Testing web sites

We observed the same task executed by 49 users on four production web sites. We tracked the rates of discovery of new usability problems on each site and, using that data, estimated the total number of usability problems on each site and the number of tests we would need to discover every problem. Our findings differ sharply from rules-of-thumb derived from earlier work by Virzi[1] and Nielsen[2,3] commonly viewed as "industry standards." We...