Requirements engineering is a systematic and disciplined approach for the specification and management of software requirements; one of its objectives is to transform the requirements of the stakeholders into formal specifications in order to analyze and implement a system. These requirements are usually expressed and articulated in natural language, this due to the universality and facility that natural language presents for communicating them. To facilitate the transformation processes and to improve the quality of the resulting requirements, several authors have proposed templates for writing requirements in structured natural language. However, these templates do not allow writing certain functional requirements, non-functional requirements and constraints, and they do not adapt correctly to certain types of systems such as self-adaptive, product line-based and embedded systems. This paper (i) presents evidence of the weaknesses of the template recommended by the IREB® (International Requirements Engineering Institute), and (ii) lays the foundations, through certain improvements to the template proposed by the IREB®, for facilitating the work of the requirements engineers and therefore improving the quality of the products specified with the new template. This new template was built and evaluated through two active research cycles. In each cycle we identified the problems specifying the requirements of the corresponding industrial case with the corresponding base-line template, propose some improvements to address these problems and analyze the results of using the new template to specify the requirements of each case. Thus, the resulting template was able to correctly write all requirements of both industrial cases. Despite the promising results of this new template, it is still preliminary work regarding its coverage and the quality level of the requirements that can be written with it.