Relevance & Research Question: In the transition to the digital economy, the implementation of e-commerce strategies contributes to reduce costs and obtain competitive advantages. Indeed, monitoring the adoption of e-commerce by firms is an important concern for national and supranational statistical offices, which include questions about e-commerce in their surveys to businesses. However, most of the information about e-commerce adoption is available online through corporate websites, which are usually public and suitable for being automatically collected and processed. In this context, this work proposes to build a predictive model to detect e-commerce availability from the automatic analysis of corporate websites.

Methods & Data: We have designed and implemented a system to automatically scrap and process corporate websites, which include techniques such as text and link analysis and Big Data learning methods. More than 150 website features were extracted and included in the predictive model. To test the system, we have built a data set consisting of 426 corporate websites of manufacturing firms from France and Spain.

Results: Results show that the proposed model manages to predict the availability of e-commerce with a precision about 85%. A more detailed analysis evidences that websites with e-commerce tend to include some specific keywords and have a private area.

Added Value: The potential of online data for providing fresh and precise knowledge about economic activities is enormous, as some incipient research is evidencing. In this vein, this is the first research work that proposes and implements a system for detecting the adoption of e-commerce strategies by scraping and analyzing corporate websites. Given the automatic nature of the proposal, a large scale monitoring could be developed, providing policymakers with prompt information for making decisions related to Internet and law regulations on this matter.