Websites quality models are important to rank presences in the web, to identify best practices or pinpoint poor implementations. This paper presents a preliminary step of a website quality evaluation approach based on fully automatic collection of quality metrics, grouped by ISO9126 quality characteristics. In this step, over sixty candidate metrics, mostly proposed by other Web Engineering researchers, were collected. Besides providing evidence on the feasibility of automatic collection, we present the results of an empirical study where we assess which metrics of website quality depend on the application domain. Using a modified web crawler and a database, we collected a sample of more than one hundred sites, organized in three balanced groups, each corresponding to a different application domain area (banks, newspapers and airlines). Then, we tested if the application domain had a statistically significant influence on the values of the collected quality metrics. Our findings point to the conclusion that roughly half of the selected quality metrics are domain dependent.