In-depth articles, interviews and debates covering local and global topics related to the Experian business expertise.
Managing risk in product bundles for mobile phone customers
Global Consulting Team
Experian Decision Analytics
Just as most telco’s start control of the risk associated with their mobile subscribers, the organisation throws a new challenge their way, product bundles or packages. So now it is necessary to calculate not only the risk of non-payment, fraud and the level of profitability associated with a mobile customer, but now this same customer may or may not have a landline, or digital TV, or fixed broadband. What impact does that have on the risk they represent? This article suggests a three stage approach to solving this problem.
The first stage is to make a risk and fraud assessment of the subscriber, regardless of type i.e. individual, business etc. From a risk point of view, this is a well understood process, involving elements such as subscriber age or time in business, time with the telco, number of other products held either at same person or same family level, payment history, credit bureau data, for new or returning subscribers, if this is available in the market, geo-demographic information such as Mosaic etc. All these many and varied characteristics are assessed, usually by using a scorecard, to give an indication of the risk of non-payment for Telco services. This assessment should be made at point of first application, and then on a regular on-going basis of at least monthly, unless intermediate trigger elements warrant a re-score is carried out sooner.
The second verification which should be made at subscriber level is an anti-fraud check. It is only this way that a Telco can differentiate between the perfect subscriber and someone impersonating the perfect subscriber. These checks work on the principle that some application details are harder to put in place than others, for example the same bank or card details being re-used for ‘different’ people, and the more anomalies that are seen in similar applications or the more similarities between supposedly different subscribers are strong indicators of fraud at point of application. This second check allows more confidence to be placed on the outcome of the risk assessment.
The second stage is to classify all the offers. This can be as complex as the Telco has the appetite for and the resource it has to put it in place and manage given the rapidly changing product and offer catalogue which most operators provide. At a minimum, there could be three categories, High, Medium and Low, where by default all offers not categorised fall into the medium risk bucket. Then the organisation will need to identify products and offers that by their nature represent a high risk (tablet type devices, the latest smart phones, unlimited contracts etc), and products that carry very little risk, (SIM only connections, fixed broadband, pre-pay contracts etc) to be placed in the low risk category.
This assessment also should be reviewed regularly to follow fraud and non-payment trends (i.e. fraudulent cards being used on pre-pay offers) and allow this knowledge to be fed back into and to update the product and offer ratings.
As a second level, ideally an idea of the profit associated with each offer should also be stored, i.e. the ARPU over term less any hardware costs and other overheads.
Unlike both of the above assessments which can be made pro-actively, this assessment is made on a more reactive basis at a certain point in time, either because the subscriber wishes to update their offers, or because the Telco has an outbound marketing campaign. The deal assessment brings additional elements into consideration, such as the dealer used, the channel employed, whether or not there is a port in of an existing number, term (unless this is always tied to an offer) etc. These elements are placed in this assessment as opposed to the customer one as they are not static and could change with every new request.
Again from a profitability point of view, these elements will also determine the SAC applicable to the deal, which can be used to refine the profitability equation of the deal.
Once the above, not inconsiderable, work has been completed, and the necessary feedback loops have been put into place to ensure the assessments remain predictive, all the elements are available to calculate a customer level view. Again here there are multiple options available.
This can be done by a class matrix approach: For example, a low risk subscriber with a high risk offer could give a medium risk combination, and this combined with a medium deal assessment would give an overall medium risk assessment. Or the deal and the offer assessment can be combined to give the risk of the package, and then this can be compared with the risk of the applicant, and if the applicant risk exceeds the package risk, remedial action to reduce the package risk can be suggested.
It would also be possible to use a point basis, using the scored outcome from the subscriber assessment, with niche scorecards reflecting the adjustments based on the offer and the deal assessment. Then based on the final points outcome, a classification or index could be overlaid if required.
In both of these bases, the global risk outcome, either as points or a class can be aligned with the monetary value associated with the deal. For example, a deal given a low profit margin for a high risk subscriber is more likely to be rejected than one with a high profit margin.
Finally a monetary approach could be taken, using the SAC and ARPU from the deal and offer assessment, factored by the risk that the subscriber represents. So a high risk customer with a net ARPU of 500€ would finish with a risk level of 500 x 1.25 = 625€, whereas the low risk subscriber for the same offer would be 500 x 0.20 = 100€.
The other significant use of the customer view is in gauging subscriber potential. After each monthly cycle, it is possible based on the above process to generate the maximum entitlement a subscriber is entitled to for each product: mobile, ADSL, Digital TV etc. This improves the customer service in case the subscriber walks into a retail outlet, or contacts the service centre, the sales team already know what can be offered and will be accepted.
Of course the customer view is not the end of the story. Having categorised the subscriber, an appropriate treatment path needs to be applied. Is the outcome of the assessment that the risk is acceptable, and the subscriber may have the offer requested; or as in the example above of a subscriber with a higher risk than the package requested, do terms need to be applied to reduce the risk, i.e. a deposit asked for; or is the risk too high altogether, and the whole application needs to be reworked, i.e. a different offer of lower risk? It would also be possible to suggest which level of risk would be acceptable, to help from a customer experience point of view.
To arrive at a level of subscriber assessment covering all products from a risk, fraud and profitability point of view is not a simple piece of work, and nor is it ever finished, as the feedback loops in place are equally important to ensure the process stays predictive. The organisation attempting this task will need significant time and resources put in place and because of the widespread impact of the project, it would need to be supported as a strategic objective with suitable management level support. However with Telco organisations becoming more and more complex, it is the direction the subscriber credit risk management is going.