The company-client interactions are usually registered in a CRM (customer relationship management). When properly exploited, these data are very valuable as they allow for client targeted offers and improve the sales department performance. The input of new data to CRM in a company with a number of salespersons is decentralized. The lack of a central entity that controls the CRM inputs, speeds up the addition of data by salespersons, but it may also have a negative impact on the quality of these data. Typos, duplicated registers, or a lack of normalization are shortcomings that may complicate or even prevent from the exploitation of these data. The challenge is to design a tailor-made, intelligent algorithm that improves current and future CRM data quality, keeping it a decentralized system.