Web access panels have received a widespread use in the survey industry for cost-effective surveys. It is not clear, however, if web panel surveys are plausible for high quality products. Can these surveys be adapted in their design to meet the quality requirements of an agency producing official statistics? The self-selection problems of web panel surveys, arising from the recruitment method and the nonresponse, obviously need to be counteracted. This paper presents the ordinary methodology used for web panel surveys as well as a few ideas for combinations of web panel surveys and traditional surveys in order to achieve a more controlled inference situation.