Growing competition in the retailing of domestic utilities (electricity, water, telephone, etc.) means that companies must tailor their commercial offer to the needs of each customer, which means knowing their habits, their consumption, and the means by which the offer may improve the cost or service to the customer.
In the case of electricity and gas retailers, much of their customer’s information is available on the bills.
Our automated bill reading and structuring system enables tailored offers to be made in real time, making marketing campaigns and sales forces much more efficient with minimal effort by the end user.
This system also allows retailers to ascertain the competition for the customer and design offers and strategies to improve retention and expand the customer base.
European utility interested in automating the process of extracting data from the bills of different gas and electricity suppliers in order to make offers tailored to potential customers in real time.
Those bills have different structures. They may feature tables or plain text. They may be presented in color or black and white and have different lengths. They may be in different formats like pdf or png. Therefore, the difficulty comes from creating a single tool to structure the information of various types of bills. Due to those differences in the structure of the documents, it is important to read them using a key-value system, not as a simple text.
The selection and extraction of specific information must be clear, structured and immediate in order to speed up and consolidate the task of the sales agents.
Foqum has created an advanced OCR system capable of structuring information from the electricity and gas bills of different suppliers. The process was completed with the development and deployment of a Web application to query, control and upload data relating to the applicants.
After a one-month phase fine-tuning the proper reading of data, the tool is able to perform a well-structured upload in less than 8 seconds per document, with 99% reliability in personal information and 97% reliability in billing information.