In 2016, among the 28 countries of the European Union, more than three quarters (77 %) of enterprises with 10 or more persons employed answered that they are concerned by online visibility and had a website or a home page. This result comes from the community survey on information and communication technologies (ICT) usage and e-commerce in enterprises. It is a mandatory survey conducted annually since 2002 on enterprises with 10 or more persons employed. The aim is to collect data on ICT in enterprises to supply indicators about Internet activities, connection used, e-commerce or ICT skills. In France, this survey is conducted by the National Institute of Statistics and Economic Studies (Insee). Reducing respondent burden, using administrative data or innovative techniques are becoming increasing issues: “non-excessive Burden on Respondents” (principle 9, European Code of Practice) “harness new data sources” (European Statistical System’s Vision 2020), “yet official statistics are an area that is constantly innovating, in order to reduce the response burden on individuals and enterprises who are surveyed, to make full use of administrative data” (Insee Horizon 2025). This research proposed modelling to reduce respondent burden using auxiliary information like administrative data which is one of the objectives of National Statistics Institutes. The aim is to model enterprises responses to items asked each year and remove them from the French ICT survey while providing national aggregates required by Eurostat. To do this, adapted, available and quality auxiliary information is needed. Six dummy variables to be modelled were selected from two modules : ICT specialists and skills, and access and use of the internet. To model questions about ICT specialists, the auxiliary information from the annual declaration of social data (DADS) is very useful. In this document, employers supply information like the description of the job using French classification “PCS-ESE” for each employee. In 2016, for the question about the employ of ICT specialists, the variable constructed with the DADS matches with survey responses for 80 % of enterprises. If we improve the ICT specialists classification it may be possible to remove this question. To model website characteristics, the answer to the question “during the last year, did your enterprise receive orders for goods or services placed via a website or “apps”?” is used. Also, when available, using the answers given by an enterprise in the previous edition of the survey improves the quality of the estimated models. But for this module, the lack of auxiliary information and the answers instability between different surveys editions make the estimations difficult. This research may be a starting point to improve jobs classification or use innovative techniques like web scraping to obtain website characteristics (cost and efficiency remain to evaluate). Moreover, it show why it is difficult to model survey responses: do we want to model the reality or enterprises responses? Finally, model responses to a community survey open up a scope of possibilities: it would be interesting to try to reproduce this research on others countries, using their own data sources.