The term “Artificial Intelligence” speaks to our collective imagination. We believe AI can make our dreams come true and all we need is a magic wand to open the machine learning box. Unfortunately, that’s not quite how it works!
But there is a silver lining, and it’s becoming brighter. The practical applications of AI are maturing, and new techniques are enabling us to push boundaries even further. For instance, in May 2017 Google’s DeepMind AlphaGo defeated a human Go champion. We now have algorithms that can colorize black and white photographs and paint paintings in different artist’s styles from Van Gogh to Picasso. AI is developing by leaps and bounds.
As a deep-seated technologist and software engineer, I have had a keen interest in where we are taking machine learning and AI. And finally, I’m very excited to see the trend emerge where I can swap those words around. I now have an interest in where machine learning and AI can take us.
We still don’t have a magic wand; we need people, smart people, and data, lots of data. We need to turn intelligent questions into concrete problems. We need to put our wizard hats on and imagine where we can go in our flying car.
In the field of CX, we have all the right ingredients to magnify the benefits AI has on offer. Through machine learning, we can unlock the secrets to address poor customer experience, and truly change the way our customers feel about our brand.
Let’s look at how machine learning can help us answer 5 key CX questions:
Q1: How can we find out how the customer really feels about an experience with our business?
Identifying how the user feels when going through any experience offered by our business is key to understanding what to action and how. A straightforward way to figuring this out is simply asking the question.
Working with unstructured text has been difficult in the past, we can’t possibly read every comment and codify it. Typically this problem has been addressed by having a customer select from many options in a feedback form to indicate which ones best describe their experience, an approach which has multiple pitfalls:
- Providing options can be suggestive and trigger the customer to select subconsciously insincere choices
- We force classification on our terms as a business, but to the customer, they may have a different meaning
- The real issue may not be captured by the available options.
Using machine learning we now have a powerful alternative to this approach. Allowing people to express their feelings and provide feedback using their own words is invaluable and provides much better outcomes for multiple reasons:
- Customers will write down the most relevant highs or lows of their experience, without bias from suggestions.
- Using machine learning we can identify concepts that are important to the business and use these categories in our structured data set.
- Exploratory analysis is a way to use machine learning to identify emerging trends and spot new and emerging themes the business might not yet be aware of.
- Machine learning provides us with the ability to extract sentiment – and magnitude of that sentiment – from the words provided by the customer.
Q2: How do we get to the bottom of our customer’s biggest pain-point?
Nobody likes to sit through 20 pages of questions. Unfortunately, many feedback surveys are still designed in this way. The business tries to get to the bottom of the customer’s potential dissatisfaction in as much detail as possible, so every question comes with a conditional branch and a rating.
Using machine learning we can bring our customer into a conversational experience, very similar to your favourite messaging application. The algorithms sitting on the other side will interpret the conversation and ask directed questions that will help identify the root issue in a natural and short way, keeping the impact of acquiring the feedback low. Survey fatigue is very real, and engaging with the customer in smarter, “more human” ways allows us to capture better quality feedback at a higher frequency.
Q3: How would our company be perceived if we take our customer’s biggest pain away?
Using predictive analysis, we can use machine learning to simulate the tendency of people to be happier if their number one pain point was taken away. We can then expand that concept and model the impact of addressing other or even multiple pain points and bring in the business case to action these, comparing cost against sentiment and loyalty.
Imagine we found out that our staff’s lack of in-depth product knowledge is responsible for 40% of our customer’s disgruntlement. We can put a business case together for staff training, but how much change will it effect? Is it truly the most efficient investment? Using predictive analysis, we can model this and find out that this would merely shift the needle by 5% because staff attitude is a strong driver of disgruntlement as well. Lucky for us, we can calculate that by including staff sensitivity training into our business case as well, we can shift the needle by 35%!
Q4: How can we put customer feedback in the hands of our business leaders, so it can be put to efficient use?
Being able to organically identify drivers of customer loyalty and happiness can be achieved with correlation analysis using machine learning techniques. These can be aggregated and reported on at the various levels in our organization and contribute as a key tool for our business leaders to drive iterative improvement.
Business leaders make informed decisions at a high pace. With large amounts of information in multi-page reports, the message often gets lost. Using machine learning, we can put the right information in front of our leaders in a contextual manner.
- The answer to the question “What are the 3 most important things to consider right now?” is different for a store, regional or national manager.
- “What is the most important insight I can actually action?” is constrained by the responsibilities and specific to the role the leader has within the organisation.
These questions can be asked again, every day, and they evolve over time. Lucky for us we have the technology to keep on top of them.
Q5: How can we delight our customers?
Machine learning can be used to drive a high degree of user personalization. Understanding our customer’s needs, and being able to predict the type of experience they will, or even want to have with the brand is one of the most powerful ways to use artificial intelligence. The data collected from previous experiences is the key to predicting future experience. This information can then be used to help our customers side-step the bad experiences, and be delighted by the great ones.
We are truly at the dawn of an exciting time where machine learning applications are opening new doors for us in the area of customer experience. Stepping through each of these doors opens up possibilities that can transform the customer experience and the shape of your organisation.
If you would like to learn more about how AI and machine learning could supercharge your NPS programs click here.