciplined entrepreneurship’ method from MIT.
Stuart Marks, technology entrepreneur and partner of the Innovation Lab, said: “Every start-up that attended the pitch day presented a new idea for changing automotive finance, which meant that selecting only five to participate was a difficult challenge.
“However, we’re certain we’ve selected five strong companies who will benefit tremendously from being members of the Innovation Lab – it’ll be thrilling to watch them develop over the next ten weeks.”
The Innovation Lab will officially launch October 3, run for 10 weeks, and close with a demonstration day on December 8.