3 Ways That Predictive Analytics Have Improved Customer Experience

The ability to predict actions is about more than greater profits


Using predictive analytics is not new, it is something that companies have been doing for several years to varying degrees of success. Use cases are plentiful and there are some exciting developments in the area, none more so than in improving customer experience.

When many consider predictive analytics, it is through looking at how to increase sales and improve margins for companies. The more cynical amongst them think that this is done through exploiting their customers. However, the truth is often far from this, instead predictive analytics is often used to improve customer experience. Below, we look at three examples of how this has been the case.

Suggestion Engines

One of the keys to the success of Amazon has undoubtedly been it's use of a suggestion engine. Here, the website can display products that their customers might be interested in based on their viewing patterns and previous purchases. The huge amount of money that the company turns over yearly can be clearly traced back to their use of a powerful suggestion engine. Upon expanded implementation across the site between 2011 and 2012, Amazon saw a 29% increase in sales, a jump far too big to be coincidental.

It is not simply in online shopping and purchasing new products that suggestion engines are powerful, as both Spotify and Netflix have found. With both services it is possible to get almost exactly the same content through both legal or illegal means. Sites like Youtube allow you to listen to almost any album, and almost every film ever made will have some kind of pirated version available on the internet. What sets Netflix and Spotify apart is their customer experience and the ability to suggest what watchers/listeners may be interested in next. The importance of these recommendations on these sites cannot be understated, with Netflix claiming that 75% of what people watch on their platform comes from their recommendation engines.

Insurance Companies

Nobody really likes insurance companies. Almost everybody will have had some kind of negative experience with one at some point, from overcharging based on oversimplified segregation, through to claims being unfairly declined.

However, using predictive analytics, insurance companies are looking to decrease the number of negative experiences that customers have.

A well known and widely used element of this is through the use of in-car devices to log the merits of individual drivers. Progressive Insurance found that 'After analysis of billions of miles in driving data, Progressive has found that key driving behaviors—like actual miles driven, braking, and time of day of driving—carry more than twice the predictive power of traditional insurance rating variables, like a driver’s age, gender and the year, make and model of the insured vehicle.' This tracking then saw those who agreed to be tracked receive a 10-15% discount of premiums.

There are also several examples of how predictive analytics have been used to make accurate claims, preventing nefarious individuals trying to defraud insurance companies and increasing the premiums for other customers. Having the ability to track historical actions allows claims to be fairer and more accurate for each individual case, meaning fewer unhappy and frustrated customers.

Fraud Detection

Fraud has the potential to be hugely damaging for a wide variety of people. This could be from any one of the 9 millions individuals in the US who have their identity stolen every year, through to the US Government, who lost a predicted $3 trillion between 2000-2010 from tax fraud. Regardless of where the Fraud occurs, it will damage people, if it's in the financial sense of losing credit card information or having funding pulled on an important state project.

However, predictive analytics is at the cutting edge of fraud detection for individuals and organizations across the world. Credit card companies can look at your purchase locations, what you have been buying and at what time, then if purchases are made outside of these parameters it will be flagged to make sure that you don't lose money. The amount of fraudulent activity that has been stopped in this way is difficult to work out, but the regularity in which banks pick up anomalous activity suggests that it has been a considerable amount.

However, more can be done in the area and as analytical solutions improve, so will the security of customers. The total amount lost due to identity fraud in the US has doubled in 4 years, from $13.2 billion in 2010 to $29.35 billion in 2014. The average financial loss for each individual targeted is $5,130, showing how it can be a huge problem for customers. The more that predictive analytics can do to stop these kinds of activities, the better off every customer will be in the long run. 

University lecture small

Read next:

How Are Higher Education Institutions Using Analytics?