Conversation Analytics is the new game-changing way to extract value and insight from customer conversations.
It’s clear for everyone to see now that how we interact with and support customers has fundamentally changed. The pandemic has accelerated technology adoption at an incredible rate, or as Satya Nadella at Microsoft put it back in April 2020:
“We’ve seen two years of digital transformation in two months.”
This mega-shift towards a new remote working world has created a clear dependency on digital services and products that form part of the new ways that consumers behave and buy.
But what might have gone amiss for some, hidden beneath the commotion and turbulence of the eight month rollercoaster we’ve been on, is that this change towards digital was actually already in full flow pre-crisis. Although the crisis massively accelerated the process of digital transformation, the trend itself was already well established and creating a long list of winners and losers.
Digital transformation, although an often maligned term which can give people pause at it’s many interpretations, is fundamentally straightforward: by embracing technology, companies make changes in an attempt to get closer to their customers, with the hope that they may perform better in winning them over. In other words, it’s all about ROI of a better customer experience.
Consumer & Market Trends
So the trillion dollar question: What kind of experiences do customers want?
Well, we can start by analysing the trends pre-crisis, which the crisis itself accelerated: namely the push towards more instant forms of connectivity for literally all human beings on the planet. There are two main forces at play here:
- Technology
- Market Demographics
Technology:
We live in a world where it’s clear that whatever it is that we want, companies are out there trying to get it to us as quickly and efficiently as possible. For them it’s logical: the faster we get what we want, the more money they make.
Be it easier ways to stay connected via social media, ordering products which arrive the next day, streaming shows and films instantly online, summoning cars to ship you to where you need to go, or renting some strangers house without ever even talking to another human being. Technology in that sense has made modern living a piece of cake (of course all things being relative).
However, what many are starting to realise now is that it’s not just about getting customers what they want as quickly as possible: it’s also largely about how you do it.
For many, there seems to be something missing from what technology now allows us to do. Let’s call it the human touch: an element that goes some way towards satisfying our fundamental need for person to person connections and social belonging.
Although everything is comparatively easier now, there still isn’t really a substitute to speaking with another person: something that many companies had already started to clock pre-crisis. No matter how advanced your ‘chat bot’ is, or how many ways you allow customers to get what they need without speaking to anyone in your company: people still want to deal with people. And they want to do it in whatever way suits them best.
Demographics:
So what ways suit them best?
The logical next step is to understand what the largest current consumer demographic is, and what it is that they prefer. There's a double-edged sword effect in terms of the benefits: not only are you more inclined to set your business up for success today, but by aligning yourself with the dominant consumer group now, you’re also in turn safeguarding its future performance. And right now, that’s Millennials. As baby-boomers head towards retirement, Millennials are coming into their prime consumer years - now making up 25% of the market (globalwebindex).
And here’s the kicker: Millennials, more so than any group before them, care about their experience as a customer. This becomes especially evident in how you’re set up to support them when there is either something they need, or something goes wrong. Up to 33% will consider switching to a competitor after just one poor service experience. That said, although the stakes have never been higher for getting it wrong, it also represents a huge opportunity for those who can get it right. As a group, Millennials are willing to spend the most (an additional 21%) for great a customer experience (American Express Customer Service Barometer).
What it means for businesses
On top of the shift to consumer trends in perceived service value, the pace of technology and innovation continues it’s relentless march towards outstripping existing business processes. Zoom, Slack, Microsoft Teams, you name it.
Everything is now geared towards connecting people digitally as people struggle to connect in the physical world. Customers are now being pushed more than ever towards a world of instant connectivity, always on, and real-time. And, if they need you, they are no longer willing to send in a request into the ether and hope for an answer; they want an immediate, effortless experience based on personalised, real conversations, with real people.
With such little human contact taking place in a pandemic rattled world, a significant opportunity presents itself for those who can instil a truly human touch to their products and services, which go beyond a ‘jump-on-the-bandwagon’ ad spot on how they are 'still here for us'...
The problem is: customers don’t care just about what a business has to say; they care about how that business acts, and what it does to back up their message or brand. If showing customers’ you’re still there for them is important, aside from launching a new ad campaign ask yourself what are you doing that truly signals to your customers that you are still there and that you do indeed care.
