Over the past 30 years, self-administered establishment surveys have increasingly transitioned away from using mail to more online and mixed-mode data collection. To examine the potential impact of this transition on survey participation, we evaluate several mail and web single- and mixed-mode designs implemented experimentally in a large-scale job vacancy survey. We find that neither response rates nor nonresponse bias significantly differed between the alternative designs. Subgroup analyses revealed that establishments of all size classes showed a preference for the mail mode in the concurrent mixed-mode design, but larger establishments were more likely to participate via web than mail in the single-mode designs. Potential cost savings (over 50 percent per respondent) were evident when utilizing the web mode in either a single- or sequential mixed-mode design. Qualitative follow-up interviews indicated a general preference for the web mode due to easier handling, smoother collaboration between colleagues, avoidance of a cumbersome mail return, and being seen as a modern sustainable solution.