Building Customer Trust by Melding Human Insight with Machine Learning

Dec. 11, 2017

BY TARA MATAMOROS CARTER AND CHRISTIE LI

 

It’s 2018, but companies are still missing one critical element of machine learning: Human Insight.

Admit it. A significant aspect of your satisfaction as a customer depends on emotion. Emotion, and in larger scale, human insight, is crucial to a customer’s overall experience when interacting with a company’s services. So, with human insight’s increasing presence in all industries (especially technology), companies must be aware of the role human emotion plays in machine learning.

In this blog post, we expand on the relationship between emotion and machine learning. To do so, we provide a deeper look into machine learning, provide 4 key steps for companies looking to blend human insight with machine learning, and finally, explore the how the customer experience remains important even in a world consumed by machine learning.

I. Machine learning should be more than a buzzword

It’s tempting to nod our heads and accept the term “machine learning” as yet another Silicon Valley buzzword. However, we at Tangelo advise you to dig a little deeper - that way, you can leverage the craze over machine learning to your advantage by prioritizing consumer trust in your company’s services.

Say you’re making a transaction, a process that commonly involves money. From the perspective of a successful company, transactions are ordinary events. However, put yourself in the customer’s shoes for a moment. From your perspective, this may seem like a short minute interaction; however, for the customer, this transaction is actually a major event. Your clients invest a great deal of thought into deciding where and how to place their hard-earned savings.

Attention and a personalized experience not only makes the customer feel more secure and at ease with letting go of their hard-earned savings, but also provides an opportunity for you to infuse the unique flavor of your brand into the user’s experience.

It is important to remember that despite the efficiency and added insight that machine learning brings, the human factor cannot be neglected. This factor helps establish trust and credibility in an otherwise depersonalized world.

II. A deeper look Into machine learning

Now that you know the value of prioritizing customer emotion in an otherwise automated world, let’s combine the two: In this portion, we explain how companies can use machine learning to efficiently glean insights about their customers, while also ensuring that the company’s biases do not interfere with these insights. By using a specific branch of machine learning - “supervised machine learning” - companies are able to learn from previous associations. As Paul Garbett writes in Harvard Business Review:

“ Nobody can write or articulate all the rules for classifying all things, and they certainly can’t document all of the ways human emotion is expressed. As humans, we learn, classify, and act based on pattern recognition and past associations. We make lightning-fast assumptions based on patterns, purpose, and context.”

III. How to measure emotion using machine learning

We often don’t think of emotion as quantifiable. However, with machine learning, we can set metrics to evaluate customer’s experiences. One way of doing so is assessing positive and negative sentiments, which can be measured retrospectively. This way, you are able to focus your attention to the negative sentiments and understand why your customers may be feeling unsatisfied. With these metrics, you are able to analyze emotional data. One tool to leverage in doing so is Plutchik’s Wheel. This tool, a wheel of eight primary emotions, allows responders to accurately choose a representative emotion, rather than feel restricted by a survey question. Using customer experience (CX) metrics to drive improvements, your company can efficiently identify unsatisfying customer experiences by measuring negative emotion.

Plutchik’s Wheel of Emotions

Wheel of Emotions.png

IV. Limitations of Machine Learning

It remains crucial for companies to continuously audit their algorithms to search for potential biases that could provide a faulty response. Additionally, your customers are (for now) humans, not machines, so their experience is subjective. For example, their response could depend on whether or not they were having a difficult day, impacting their experience. As Adrian Swinscoe, a Forbes contributor puts it, machine learning has it’s limitations.

“Instead of trying to design emotion into their customer experience, companies should focus on making their customer experience easy, pleasant, consistent, proactive..”

An important component to machine learning goes beyond integrating human insight into machine learning, it involves valuing the customer experience.

Conclusion - Two takeaways: 

1. Companies must be flexible regarding how and where data is collected. As Harvard Business Review claims, “but often the greatest insight is found in spontaneous conversation with customers—not in the online survey that shoppers are asked to complete, but in the photos they take, the tweets they post, and the advice they offer in online forums.” Data is not only found in the traditional format of survey responses, but also in these everyday artifacts of the customer experience.

2. In a world consumed by machine learning and artificial intelligence, we can always go back to the basics. Strive to create a desirable customer experience, not simply for the sake of machine learning, but also because it is a time-tested way to generate loyalty and returns.

BY TARA MATAMOROS CARTER AND CHRISTIE LI
CUSTOMER EXPERIENCE, CONSUMER TRUST, EMOTIONAL BRANDING, HUMAN INSIGHT, MACHINE LEARNING