As always, actions will speak magnitudes louder than words, especially during difficult times. If your strategy for 2021 is to find ‘smarter ways’ of ‘deflecting customers away from support’ or ‘deploying chat-bot solutions’, you might need to think deeper about what kind of experiences these strategies will create for customers.
So to quickly recap, right now we live in a world where:
- Digital transformation means embracing technology in the hope to get closer to customers and win more of them over.
- Millennials are the emerging dominant consumer group, and represent a competitive opportunity for businesses who can win them over through personal, memorable customer experiences.
- The prevailing technological trend of instant connectivity and communication has been accelerated by the crisis, placing strain on existing business processes.
- For companies looking towards differentiation, social belonging and interaction i.e. conversations, where people speak to another person directly, have become a valuable commodity and potential competitive advantage.
- Customers now prefer message-based conversational communication channels; and so if you’re not set up to let them engage with you via conversational channels, you’re already falling behind.
A new competitive advantage
For those who stand at the front line of a business and act as it’s support structure, the current climate proves to be one of the biggest challenges yet in the pursuit of customer experience holy grail: being a 'customer-first' company.
All of a sudden, contact channels are exploding with agitated and confused customers, who are facing extreme levels of uncertainty. For customer service leaders, this brings to light a deep underlying problem that has obstructed their ability to act quickly and effectively to adapt around these emerging consumer trends.
Even though data is comparatively plentiful within most customer service organisations, it is also comparatively opaque as there’s just too much of it which is unstructured text. Extracting clear, accurate, consistent and actionable insights from customer conversations at scale is simply impossible when you’re in the process of deciding where to focus on to improve the customer experience.
To make things worse, the proxies we've used as a gauge for a 'good customer experience' like Ticket Duration, CSAT, or Response Times are no longer effective in truly understanding the depth of a customer conversation, and how well we perform in the eyes of the customer in terms of getting them what they need.
So how do you stay on top of customer service performance trends and insights? How do you stay on top of measuring, assessing and fixing service quality issues based on how your customer conversations are going at scale? How do you do more with less, and simultaneously create a better customer experience which also reduces costs?
The answer lies in the data. Conversational channels like live chat are now the preeminent communication channel of preference for customers looking for support. However, although it’s the preferred channel for most customers, conversational data is long, text-heavy and near-impossible for any one human to read through at scale. All that adds up to a big challenge for teams looking to make the most of current trends and stay at the cutting-edge of customer-centric best practices.
How AI can help us analyse a conversation more accurately and consistently
So how on earth can you begin to get a sense of not just what each customer contacts customer support about, but what actually happened in those conversations?
The conventional methods being used today are unfortunately still quite basic. We are essentially limited to the basic approach of agents classifying case topics with tags that are collated in the Helpdesk or CRM being used for support and combining these with customer satisfaction survey feedback (CSAT) data.
The problem is, this isn’t a great way to understand a customer interaction, because at best, it only provides a view of what the customer reached out about, and not what actually happened during the conversations.
Even with CSAT, unfortunately most customers don’t answer the survey question, and getting a 20% response rate is pretty much as good as it gets across most industries. The proportion of people who leave open text/verbatim within that 20% being even lower. In other words, it’s nearly impossible to get a complete view of customer outcomes and experience in the context of analysing all of your conversation data.
There are a few obstacles that we need to overcome in order to effectively analyse conversations:
1. Language is and will always be subjective:
Although agents can definitely do a great job of tagging cases retrospectively / post interaction, each agent can interpret the same concept slightly differently.
If you were to give any two people a list of 100 independent conversations to go through and classify in terms of why the customer has reached out, without being able to talk to each other, there is a strong likelihood that there will be variance in the tags that are used to classify the conversations. This is because language is inherently quite subjective, and open to interpretation. It means that the list of tags that are used can grow exponentially in the mid-long term.
Duplicate tags and mistags from agents, means it's therefore much more difficult to get a bearing on what the 'data reality' is, and subsequently what should be done to improve upon the status quo.
On top of that, adding more agents into the agent-tag model exacerbates the problem which also makes it more difficult to scale as a process (I've personally lived this type of problem first hand at a big multinational early in my career when i worked in customer support - and it meant that no one really knew how to ever prioritise anything or what to fix or work on in a way that would create the biggest impact for customers - AKA the opposite of customer-centric decision making).
