Source-linked AI summary
A Systematic Review of Productivity Factors in Software Development
Stefan Wagner, Melanie Ruhe
TL;DR
Software productivity remains difficult to assess, and the factors influencing it and their relationships are not fully known. This paper systematically reviews relevant literature and derives factor lists, finding that soft and technical factors warrant equal consideration. The lists can support productivity modelling and improvement, although further detail on factor direction and interrelationships is needed.
Problem
Not all productivity factors and their relationships are known, while productivity must be measured to support its control.
Source: §6 · p. 5Method
The paper systematically reviews productivity literature and derives lists of factors based on their use in studies, considering soft and technical factors in equal detail.
Source: §6 · pp. 5–6Results
The review provides separate lists of important soft and technical factors, and reports that software reuse can have a strong positive influence on productivity.
Source: §3.4 · p. 3, §6 · p. 6Takeaways & Limitations
The compiled lists can support productivity modelling and improvement methods, including interviews to identify influential factors in a specific organisation.
Source: §6 · p. 6The lists need further detail on whether factors affect productivity positively or negatively and on influences between factors.
Source: §6 · p. 6
Abstract
from arXiv · showhide
Analysing and improving productivity has been one of the main goals of software engineering research since its beginnings. A plethora of studies has been conducted on various factors that resulted in several models for analysis and prediction of productivity. However, productivity is still an issue in current software development and not all factors and their relationships are known. This paper reviews the large body of available literature in order to distill a list of the main factors influencing productivity investigated so far. The measure for importance here is the number of articles a factor is mentioned in. Special consideration is given to soft or human-related factors in software engineering that are often not analysed with equal detail as more technical factors. The resulting list can be used to guide further analysis and as basis for building productivity models.
1. Introduction
Software productivity is difficult to measure, and research has not established a sufficient empirical basis for understanding its influencing factors. This review compiles factors from prior literature while giving soft factors consideration alongside technical ones.
1. Introduction
- Productivity is difficult to measure in software development, and terms such as performance and efficiency are sometimes used synonymously.Output divided by input is described as the most general definition.Source: §1 · p. 1
- Studies have used LOC or function points produced per developer-hour as traditional software productivity measures.Source: §1 · p. 1
- Prior discussion of productivity factors was characterized as having an insufficient empirical basis and relying on shallow surveys and qualitative experience reports.Source: §1 · p. 1
- Software engineering literature often emphasizes technical factors, although more than a third of a typical developer’s time is spent on nontechnical work.Meetings and talks account for 21.1%, presentations and project organisation for 9.6%, and independent qualification for 6%.Source: §1 · p. 1
- The review distills factors from software engineering, management, and organisational psychology literature to support model building, productivity improvement, and further research.Factor importance is based on how often factors appear in the analysed literature, with care taken to consider technical and soft factors equally.Source: §1 · p. 1
2. Review approach
The review combines automated database searches with manual collection of relevant publications. It extracts and consolidates factors, then ranks them by their appearances in the literature.
2. Review approach
- The review combines automated and manual searches across software engineering, management, and organisational psychology literature.Automated searches used a query on four scientific-literature portals.Source: §2 · p. 1
- Researchers inspected the first 100 results from each portal, excluding narrow analyses of individual factors and studies reporting only no influence.The large search-result counts made manual review of every result impractical.Source: §2 · p. 2
- Manual journal searches and reference following supplemented portal results, while books by Boehm and Jones were included as a baseline.Source: §2 · p. 2
- The review extracted product, process, and people factors, unified synonymous terms where possible, and ranked factors by appearances in the literature.The ranking primarily sought to identify how many different authors used each factor.Source: §2 · p. 2
3. Considered studies
The considered studies span early technical productivity measures and later work on human and organisational factors. Across these studies, productivity factors include both technical and soft dimensions, with findings that vary by context.
