How to authenticate a good writing of web is ambiguous. Our studies is trying to find the answer to this question. In this paper, we described three study on this issue. In Study one, directed tasks, we try to compare the reading behavior of technical and nontechnical users. Base on it, we have found some main differences between the two gruop which will help us know what's the reader need. In Study two, exploratory study of reading, we try to make the participants read amounts of text rather than simply picking out a single fact from the page. Through this research, we have found lots of factors affecting a reader in reading web article such as text should be scannable, concise,etc. In study three, measurement study, we asked the participants to read 5 variations of a web site all of which contained essentially the same information. Base on the findings in Studes 1 and 2, we use six hypotheses to do the test. Finally, our study suggest that scannable, concise, and objective writing styles each make a positive difference in Web users' proformance and subjective satisfaction.