AES Eletropaulo’s evolution in relationship management


AES Eletropaulo’s evolution in relationship management

Energy distribution utilities are in possession of information about their customers – such as consumption history and location – that is extremely useful in preparing profiles that identify the economic, social and behavioural patterns of those customers. This allows energy utilities to understand more about the levels of service that customers require, and to establish the value to the utility of a single customer or a group of customers. Utilities can – and should – use this information to prepare solutions tailored to customer needs, but often this does not happen in practice.

The generation of such a knowledge base, which is developed from proprietary indicators associated with external databases, constitutes a rich data repository for the creation of predictive models that can be applied to the most diverse business and support programmes for utility management processes.

The main marketing drivers in an organisation should be pursued by means of a realistic strategy and aligned with the financial and image building goals. The search for alignment must be supported by a set of quantitative marketing goals over the short, medium and long term, such as commercial loss reduction, identification and management of bad debt, enhancement of the profitability of the existing customer base, widening the segmented customer base, and cost reduction.

A company’s desire to broaden its knowledge base, based on the integration of both internal and external information sources, enables the generation of new and efficient support elements in marketing management. Such a situation enhances the strategic repositioning of the business processes within the company’s market, and also enables the creation of a differentiated intelligence base via the generation of a structured behavioural and energy consumption profile database.

Operating within the framework of their customer relationship strategy, AES Eletropaulo, one of the largest energy utilities in Latin America supplying electricity to 5.3 million households in the city of Sao Paulo, has been developing complementary market segmentation projects since 2005, using both traditional and behavioural approaches. The Payment Behaviour Customer Classification Score (Receivable Risk Score) and the ‘Clusters BT’ are examples of such activities.


Faced with the requirement to minimise the impact of bad debt on company results, mainly caused by the B group (low voltage) customer in the private sector, AES Eletropaulo has identified the importance of introducing scores based on customer payment behaviour. The main goal is to create a credit score for each customer, based on his payment behaviour over the last 13 months. Such a score is achieved via a weighted average drawn from 11 variables, namely:

  1. Payment of energy bill by the due date
  2. Payment of energy bill after the due date but within a 29 day period
  3. Payment of energy bill 30 to 59 days after the due date
  4. Payment of energy bill 60 to 89 days after the due date
  5. Number of outstanding bills with an overdue date greater than 90 days
  6. Number of outstanding bills reflecting regular consumption
  7. Number of outstanding bills reflecting irregular consumption (fraud)
  8. Number of disconnections carried out
  9. Number of self-reconnections
  10. Whether an acknowledgement of debt has been signed
  11. Number of instalments outstanding in terms of the acknowledgement of debt agreement.

Each of the 11 variables contributes to the final score allocated to all AES Eletropaulo customers. The higher the score, the higher the customer is ranked as a possible bad debt risk. The risk level is described as follows:

Risk diagram.JPG

The implementation of the receivable risk score system has enabled the company to segment its customers, develop relationships, and differentiate between bad debt prevention actions. It has also allowed us to define negotiation policies and strategies.

An advantage of the receivable risk score is its applicability in discriminating analysis algorithms, which are an excellent credit management instrument. They allow the utility to assess customers’ risk history, and help analysts decide what action to take, depending on the customer’s risk assessment. For example, if a customer has a low risk profile, we may decide not to take immediate action if a bill is unpaid. On the other hand, we may implement different policies and strategies for customers with a high risk profile that are focused on migrating and/or reducing the bad debt, such as:

Eletropaulo customers.JPG
  • Specific, shorter-term communication regarding actions to be taken.
  • Energy supply disconnection (disconnection followed by device removal).
  • Enlisting the customer in the credit protection system.
  • Requirement of collaterals in the negotiation process.

Finally, the receivable risk scores should be used as a mechanism for structuring the company’s credit and collection policies, which should be based on actions needed to minimise credit losses while maximising each customer’s potential.

Customer segments.JPG


Technical, commercial and operational processes are traditionally hard to integrate at energy utilities. ‘Clusters BT’, a partnership with CPqD (Foundation Center of Research and Development in Telecommunications (Brazil) has developed a specific behavioural segmentation methodology aimed at characterising low voltage customer segments at AES Eletropaulo.

The project was targeted at maximising the usage of commercial and technical data from several of the company’s management processes, using the Two-Step Clusters algorithm in order to perform the segmentation. The methodology, defined by Fayyad around the knowledge discovery process (KDD – Knowledge Discovery in Database) focuses on acquiring knowledge which up to now has been ‘hidden’ in huge databases. The starting point is the predefined business goals. This innovative methodology can be used by any energy utility, from the identification of the data system (database) up to the data mining techniques used for the identification of standards, associations and customer profiling.

The main innovation is the geopositioning of all premises served by AES Eletropaulo. By means of special analysis starting at the customer’s geographic location, the methodology integrates external information for the segment characterisation, such as the micro data from the 2000 demographic census and the ABRADEE (Brazilian Association of Energy Distribution Utilities) Research on Urban Residential Customers, conducted in 2005. The receivable risk score mentioned above is also used as a database describer. Figure 1 shows the superimposing process used for external source data regarding the customer behavioural segments.

The first step was to focus on AES Eletropaulo’s LV customer segment. The business objectives were mainly aimed at understanding these customers, in order to improve communications and to identify the needs and problems that customers have. The analytical mapping process has begun; it consists of identifying the data required for reaching the goals previously defined by means of data mining techniques.

Fourteen clusters have been identified, enabling the generation of new knowledge and market estimates and definition of solutions, appropriate services and products for each segment. The process supports utilities’ financial loss reduction programmes and results in improvement of revenues. Additionally, the identification of consumption habits and patterns for each customer segment has allowed us to improve our knowledge of key customer consumption profiles, enabling the adoption of specific commercial actions and/or differentiated service offerings. Figure 2 shows the clusters that were identified.

The incorporation of secondary source data, the creation of geographic influence models (GIS) and the search for patterns and trends enable a more thorough demographiceconomic- social characterisation of customers and the utility market. This provides a higher level of efficiency when actions such as research on fraud, bad debt/default/delinquency assessment, and improved network planning are undertaken. It also helps to correct discrepancies in the customer database. AES Eletropaulo has built on this approach by incorporating technologies that are seen as relatively new within the energy sector, aiming at improving its relationship with customers.