As AI evolves and new customer service trends enter the picture, the industry’s speed limit of change is picking up. This may feel as exciting as it is daunting.
Is automation going to replace human interaction? How will natural language processing affect customer experience? Are any cats safe when cucumbers exist in the world?
‘Companies everywhere are realizing that their service is not measured against that of the local sausage factory but the best global brands of the world, which have now become part of daily life.
Therefore a strong incentive exists to get your act together – if you have not already.’
Martin Kõiva, CEO
Maybe you don’t believe in fortune-telling. Luckily, our resident expurrts have some experience between their ears and are willing to impart their thoughts on the challenges and opportunities facing the industry.
Well, do they have psychic powers?
“Kibbles and yarn.”
The Absurdist – Martin Kõiva
“My sixth sense is knowing how machines ‘think’ – creating a model of any space or process so I can automate the ‘finding of things’.”
The Scientist – Mervi Sepp Rei
“Human communication is a complicated phenomenon.”
The Linguist – Kristel Uiboaed
“One thing is always certain: mistakes may happen, as will learnings. It always comes back to humanity, and that’s a value that can’t be overstated.”
The Humanist – Valentina Thörner
“Intuitions are reasonably explained as the use of known senses paired with intellect, experience, and evolutionary developments.”
The Empiricist – Martin Ojala
How is customer service changing for the better?
Ease of use
Ojala – Technology is making the experience of getting and giving assistance more efficient. You can easily send screenshots or even get on a screen-sharing call that will help the agent assist you. Caveat: to be more accurate, this is my expectation with any decent service provider nowadays. But many companies still fail in this respect, and they’re unlikely to get my business moving forward.
Valentina – Help your customers to find answers without depending on your support team. Companies who create an easily accessible knowledge base have a clear edge here.
Mervi – We know more about human-computer interaction, and clever companies have translated this knowledge into digital customer support. Keeping the communication going as the thread moves from one channel to another is still a bit clunky but improving.
Smarter AI practices
Kristel – AI reached the peak of its hype several years ago. This was because very few companies had real-life experience with machine learning applications. Industries have learned to be more critical and realistic. Also, I think the customer service field will become more informed with the experience. And can more realistically assess what actually makes sense to automate.
Valentina – Automated chatbots can’t solve most problems. Instead, they are excellent information gatherers, never getting tired of asking the same set of question for every new conversation. Meanwhile, your support reps have all necessary information as soon as they take over, continuing the conversation where it matters.
Human interaction
Ojala – It seems that chats have made the experience less formal, which really makes it more pleasant for both sides in most cases.
Mervi – All the capitalised personal pronouns, lengthy essay-like structure, very official language – it puts the receiver in a position that you’re first a customer and then a person – kind of objectifying. It’s especially creepy when it comes from a bot or an automatically generated message. This is changing and this change is good.
Kristel – Companies are learning that human communication is a complicated phenomenon to automate and they might return to more ‘human’ solutions. I think quality human interactions are becoming more and more a competitive advantage.
Valentina – There’s a reason as to why two of the most popular Klaus rating categories are “solution” (technical) and “empathy” (psychological). Companies who want to excel in their customer relationships need to hire appropriately and pay accordingly. And honestly, it makes me happy to see well-paid customers support reps because it is THE metric that ascribes value within companies.
What are the biggest challenges that lie ahead in customer service?
AI excites, but companies aren’t versed enough in Natural Language Processing and Machine Learning.
Kristel – There’s a lack of experience and awareness of how the potential of NLP can really add value to the customer service field. Solutions in this field are still often rather primitive in the linguistics sense. And people often underestimate the extent of linguistic and natural communication in final solutions. NLP problems are not just regular ML problems – people need to learn more, ask more relevant questions, and build more useful applications.
Mervi – I’ve been doing machine learning in customer communication for six years. The biggest difficulty in automation or machine learning models is not the science or model accuracy, but how this automatic decision-maker with its uncertainty and errors is incorporated into the user experience. Companies need to keep a cool head about data specification and automation and realize that there are things that machines cannot do.
Ojala – Bots are excellent for people with simple questions who have not bothered to search through the knowledge base or other public sources. On the other end of the spectrum, if a person has scoured through 10+ articles on your knowledge base and then turns to your chat, it is probably not wise to throw your article-suggestion-chat-bot at them.
The aim cannot be to replace human agents. It should be to free them up from trivial questions, so they have more time to deal with issues that really need their expertise.
Businesses need to invest more in their customer service teams.
Martin – It is becoming painfully evident in many sectors that customer support is not just customer support but also sales, marketing, branding, customer success. Customer support quality is directly impacting business results as these disciplines blur.
There’s a flipside to rapidly increasing customer expectations: providing fast, excellent service to everybody at all times is VERY hard. Making that balancing act work is definitely making customer support now a much steeper challenge than 10 or 15 years ago.
Valentina – With ever more data available, it’s important for companies to decide what quality means to them. What do they want to achieve with their customer support?
If an organization isn’t clear about what customer service means to them, then it’s almost impossible to create a coherent strategy around resources, training and results. Ask any support leader how they define and operationalize quality – it differs from company to company. And that means copying someone else won’t work.
What is on the horizon for Klaus?
Ojala – More cleverness that will be useful for customers to identify conversations that should be reviewed and areas where they can improve. More features to organise the reviewing process and help agents learn from their mistakes.
Valentina – At the end of the day we want our customers to be able to make meaningful decisions based on real data. If your goal is leveling up your individual support reps, you need to be sure you are reviewing the right conversations. If your goal is improving and streamlining processes and practices, you need to be able to discover customer intent and bottlenecks that aren’t tied to individual support reps. If your goal is channeling meaningful feedback to the product or engineering teams, you need to figure out topics and trends as they arise.
Mervi – Klaus has the best data pipeline setup that I’ve come across anywhere – data and its processing is democratized, meaning the data science research can move very fast from prototype to production. Klaus’s data science ambition is to move deeper into analysing textual content for product application, forking to realtime analysis and voice transcriptions for research. Our ambitions will a) surface more valuable review content, and b) enrich the analytics for better business optics.
Kristel – We already have experience and knowledge with different types of NLP applications. We have realistic goals and an ambitious vision for the future; the pace of development is also steady.
- Sentiment Filter: analyzes 100% conversations to detect negative or positive sentiment
- Complexity Filter: analyzes 100% conversations to detect which are the most complex (tailored for each company)
- Conversation Insights: neat data visuals to help with conversation discovery and coaching
Tell us the most important thing you’ve recently learned
Mervi – I’m obsessed with finding things. I like to dig and explore and play. This is often tedious and there’s a lot of routine work but because at the end I always find something exciting – it is thrilling to me.
Ojala – I’ve learned to accept that I might not make groundbreaking discoveries or personal development leaps as frequently as I have in the past, and it’s OK to have periods where you cruise instead of climbing.
Kristel – Productionizing NLP applications is hard. But even harder is finding the mutual language to communicate potential solutions between the data scientists and the customers.
Valentina – Change is hard. People love to stick to what they know, even (or especially) when the world seems to be scarier and scarier. This is true when it comes to adopting remote-first practices, when you implement peer reviews for the first time, or when you change your kids’ breakfast regime because your parents didn’t buy the requisite stracciatella yogurt before your arrival.
Martin – How to pick locks.*
*Disclaimer: Do not use any of the information presented for illegal purposes.