At a fundamental level, the key thing is not just having accurate data tagged in a reliable way, but also making the process scalable and the output easy for everyone to understand. Additionally, it’s no secret that agents' time should always be optimised towards serving the customer. The less admin they have to do, the more efficient their work is and more customer-centric you can become.
2. Key-words don’t provide meaning:
If getting agents to tag the conversations themselves isn’t the best solution, then what about using key-word searches.
Again, unfortunately we see quite a lot that can go wrong with this. For example, think about how many ways people can complain about price: “it’s too expensive”, “very costly”, “these are crazy prices”, “I can’t afford it”, “didn’t realise it would cost me an arm and a leg”, etc.
There are so many ways that people can describe their experience at a semantic level of language, that looking to just capture all the key-words will produce a very similar problem to before, namely: we’ll have a lot of words, but not much meaning.
Words on their own simply don’t provide enough meaning to derive value and a specific action to take - and that is fundamentally the goal of analysing conversations data in the first place.
However, by looking to understand language at the conceptual/semantic level, the words that people use begin to matter less, as we can gauge the meaning of what it is they are saying - i.e. that in the prior example the customers perceive that they aren’t getting Value for Money.
3. Contact reason topics + CSAT don’t give a complete picture
A big issue most folks come across is that only having contact reason topic tags, doesn’t actually provide them with a useful gauge of what happened during the interaction itself.
At this point, adding CSAT surveys into the mix would be the conventional wisdom, as customers will of course tell you everything you need to know about their interaction, after they’ve already had it… wrong.
Although CSAT can indeed be a useful barometer for quality, unfortunately once a customer has got what they need out of the support interaction, they simply want to be done with it. It means that a successful cross-industry benchmark for CSAT response rates today is around 20% of total support case volumes. That means that most companies literally have no idea about the quality of the customer experience of 80% of customer interactions with their support teams.
You also have to consider that the proportion of customers who are willing to leave open text feedback is even lower than the 20% - which means for most companies, all you really have is a Yes/No response to “Were we able to help you today?” question, or at best an arbitrary response to a likert scale survey asking “How would you rate your support interaction?”.
Enter Conversation Analytics
Imagine if you could read every single customer interaction your support team handles, and take from each conversation the most relevant insights.
You could in theory, start to piece together not only what drives a customer to get in touch with your support team, but also how well the team did in providing them with a positive experience i.e. gauge the quality of the interaction without any need for human intervention .
Over the past year, Chattermill has worked towards understanding what would help unlock the most value out of customer conversations data for businesses.
Having researched, tested and actually deployed several methodologies at the cutting-edge of AI text analytics, the results are clear:
a) Businesses need to understand what drives a customer to reach out to customer support: Contact Reasons.
b) Businesses want to understand not only the reasons for contact, but what actually happened during the interaction from the customer's perspective as a way to gauge quality and impact: Customer Outcomes.
They look to do this primarily for two reasons:
- To better understand the user experience
What drives a customer to get in touch with support is a good proxy for where in the customer journey the pain points could be. These insights are useful for not just support leaders who’d like to understand better ways of helping these customers, but also product, marketing, user experience and beyond.
- To better gauge the quality of the experience agents provide customers with during their interactions
Understanding quality is a key driver in particular for customer service leaders who are tight on time, and want an automatic way of assessing quality over 100% of customer service interactions.
Final Thoughts
AI is a term that is thoroughly overused in a business context nowadays, and one can safely say that the novelty and excitement attached to it has depleted since it became the in-vogue way of marketing a tool or service. However, true AI is real - and the possibilities it provides are ground-breaking in many areas.
By deploying state of the art artificial neural network models that are custom trained on a businesses support conversations data, we’ve been able to start unlocking these benefits for the most forward thinking and customer-centric organisations in the world.
We call this new approach Conversation Analytics, and believe it’s here to shake-up the conventional best-practices of the customer support industry.
We’ve only just begun this journey, but it is a truly exciting time to innovate on an age-old problem in business: how to improve customer experience and business performance at the same time.
If you’d like to hear more about how Chattermill is innovating on customer experience best practices through Conversation Analytics, get in touch through our website here.
