Internet search engines today are facing problems in keeping upw ith the pace of web growth. Two facts are responsible: bandwidth bottlenecks due to central indexing; deep web (or invisible web) contents that are inaccessible for search engines. Powerful and flexibly extensible object-oriented frameworks are available that assist in the implementation of distributed search infrastructures, thus addressing the first problem. In order to address the second problem, searchability has to be designed into the online applications constituting the deep web, and integrations to the distributed search infrastructures have to be implemented. A model-driven approach to software construction can be used to specify an application’s searchability. This paper presents an extension to the UML that can be used to specify an application’s searchability in an efficient way. The resulting models can be used to generate large parts of the searchability implementation automatically.