The paper assumes data of various web shops to be available. These data are used to calculate price index information. The assumption is that at the lowest level no turnover information is available. At a higher level of aggregation proxy turnover information may be available, so that price developments for various articles can be weighted according to economic importance. The paper concentrates on methods to calculate (unweighted) elementary price indices, using stratification of the articles. The method is applied at the subgroup level and yields price indices at the group level. These can in turn be aggregated to the group level, using weights. This step is not considered here, however. The method starts with price comparisons at the subgroup level and aggregates them to the group level. The price indices calculated in this fashion need not be transitive. They can be transitivized, however, as is shown in a companion paper (Willenborg, 2017). The methods described in the present paper are a first step in the process to produce transitive price indices for web shop data.