Mobistar

Mobile phones or ice cream, you have to be able to measure satisfaction

CUSTOMER CASE

Industry Telecommunications

download .pdf

Mobistar was the second of the three mobile telephone operators to set up a network in Belgium. It now serves over two million customers, which represents a market share of more than 30%.

The Challenge

There is relatively high mobile phone penetration in Belgium, and competition - still a relatively new phenomenon in the country's telecom sector - is resulting in a dynamic market, where customers are often prepared to chop and change: the notorious phenomenon of 'churn'. Understanding the reasons for a customer to change operator requires being able to measure that customer's level of satisfaction, or dissatisfaction, with the service being offered. As anyone in marketing will tell you, it is notoriously difficult to measure how satisfied a customer is. How do you get through the subjective notion of 'how people feel' to reach the hard facts of 'what needs to be changed'?

The Solution

Mobistar's Network Quality Group used the SAS solution to process network information and customer surveys, and to construct a new algorithm, which provided it with a reliable measure of customer satisfaction. Placing satisfaction alongside network performance allows it to pin-point the areas which matter to users, so as to keep existing customers happy, and attract potential new ones.

Towards reliable measurement of satisfaction

Mobistar's Network Quality Group has linked up with its Technical Department to look at gaining a better understanding of what its customers want - and what turns them off. One way of doing this would be to look at the actual performance of the network, as measured on a daily basis, and set this alongside a survey of customer satisfaction. In this way, the company would learn what sorts of technical problems have a real impact on customer satisfaction, and what the relative importance of these significant technical problems is.

This is what Mobistar set out to do, in a project which involved three departments, all of which had stood to learn vital information from a better understanding of the customer's view. The Marketing Department is interested in how to attract new customers, or keep existing ones. The Customer Valuation Department needs to know who are the potential churners, and the Technical Department needs to know what to improve on Mobistar's network, and where to do it.

Central to building a picture of customer satisfaction is a survey, which asks a sample of customers:

• to rate how satisfied they are (1 to 5);
• to state whether they have experienced one or more of eight problems;
• to state how often they have encountered any of the eight problems (on a five-point frequency scale).

The combination of a soundly-designed survey, and thorough analysis driven by SAS/STAT presented Mobistar with a powerful tool for increasing understanding.

"It's not a methodology or an approach which is limited to mobile phone operators," explains Laurent Van Belle, Mobistar's Network Quality Project Leader. "Understanding and measuring customer satisfaction with a product or service is something which every company has to understand."

Laurent Van Belle likes to take the methodology out of the technical context of GSM technology (and avoid getting into market-sensitive discussions) by making the analogy with an ice-cream manufacturer. The Quality Project is looking to understand more than just the fact that problems occur. You need to know how much it upsets a customer if he finds ice crystals in his ice-cream, if the cone is broken, if the wished taste is absent, or if there is a trace of an unusual taste. You can then base the action you take to improve the product not only on a recognition that there are problems with it, but also on an assessment of how annoying these problems are. "Mobistar ice-cream rather than mobile phones might be a dream come true for some of our users," jokes Laurent Van Belle. "Except perhaps people who would still try to wear it on their belt!"

To calculate satisfaction, the starting-point is a carefully constructed survey which produces data on the types of problems customers face, and the frequency with which these problems occur. Mobistar's questionnaire was targeted at groups of users in areas selected on the basis of the network quality there.

"I couldn't get no…."*

Satisfaction ratings are usually calculated by combining this type of problem-related data with an odds ratio. Setting this odds ratio is the critical issue. All too often, traditional odds ratio methods are unable to combine different types of variable in a satisfactory way. This can lead to discrepancies, or even entirely illogical conclusions. "Algorithms which maximise likelihood do not make a distinction between a simple categorical variable and an ordinal categorical variable. This can lead distortions, or even the extreme of a customer who has a problem often being less likely to be unsatisfied than one having the same problem rarely!"

Mobistar used SAS OR software, with its sophisticated mathematical programming solutions, to construct a new methodology known as isotonic logistic regression, based on research carried out by top mathematicians on two algorithms. This new methodology team uses a much more reliable odds ratio, able to combine different classes of variables. It is able to deal with a combination of binary variables (does this problem exist?) with ordinal variables (how often have you encountered the problem?).

"Hey, hey, hey. That's what I say"*

For us it is indispensable to make a hierarchy of technical problems in terms of the effects these might have on customer satisfaction," explains Laurent. "This method allows us to analyse and interpret information much more accurately. We can use it to produce objective results from an objective survey of how customers feel. We can get past the subjectivity".

Mobistar is using the new methodology to identify types of clients it may risk losing, and the areas in which network quality might be a factor in this. Laurent Van Belle sums it all up: "It really places clients on the map for us, and helps us plan really effective changes in the way we use our resources."

Read More ...

• Visit Mobistar on the Web
• download the success story in PDF format.