Source-linked AI summary
What do we know about software development in startups?
Carmine Giardino, Michael Unterkalmsteiner, Nicolò Paternoster, Tony Gorschek, Pekka Abrahamsson
TL;DR
Software development in startups occurs under high uncertainty, but empirical knowledge about these practices remains limited and premature. The paper maps existing evidence and startup practices, finding that startups adapt agile and lean methods ad hoc to validate products quickly, while emphasizing flexibility, customer feedback, and continuous value delivery. It calls for more transferable and reliable evidence across startup contexts.
Problem
Software engineering practices in startups remain largely unknown, despite startups operating in highly uncertain and rapidly evolving contexts.
Method
The paper conducts a systematic mapping of empirical software engineering studies on startups, assessing their rigor and relevance while synthesizing reported practices.
Results
The evidence base is premature, while startups commonly use agile and lean methods ad hoc to validate products quickly and adapt development to uncertainty.
Takeaways & Limitations
Startup engineering activities should be tailored for flexibility, rapid change, customer acceptance, continuous value delivery, and easy-to-implement tools.
Takeaways & Limitations
The paper calls for more transferable and reliable results that account for diversity in startup contexts and viewpoints.
Abstract
from arXiv · showhide
An impressive number of new startups are launched every day as a result of growing new markets, accessible technologies, and venture capital. New ventures such as Facebook, Supercell, Linkedin, Spotify, {WhatsApp}, and Dropbox, to name a few, are good examples of startups that evolved into successful businesses. However, despite many successful stories, the great majority of them fail prematurely. Operating in a chaotic and rapidly evolving domain conveys new uncharted challenges for startuppers. In this study, the authors characterize their context and identify common software development startup practices.
What is a startup, anyway?
A startup is characterized less by newness alone than by exploring uncertain opportunities in rapidly changing technologies and markets. These conditions distinguish startups from established companies.
- What is a startup, anyway?: Startups are small companies exploring new business opportunities where the solution is uncertain and the market is highly volatile.
- What is a startup, anyway?: High uncertainty and rapid evolution are the key characteristics differentiating startups from more established companies.
Empirical body of evidence (SIDEBAR)
The study maps empirical evidence on startup software development and assesses its rigor and relevance. The evidence base is still premature, with most studies lacking either reliable transferability or strong scientific rigor.
- Empirical body of evidence (SIDEBAR): 43 studies investigate different aspects of startups and their software development processes.
- Empirical body of evidence (SIDEBAR): The mapping assesses rigor through reporting precision and thoroughness, and relevance through environmental realism and transferability to practitioners.
- Empirical body of evidence (SIDEBAR): 10 of 43 mapped studies provide transferable and reliable results to practitioners, while 10 provide low rigor and relevance.
- Empirical body of evidence (SIDEBAR): 23 studies show moderate industry relevance but low scientific rigor, leaving the evidence on startups rather premature.
- Empirical body of evidence (SIDEBAR): The authors conclude that researchers need efficient ways to collaborate with and study startups because resources are limited.
Startup software development
Startup software development spans more than 200 practices shaped by uncertainty, rapid learning, customer feedback, and limited resources. Startups commonly adapt agile and lean ideas pragmatically rather than following any process strictly.
- Startup software development: More than 200 working practices are synthesized to characterize startup software development and identify evidence gaps.
- Startup software development: Agile methods support frequent change through fast, iterative, incremental releases that shorten the path from idea to production.
- Startup software development: Lean Startup identifies risky business assumptions and uses a minimum viable product to test them and plan the next iteration.
- Startup software development: Startups do not strictly follow development processes; instead, they tailor minimal process management to short-term objectives and user learning.
- Startup software development: Customer discovery and minimal functional requirements help startups test whether a proposed solution addresses real needs.
- Startup software development: Modular architecture, refactoring, and focused quality assurance help accommodate changing functionality and growing user demand.
- Startup software development: Loose structures, empowered teams, adaptable expertise, co-location, and change-friendly tools address resource constraints and evolving requirements.
Practical takeaways
Software startups operate under high uncertainty, so their engineering practices emphasize rapid validation, flexibility, customer feedback, and focused value delivery rather than rigid processes or comprehensive quality. The evidence also highlights lightweight tools, adaptable frameworks, experimentation, and empowered teams as practical responses to changing markets and limited resources.
- Process and priorities: Startups use agile and lean methods ad hoc to validate products quickly, while software quality is not their most critical concern.Revenue generation and obtaining funding often take priority during development.
- Process and priorities: Engineering activities should be tailored to preserve flexibility and reactiveness because startup outcomes are difficult to predict in advance.The paper contrasts flexible, context-sensitive work with rigorous methodologies that cannot fully control development under continuous unpredictability.
- Product development: Startups favor well-known frameworks, evolutionary prototyping, and experiments using existing components to change products quickly as market needs evolve.These practices support fast changeability without requiring every product element to be developed from scratch.
- Customer learning and delivery: Ongoing acceptance testing with early adopters and continuous delivery of core functionalities help startups learn customer value and focus on paying users.The evidence also identifies metrics for learning from consumer feedback and demand.
- Organization and tools: Easy-to-implement tools help manage rapidly changing product information, while empowered teams can influence final outcomes and adapt responsibilities.The reported practices include lightweight tools for product development and management and greater team responsibility.
- Evidence boundary: The review calls for more transferable and reliable evidence across the diverse contexts and viewpoints involved in addressing startup uncertainty.The authors frame the startup phenomenon as an opportunity and challenge for further research.