Improve Customer Experience with Machine Learning
Aim for simplicity in Data Science. Real creativity won’t make things more complex. Instead, it will simplify them.” --Damian Duffy Mingle
If you want to improve customer experience, then measuring the omni-channel customer journey at an individual level is essential.
Machine learning’s (ML) many applications make it a powerful tool in creating amazing customer experiences. The ML approach enables true digital innovation for customer care that has the potential to deliver world class customer experience if understood and applied correctly.
What are the key takeaways for me so far? The first is maybe the term ‘machine learning’ creates misconceptions when compared to what it means. The obvious one being that the task of machine learning is to think and make decisions like humans – this is incorrect. It’s “applied statistics and algorithms to identify probabilistic relationships in data”
Machine learning is about learning patterns in data and it's good at predicting the future when the future looks like the past - I did a double take on that one!
Once deployed and over time, ML models become less accurate until they are re-trained.
Accuracy is not a good measure of model performance and ML learns whatever biases you and your data have. I am learning this myself having done a data science course for machine learning, but I've quickly realised that business leaders haven't grasped the opportunities and capabilities to transform the business of their companies.
Having spent more than 15 years being obsessed with perfect service and digital empathy, I can tell you that the prospects of machine learning to improve customer experience is a game changer – and should be taken seriously if companies want to truly get ahead of the competition and avoid being left behind.
Here are some well known examples where machine learning has been deployed successfully:-
There’s many more, so check out Pinterest, Wells Fargo, Starbucks, Instagram, Nike, BMW, North Face, American Express, and Disney.
Three common uses of Machine Learning
These are generally applied as part of a working platform (not necessarily bespoke) that can partner with companies to get them started with data segmentation and customer engagement being just two examples of many. From my understanding, it is important for data scientists to know when to use classification versus prediction and how to use conditional probabilities in the context of machine learning.
Applying Machine Learning to enhance CX?
The process outlined below in practice should be circular. A cycle of continuous improvement where enhancements to CX can be realised through business process optimisation. This is also a key aspect of the newly launched version of ITIL4.
The diagram below has been used before machine learning even came about, and has now been adopted with Machine learning to accelerate progress and outcomes for improved CX.
For example, reports are generated to measure performance, scope and realisation of customer activity and trends which can now translate to predictive analytics. Learnings from these enable optimisation of processes and generates the building blocks to be established for service management – with less human interaction and higher conversion rates – of course, all of this has to be managed with execution through cross functional collaboration and stakeholder engagements which leads to delivery and monitoring of functionality and outputs.
The future of Machine Learning is unstoppable. It is not a passing trend, but a mega trend that will transform business and society. It is the fundamental building block of artificial intelligence.
As a starting point, there would be no harm is working with a partner to establish a set of machine learning algorithms whether to solve business problems, or improve business process or enable a differentiation in customer experience. There is a clear case for a symbiotic relationship between business leaders and data scientists to support the advancement of AI and ML.
Business leaders cannot know what ML is good for without knowing what it is and there is a clear knowledge gap. Without a good understanding, there are cases where ill-conceived ML projects were rolled out with impact to costs and time with negative impact on the bottom line. The usual fallout in addition to losing customers is staff turnover.
With thanks to all the CX futurists that I have discussed and shared ideas with, and to the data science and machine learning course I have attended that has enabled me to translate the capability into CX possibilities. I am still learning!
Author: Renee Kalia