Characteristics and Challenges of Low-Code Development: The Practitioners' Perspective

Yajing Luo, Peng Liang, Chong Wang, Mojtaba Shahin, Jing Zhan

Introduction

Industry analysts like Gartner and Forrester have long predicted a massive surge in Low-Code Development (LCD) as a solution to the global shortage of skilled software engineers. By providing visual interfaces and pre-built components, LCD platforms promise to democratize app creation. However, while marketing materials often claim that "anyone can build an app," the lived experience of professional developers is far more nuanced. This study moves beyond the hype by empirically analyzing practitioner discussions on Stack Overflow and Reddit to define the actual characteristics, implementation units, and hurdles of LCD in the real world.

Methodology

To capture the authentic voice of the developer community, the researchers employed a multi-step empirical approach focused on two major hubs of discourse: Stack Overflow for technical Q&A and Reddit for broader experience sharing.

The multi-step empirical research process used to analyze practitioner discussions.

The process began with a systematic keyword-based search (e.g., "low-code," "zero-code"), identifying over 1,500 initial posts. Through rigorous filtering, the team narrowed this down to a high-quality dataset of 301 relevant discussions (73 from Stack Overflow and 228 from Reddit). Qualitative analysis was performed using the Constant Comparison method from grounded theory to iteratively code and categorize developer sentiment. This was supplemented by descriptive statistics to rank platforms, languages, and domains.

Implementation Units and the "Escape Hatch"

A central finding of the study is that LCD is rarely "no-code" in practice. Practitioners define LCD through the lenses of UI-driven development, visual programming, and pre-designed templates. Interestingly, the data shows that LCD is frequently used to orchestrate existing services rather than building entirely new logic from scratch.

Distribution of implementation units in LCD, highlighting the dominance of APIs and templates.

As shown above, the most common implementation unit is API-based (47 instances), followed by templates (34). Furthermore, traditional programming languages remain vital. Java and JavaScript are the most cited languages used to extend platform functionality. This indicates that "low-code" often serves as a starting point, with developers reverting to code as an "escape hatch" when the graphical user interface (GUI) fails to meet complex requirements.

Practitioner Sentiment: Benefits and Challenges

The study revealed a "double-edged sword" regarding the developer experience. On one hand, the primary driver for LCD adoption is speed; "Faster Development" was cited 65 times as the top benefit. On the other hand, the marketing promise of "ease of use" is often contradicted by reality.

The most surprising finding was the "Learning Curve Paradox." While 52 practitioners cited "Ease of Study/Use" as a benefit, "High Learning Curve" was actually the #1 cited challenge (21 counts). Once a project moves beyond a simple CRUD (Create, Read, Update, Delete) application, the complexity of a platform's proprietary logic and specific constraints can become a significant barrier. Other major hurdles include high pricing models (especially per-user licensing) and a lack of customization that leaves developers feeling "boxed in."

Application Domains and Platforms

The research identified that LCD is most heavily utilized in E-commerce, Business Process Management (BPM), and Social Media. Popular platforms like Bubble.io and Webflow dominate these conversations. Notably, 14 of the top 21 platforms mentioned are commercial or closed-source, raising concerns about vendor lock-in—a major deterrent for enterprise-level adoption.

Conclusion

Low-Code Development offers a powerful path toward rapid delivery, but it is not a silver bullet. For practitioners, the benefits of speed are often balanced against steep learning curves for complex tasks and the limitations of proprietary ecosystems. Future research will focus on larger datasets and direct interviews to identify the specific "tipping point" where the constraints of LCD begin to outweigh its productivity gains, and how developers can better navigate the risks of vendor lock-in.

Summary updated

BibTeX
@inproceedings{luo2021characteristics,
  author = {Yajing Luo and Peng Liang and Chong Wang and Mojtaba Shahin and Jing Zhan},
  title = {{Characteristics and Challenges of Low-Code Development: The Practitioners' Perspective}},
  booktitle = {ESEM},
  year = {2021},
  doi = {10.1145/3475716.3475782},
}
Figure