3.1. 1970–1979
- Early studies included programming productivity measured as effort per SLOC, and many later studies reused or adapted their factors.Some factors became less common, while user participation, design constraints, and programming-language experience remained relevant.Source: §3.1 · p. 2
3.2. 1980–1989
- DeMarco and Lister emphasized sociological influences on productivity, including turnover and workplace conditions, and introduced E-Factor, the ratio of uninterrupted to body-present hours, to measure work interruption.COCOMO is a cost-estimation model in which developer productivity plays a decisive role; its factors were derived empirically from a large project database.Source: §3.2 · pp. 2–3
3.3. 1990–1999
- Brodbeck’s survey found projects with higher communication effort were more successful, and internal communication intensity was positively correlated with project success, contrary to the belief that communication effort hampers productivity.In 31 development groups, cohesiveness and capability had the strongest influence, while experience had the weakest influence.Source: §3.3 · p. 3
3.4. 2000–2007
- COCOMO II identified soft factors including programmer capability and personnel continuity, and reported their combined importance exceeded that of the other factors.Factors also varied by business domain, while reuse was reported to have a significant productivity effect and could have a strong positive influence.Source: §3.4 · p. 3
- A project-manager survey identified soft factors including management support, business culture, team building, communication, freedom, responsibility, motivation, and appreciation.Source: §3.4 · pp. 3–4
4. Results
The review groups productivity factors into technical and soft categories, while noting that some boundaries are blurry. Its findings highlight communication effort, business domain, and developer capability as relevant considerations.
4. Results
- Technical factors are grouped by product, process, and development environment, while soft factors are organized into a non-unique categorization.The development-environment category covers tools; the soft-factor categories distinguish company-wide and team culture, individual capabilities and experiences, working environment, and project-specific factors.Source: §4 · p. 4
- Communication effort is described as positive for productivity, despite discussion that it should be reduced as unnecessary work.The authors suggest the issue is that communication effort rises strongly as team size increases.Source: §4 · p. 4
- Business domain may influence productivity or determine which other factors matter most, suggesting a need for individual rather than generic models.Source: §4 · p. 4
- Developer capability is more notable than experience, which empirical studies find rather insignificant despite its prominence in interviews.The authors suggest that long time in a profession alone does not make someone productive.Source: §4 · p. 4
- The paper presents a derived set of soft factors in Table 2.Source: Table 2 · p. 5
5. Related work
Earlier reviews and related studies address different aspects of productivity. This paper distinguishes its focus on factors influencing productivity from work on factor relevance, knowledge-worker productivity dimensions, and military software projects.
5. Related work
- Jeffery and Lawrence reviewed conflicting findings on factors such as experience and size, whereas this paper examines factor relevance generally.Source: §5 · p. 5
- Maxwell, Van Wassenhove and Dutta linked military software productivity factors to earlier studies, but their work predates many later contributions.Source: §5 · p. 5
- Ramírez and Nembhard reviewed dimensions of knowledge-worker productivity, while this review considers factors influencing productivity.Source: §5 · p. 5
6. Conclusions
The paper reviews accumulated productivity-factor literature and derives lists of soft and technical factors. The lists support modelling and improvement work, while further research must clarify factor direction, interactions, and reporting context.
6. Conclusions
- Productivity is difficult to control without knowing its influencing factors, and developer performance can vary substantially.Hatton reports variations of a factor of 10 or more in individuals’ performance in most experiments.Source: §6 · p. 5
- A systematic review of accumulated literature derives lists of important factors, with soft and technical factors investigated in equal detail.Source: §6 · pp. 5–6
- The lists can support productivity modelling and improvement methods, including interviews to identify influential factors in a specific organization.The authors cite ProdFLOWTM as an example of an interview-based method that can draw on the compiled factor knowledge.Source: §6 · p. 6
- Further work should specify whether factors influence productivity positively or negatively and account for influences between factors.Source: §6 · p. 6
- Future reports should describe factors, measurement units, and context in greater detail to enable more valuable aggregation.Source: §6 · p. 6