Research, at its core, is all about solving problems.
Engineering Management and Systems Engineering (EMSE) Professor John Helveston was conducting survey research on sustainable energy technologies and noticed a problem—the proprietary software traditionally used to generate the surveys was expensive and generated non-reproducible surveys.
Helveston conducts research using conjoint surveys, which are designed to figure out respondents’ preferences through a series of randomized choice questions. In the field of sustainable energy technologies, conjoint surveys can be used to understand under what conditions respondents might want to buy certain sustainable products. The surveys involve thousands of randomized choice questions unique to each respondent.
Growing increasingly frustrated with the existing software options, Helveston was led to the idea that would eventually be surveydown, an open-source survey framework.
The idea began as a blog post in 2023, where he outlined the idea and called for help from the open-source community to build it. Although his idea garnered a lot of interest, Helveston did not find anyone to develop the platform with him until the summer of 2024, when Pingfan Hu, one of his doctoral students, and Bogdan Bunea, an EMSE undergraduate student, joined him in an attempt to build it themselves. The team started working together to build the platform, completing about 80% of the core functionality in a six-week sprint.
Helveston shared that Large Language Models (LLMs) like Claude and ChatGPT were integral to the team’s rapid development, helping to solve some of the critical bottlenecks.
“Being very experienced programmers, but having a tool we could throw crazy ideas at – it was the fourth team member,” Helveston explained. “Without an LLM, I don’t think we could’ve built this. Or certainly we couldn’t have done it in six weeks. It accelerated our development.”
In early August 2024, Helveston officially introduced surveydown in another blog post.
Each fall, Helveston teaches a survey-based course called Marketing Analytics for Design Decisions. In the past, he taught the course using existing platforms like Qualtrics. In the fall of 2024, though, he decided to use surveydown instead. Even though the platform had only been completed just weeks before classes started, he decided that the best way to test it was to use it.
“It was a little like the wild west because we were updating the package as we were using it, and it worked really well,” he said. “The class accelerated the platform from a prototype to a well-vetted piece of software,” Helveston shared.
The software has been vetted not only by Helveston’s students but also by over 5,000 users based on the number of downloads from the CRAN, the online repository that hosts published R packages. On the code-sharing platform Github, many early adopters supported the platform’s rise by providing feedback and even suggesting and initiating specific updates.
“It’s now community-led development. If you need something, you give us the idea and we try to make it happen. Sometimes, someone identifies the problem and gives us the solution right away,” said Helveston.
Users can post issues if they find a bug or problem, and either Helveston’s team or another community member can help build a solution. For example, a user requested the ability to translate their survey into different languages. That user built out several European languages for the team and showed the team their approach, enabling them to add or crowdsource up to 67 languages.
“That’s why open source is such a powerful way to build. You get the best solutions from all over the world – anyone can add to this,” Helveston remarked.
The platform has gained traction in the survey-building and open-source communities. The team authored a paper on the platform that was recently published in the journal PLOS One, and the team recently presented the package at several conferences, including posit::conf(2025) in Atlanta, Georgia, and the 2025 U.S. Research Software Engineering conference in Philadelphia.
Helveston and the team have been intentional and thoughtful about documenting their process in building surveydown. “Software development can be underappreciated in academia, but the impact can be huge. You build something and so many people use it, but it’s not a paper, which is the typical academic research product. That’s why it’s important to also publish a paper alongside the software, so users can learn about it and cite it when they use it,” said Helveston.
The team has bigger hopes for the platform, working with the GW Open-Source Program Office (OSPO) to see if there is a way to make it even more accessible to non-programmers who need a survey solution for their research.
One step in this direction was to create sdstudio, which is a companion package for building, previewing, and managing surveydown surveys through a Graphical User Interface (GUI), similar to the way traditional survey software works.
The team hopes this approach might help grow the user base beyond early enthusiasts and become a free solution that could replace the need for paid enterprise software for thousands of researchers.