Request a Demo

Job Role
Inquiry Type

YES, I agree to receive, via email, information about Medallia solutions and success stories that show how industry-leading companies improve the customer experience and increase revenue. Read our Privacy Policy in the footer below.

Read our Privacy Policy

Thank You

Your message has been received and we will contact you shortly.

Want to Accurately Anticipate Customer Needs? Use Predictive Analytics

Medallia: Predictive Analysis

Who doesn’t want to make better decisions, take better actions and drive better experiences? You can’t if you’re only running on NPS® scores from the 20% of customers who respond. You really need the full picture. Through data analytics and machine learning the NPS of the other 80% can be predicted.  

Medallia recently hosted a predictive analytics webinar with Brian Andrews, Senior CX Principal at Medallia, and the telecommunications multinational Liberty Global, and their market research partner Gemseek. They discussed how analytics and machine learning gave them a 360º customer perspective, empowering them to deliver astonishing business outcomes. This blog post highlights the steps Liberty Global took to predict the NPS® (Virtual NPS®) of the 80% who don’t respond, as explored in the webinar.

1. Data collection

Companies sit on mountains of data. Beyond the customer experience data collected (direct feedback from surveys) there’s a bevy of existing information (internal data) that’s crucial to driving experience improvements: indirect feedback, employee feedback, and operational data. For a comprehensive view, that data needs to be combined to give a 360 degree view of the customer. But not all data is created equal.

The quality of the data is important. Using a structured high-quality data collection system is the best place to begin — because what you put in is what you get out. This really applies to predictive analytics where data is being used to make predictions. If you’re using data that’s poor in quality, then you’re not going to get great answers. Cleaning up the data is a big part of making the next step successful, introducing the data to the model.

 

The quality of the data is important. Using a structured high-quality data collection system is the best place to begin — because what you put in is what you get out.

2. Training

Then, a combination of experience data and internal data gets fed into the model. It trains itself to find patterns in the internal data that matches specific NPS scores customers have provided. Machine learning analyzes in-depth behaviors, products, interactions, usage and attitude patterns.

From there, it builds customer profiles for each score from 0 to 10. After each training cycle, it identifies variables with the greatest predictive powers. It trains and retrains itself to identify and learn.

3. Predicting

Next, the model takes customer profiles it built and looks at them against internal data and predicts a virtual NPS® score for the 80% who have not responded to surveys. In Liberty Global’s case, NPS® scores have been proven to be an accurate predictor of future customer behavior. For example, when they contacted actual detractors, they succeeded in reducing churn by 34%, and when they reached out to people they predicted might be detractors, they reduced churn by 31%. That meant their model for predicting detractors was 92% accurate. Watch the full webinar.

Liberty Global and Gemseek were acknowledged by the Market Research Society, for their “innovative and pragmatic application of machine learning models” to reduce customer churn rates.

Liberty Global and Gemseek were acknowledged by the Market Research Society, for their “innovative and pragmatic application of machine learning models” to reduce customer churn rates.

Conclusion

Predictive analytics can help you gain the full picture of your customers, recalculate NPS throughout your whole customer base, reduce churn, identify products and services that require improvements, open upsell opportunities, and target promoters.

It’s key to driving growth — by delivering a much better, meaningful experience to all customers, whether they respond to a survey or not.

To listen to a recording of the full webinar, click here.


Subscribe to our Weekly Blog Updates