DESCRIPTION AND PURPOSE. In order to achieve the highest profitability of the available assets it is necessary to focus on vigilance and early warnings in the event of non-payment of the entity’s credit assets. We propose to develop a forecasting system for provisioning needs and optimization of the measures to be taken to avoid or minimize the consequences of defaults.
SOLUTION. Based on the data provided by the entity, such as the movements of assets and liabilities of individuals, the credit history or current portfolio, models are made to predict customer defaults in some of the products of contracted credits.
RESULTS. The predictive default detection rate obtained is between 77% and 83%, obtaining an ROI of several times the investment in just one year. Applications are deployed on the solution's web portal that allow performance monitoring, through adjusted KPIs, and APIs and batch processes (files), which enable quick and easy integration, without interfering with corporate systems of the bank.
DESCRIPTION AND OBJECTIVES. Project for the commercial development of the channels of HORECA and Proximity Food Services. The goals are the detection of commercial opportunities that can be materialized by the sales representatives, as well as the implementation of promotional management.
SOLUTION. Development of a system based on Artificial Intelligence and advanced web platforms that allows, through machine learning, to discover commercial opportunities in clients, detect the best commercial levers and materialize them, define and manage promotions, and decide the best ones for each opportunity and at each point of sale. Likewise, advanced functionalities are included to support new launches that do not interfere with the opportunities detected and the available promotions.
RESULTS. Increase of more than 20% in net sales in customers in which the promotions and opportunities detected by the optimization system have been applied. In addition, the life cycle of the promotions is managed by the "trade-marketing" department, integrating it with the back-office and the opportunities detected by the Artificial Intelligence system.
DESCRIPTION AND PURPOSE. For this institution the risk control is essential for its development, so they need to foresee which consumer loans for individuals will be paid on time,, which ones may suffer a delay in the payment of any of the installments and which ones will move into default. The scope includes consumer loans, including those associated with credit cards.
SOLUTION. The data available by the institution is used, such as movements of assets and liabilities of individuals, credit histories with the bank and other institutions, whose data are public, as well as the specific data of the requested credit (amount, purpose, term, etc.). Based on this collection of information, advanced systems are implemented for the predictive detection of credit defaults.
RESULTS. With the powerful graphical tool implemented, which is based on behavior maps, an efficiency in the prediction of “defaults” about 75-85% is obtained, with the reduction of costs due to prediction failures greater than 25%, compared to systems implemented by the bank.
DESCRIPTION AND PURPOSE. The project’s goal is to identify potential clients with a greater propensity for portability, maximizing the conversion of calls into contracts.
SOLUTION. In addition to the historical data available for automatic generation of portability propensity models, Gamco's solution handles more than 18 million monthly traffic records. From this data, advanced customer and call segmentations are carried out to identify the best business opportunities, in order to detect potential customers, always applying criteria aligned with the operator's strategy.
RESULTS. From the segmentation of all the traffic, thousands of potential clients are identified and, for the first time, a database of telephone numbers belonging to the clients that the system has automatically identified as profitable clients (private or corporate) is obtained according to the company's criteria, as well as clients with a high probability of achieving their phone's portability.
DESCRIPTION AND OBJECTIVES. Due to the heterogeneous origin of the data stored by the State Security Forces, it is essential to have powerful tools to infer knowledge from their intelligence databases, preserving confidentiality throughout the process.
SOLUTION. As Gamco's learning algorithms do not need to know the “plain data ”, but can infer knowledge from coded data, the maximum confidentiality is maintained in the data collected in order to implement models able to merge data from different information repositories, filtering and discarding erroneous, duplicate or incomplete data.
RESULTS. The models for detecting inconsistencies in the database to increase its quality have an efficiency of 98%, with false alarms being less than 0,5%. Predictive models achieve an efficiency of 84,6% in the prediction of certain events not yet committed, with a percentage of false alarms less than 4,5%.
DESCRIPTION AND PURPOSE. Getting a better knowledge of customers is required to integrate with your sales and marketing processes. For this project, both at a strategic and sales force level, we pretend to increase the number of orders from its current customers.
SOLUTION. The models, obtained from billing data, customers, products, surveys, etc., are integrated into the CRM. In this way, the system distinguishes between descriptive variables and commercial levers that can be used to increase sales and/or profits, indicating on which points of sale and levers it is better to act. The scope detects sales opportunities in more than 30,000 points of sale in the HORECA channel.
RESULTS. Se logran incrementar las ventas en un 6%. Los modelos detectan nuevos productos susceptibles de ser vendidos en puntos de venta donde aún no eran adquiridos, teniendo un 33% de éxito en los primeros 3 meses de uso en cuanto a la aceptación de las nuevas oportunidades detectadas. El desarrollo, testeo y despliegue en toda la fuerza de ventas se realiza en menos de 6 meses.
DESCRIPTION AND PURPOSE. The bank estimates an excess in cash at the main cash points (ATMs and branches), so the objective is to reduce the cash that circulates through the branches. Optimal cash management mainly enables, with other advantages, to reduce cash and improve the management and cost of cash transport.
SOLUTION. From the historical data of the real cash needs, the working calendar and other public social data, prediction models of cash needs adjusted to the specific reality of each point are automatically generated.
RESULTS. Firstly, management costs decrease by several million euros and the cost savings associated with fixed assets is always greater than 30%. Secondly, the prediction errors remain contained between 15-20%. Finally, a system for the automatic generation of predictive models is implemented, without the need for manual parameterization of the cash points.