Ouvrage en collaboration
Résumé : Once seen as merely marginal data in e-government databases, email addresses have become a key strategic element in the digitisation of information and exchanges between partners. A study to evaluate and maintain quality is therefore essential. At first sight, quality management of email addresses in databases may appear to be without major complications. However, taking a closer look, it becomes apparent that email management is a highly complex task and it seems impossible to maintain a quality database. However, every cloud has a silver lining. Although email management is indeed a complex task, owing to a high level of uncertainty in various respects, email addresses present major advantages in comparison with other data, such as postal addresses and telephone numbers, in that their existence and can be verified without travel or picking up the telephone. Such steps can be carried out semi-automatically, resulting in a significant return on investment owing to improvements in the quality of email addresses. Furthermore, taking effective measures pays off: many projects have been successful, either in financial or organisational terms, in improving the quality of email addresses in databases. A deterioration in the quality of email addresses over time is the main factor resulting in a need for professional management. Our analyses have demonstrated that owing to the level of volatility in usage, less than half of email addresses provided a decade ago are still valid today. It is therefore necessary to put in place an effective management strategy, with the capacity to monitor events over time and an organisational structure capable of keeping email address lists up to date. It goes without saying that the active participation of users is essential. No effort should be spared in encouraging users to indicate changes to their email address. In terms of syntax, we have demonstrated in this report that a simple binary distinction between correct and incorrect addresses fails to capture the subtlety and diversity of the conventions in use by email provides (Gmail, Yahoo, Telenet, etc.). In order to achieve a balance between an overly rigid and a lax approach, a 'suspicious' category should be included, in order to draw the attention of the user or manager to partially incorrect addresses and put in place adequate management strategies. Furthermore, with regard to syntax, we have also shown that a number of email providers have adopted a much simpler syntax system than that recommended by standards, which allows for more precise checks and eliminates the risk of false positives or negatives. Taking account of specific syntax systems, examples of which are included in the annex facilitates improvement of the quality and precision of the tests. In the present report, we have not studied the legal aspects related to the use of email addresses (privacy legislation, probative value, etc.). We prefer to leave these issues to the specialists. The majority of tests proposed in this report have been implemented and their efficiency has been demonstrated, either in a 'Proof of Concept' developed internally, or in Data Quality Tools. However, as these results and proposals are new, they have not yet been implemented on a large scale, apart from the development and commissioning of a novel, reusable Java library with a view to testing syntactical aspects of email addresses. We firmly believe that professional management of business, client and lay email addresses will greatly improve the quality of data in comparison with the collection of information such as postal addresses, telephone numbers or other data. Users will have to be offered an incentive to keep their data up to date and all the recommendations set out in this document will have to be implemented. A significant return on investment will result.