In the past, point data only play a minor role in map generalization, as points are either already the result of generalization (by symbolization) or are used for objects which are only shown on large scale maps (e.g. real estate maps). Consequently, conflicts arise generally between point and line or areal data and not between different points. Now, with the growing availability of web mapping services the role of point data has changed: Besides route planning, the most common function of web maps is the visualization of user queries for points of interest (POI).  Due to the limited size of commonly used displays, the map extract is often rather small, which results in a smaller scale as would be appropriate for the maps content. At this, the state of the art to resolve occurring cluttered point sets, is on the one hand interactivity (pan and zoom) and on the other hand the selection of points. However, this selection is based on the ranking of the search results rather than on spatial criteria. Thus, often the available space is not optimally used. Therefore we propose a displacement method to improve the readability of dense sets of POIs.