Scoring is the optimum accompaniment to credit checks. Here, we forecast the likelihood that your customer will pay the outstanding invoice. We weight and assess various data, assign points and ultimately arrive at an overall score. Scorecards for individual shoppers and sectors are taken as a basis here. This allows us to draw up an accurate prediction of payment behaviour – and then provide you with detailed recommendations.
Your shoppers enter data every time they conclude a transaction. We can use this data to produce an accurate prediction of payment behaviour. The key here is the quality of the scorecards used. A tailor-made system of rules and standards for your shoppers and sector delivers far better results than standard applications. After all, if the scorecards are not defined sufficiently selectively, the subsequent scoring results will not be accurate.
This can lead to loss of revenue for you - for example if too many customers are turned away or the payment default rate increases drastically. We therefore secure the quality of our rules and standards by analysing your shoppers based on all relevant criteria. This is an approach that is already trusted by numerous well-known companies from eCommerce and the mail order sector.
After performing an exploratory data analysis and definitively specifying the target variables (payment behaviour), we take suitable analysis, validation and test samples in order to develop the scorecard. All potential score characteristics are checked to determine their individual selectivity. Optimum grouping helps secure the maximum information and, at the same time, stabilises the prediction capacity of the individual score characteristics.
We then use a multivariate, non-linear regression model to develop the actual scorecard. The score characteristics are selected iteratively here, taking into account their intercorrelations. To avoid a distortion of scorecards and to improve the forecasting model, we use various analysis models to scrutinize the allocation algorithms for rejected cases.
The scorecards used are regularly analysed by our risk management specialists. Checks are performed here to determine whether the necessary selectivity is still present or whether adjustments are required.
Assessments performed for new customers are referred to as application scoring. Only the criteria available at the time of the first order can be used here. Dynamic enhancement by adding credit agency information therefore plays an important part. Surprisingly accurate predictions of payment behaviour can be made in this way. The precise score scale allows the cut-off value, i.e. the point above which customers are rejected, to be controlled very accurately.
With existing customers, the scope of relevant criteria for reaching a decision is significantly larger thanks to knowledge of their past payment behaviour. These criteria are incorporated into behavioural scoring in order to predict future behavioural patterns. "Critical customers" can then be identified in advance – and your limit policy adapted accordingly.
