Big Data is one of the key topics, around which the so-called data revolution has embarked. The increased usage of Big Data in the private sector has raised expectations from the public sector as well. On their turn, statistical offices have also embarked in innovative projects to benefit from the opportunities which Big Data can offer, simultaneously feeling an increased pressure from users to digitize their production systems. In order to balance users’ expectations to the real possibilities of making a good use of Big Data in official statistics, a series of attributes for Big Data, which should be present, for it to be a good fit for the purposes of official statistics, are proposed. Importantly, the classic attributes of Big Data – the 4 Vs: volume, variety, velocity and veracity – are insufficient to explore Big Data suitability for official statistics. As a first distinctive attribute, the scope, for which official statistics needs or may use Big Data, is different from the same scope of the private sector so as a former scope is decision-oriented analysis whereas a latter scope is an action-oriented analysis. Starting from this argument, attributes for Big Data for official statistics are proposed and exemplified through an application of scanner data in Luxembourg. Nowadays, Luxembourg (STATEC) is using bilateral dynamic index complication approach for scanner data on «Food» and «Non-Alcoholic Beverages» COICOP groups calculation, which implies that prices of goods of two consecutive months are taken into account with basket of goods resampled every month. STATEC is utilizing machine learning algorithms in items classification process to make classification process faster and almost automatic. Future goals regarding scanner data integration in CPI/HICP estimation are to include COICOP groups of «Fresh Fruits», «Fresh Vegetables» and «Alcoholic Beverages» into STATEC production system as well as to integrate multilateral index complication approach for scanner data for the above mentioned COICOP groups.