During the COVID-19 pandemic, the retail sector experienced ten years of growth in digital penetration in a matter of months. At the same time, their customers are becoming more demanding, driven by customer-centric digital giants. The bar for customer experience (CX) is not set by your direct competitors, but by technology leaders like Amazon, Apple and Uber.
The resulting surge in data has not provided marketers and CX specialists with substantially better understanding of their customers and their behaviours, since their methodology for measuring and analysing customer experience has not kept up in most cases. Most companies are still relying on a survey-based metric, such as NPS, CSAT or CES, as their primary means of measuring CX performance. But, as commonly known, surveys have some serious limitations:
- Biased: survey records what people say that they will do, not what they actually do. There is even plenty of anecdotal evidence which shows that the very process of conducting customer satisfaction research is destined to decrease customer satisfaction.
- Limited: the typical response rate is between 5%-30%, and thus provides an extremely limited view of what customers experience and how your core processes are performing.
- Lack of insight in drivers: surveys often do not reveal the drivers of satisfaction sentiment.
- Reactive: surveys look backwards
Given the growing importance of winning in CX in order to stay in business as a brand, should you base your strategy on the seriously limited data source of surveys only?
The reliance on surveys in the past is understandable, since most companies simply did not have the data modelling capabilities to capture the behaviours of customers and predict satisfaction. But the current availability of data, thanks to digitalisation combined with the analytic possibilities that data scientists bring, make predictive CX possible for most companies.
Predicting Customer Experience
Developing analytics based on algorithms can reveal the actual drivers for customer satisfaction. Based on certain features in the customer journey of a specific customer, the satisfaction can be predicted. Think about variables such as contact with your service staff, questions asked in the Q&A section, delays in delivery, feedback on product quality, etc.
Moreover, the ROI of investments in customer experience initiatives can be calculated much more granularly, building up on a customer-by-customer level. So not only the data driving your CX strategy become much more complete and real time, it also provides a direct relation to customer value variables such a churn, average spend, etc. This way-of-working, linking features in the journey directly to the satisfaction and value of a single customer, can become the foundation for building convincing business cases for CX initiatives. This will help CX leaders getting support from finance and the executive team.
Next to the fact that this kind of analysis captures a much larger part of all customer interactions, the (near) real-time nature offers another big opportunity. Companies can start pre-empting possible pressures on satisfaction and value by predicting it and prompting automated action or by service staff (such as redirecting the customer to a live conversation).
How to Start with Predictive CX
Given the fact that this is quite a fundamental shift in steering CX operation at a company, moving from survey based to predictive CX can meet quite some resistance in the organisation. In most of the data science projects, managing the impact on people and processes is just as important to creating lasting value as data collection and model development. Also, there will probably be limited resources in the data science team and other limitations due to availability of data and legacy CX systems. So, you will have to start one customer journey at the time.
Start by selecting a journey that is important in terms of number of interactions or that has serious CX challenges. The project lead should be in the business, not someone from the Data & Analytics team. A senior executive should be a sponsor and help bridging internal silos (e.g., for data collection) and take away barriers when they appear. Once a prototype model has proven that satisfaction or customer value can successfully be predicted, and the most important stakeholders share the enthusiasm of the project lead, only then the further scaling of the modelling to other processes and journeys is prepared.
Predictive CX allows companies to take full advantages of the wealth of data that the recent digital acceleration has provided. It makes CX strategy truly data-driven and will help getting more attention in the board room due to the direct connection to business value.