Statistics Belgium has been using scanner data from supermarkets in the calculation of the CPI since 2015. The applied method is a version of the so-called “dynamic method” using an unweighted chained Jevons index. Unweighted means that the turnover information at the product level is currently not explicitly used. Incorporating the available turnover information explicitly into chained monthly index calculation (e.g. superlative formulae such as Törnqvist) leads to chain drift. However, using such turnover information could lead to a more representative index calculation therefore methods have been proposed that make it possible to use this information while calculating drift free indices. These methods are called multilateral methods, because they use information from more than two periods. These multilateral methods (GEKS-Törnqvist, Time Product Dummy, Geary-Khamis and augmented Lehr index) are evaluated and compared with the dynamic method. These comparisons and evaluations will be presented. It will be shown that the differences between the methods aren’t that large apart from the augmented Lehr index. To calculate non-revisable indices rolling windows have to be used together with various splicing and extensions options. These methods are also evaluating by applying them on scanner data. It will be shown that some of these options might still cause some drift. A final issue that will be highlighted is how product relaunches (e.g. same product smaller content) are dealt with. We currently combine text mining with manual verification, more efficient ways of creating homogeneous product groups will be examined. The results will show that to handle relaunches good metadata is necessary.