Data-limited Continuous Experimentation (dlCE) : a literature review
(2025) 15th International Conference on Software Business, ICSOB 2024 In Lecture Notes in Business Information Processing 539. p.324-339- Abstract
Continuous experimentation (CE) is a software development approach where product decisions are data-driven. Large global internet-facing companies such as Microsoft, Google, and Facebook apply the practice by leveraging their massive user bases to obtain high statistical significance in experimentation results. However, companies with smaller user bases, such as small- and medium-sized enterprises (SMEs) and early-stage software startups, struggle to adopt CE due to limitations in user data. Our goal is to increase understanding of situations in CE where data limitations occur, as observed in the research literature. We investigate data limitations and challenges related to them, including their characteristics, and solutions and... (More)
Continuous experimentation (CE) is a software development approach where product decisions are data-driven. Large global internet-facing companies such as Microsoft, Google, and Facebook apply the practice by leveraging their massive user bases to obtain high statistical significance in experimentation results. However, companies with smaller user bases, such as small- and medium-sized enterprises (SMEs) and early-stage software startups, struggle to adopt CE due to limitations in user data. Our goal is to increase understanding of situations in CE where data limitations occur, as observed in the research literature. We investigate data limitations and challenges related to them, including their characteristics, and solutions and practices that may address these. We conducted a rapid review of CE papers from a previous systematic literature review, and analysed these to identify scenarios that exhibit data limitations in CE. We present a framework that illustrates different dimensions of data-limited CE (DLC) and connects scenarios to challenges and their potential future solutions. Most challenges and potential solutions that we found are related to the amount of data. We also note examples of other limitations related to, e.g., evaluation metrics, that provide interesting avenues for further research.
(Less)
- author
- Chren, Stanislav
; Fagerholm, Fabian
; Bjarnason, Elizabeth
LU
; Linåker, Johan LU
; Rafi, Saima ; Lehtelä, Bettina ; Runeson, Per LU
and Kauppinen, Marjo
- organization
- publishing date
- 2025-03-23
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- Continuous experimentation, Software engineering, User data
- host publication
- Software business : 15th International Conference, ICSOB 2024, Utrecht, The Netherlands, November 18–20, 2024, Proceedings - 15th International Conference, ICSOB 2024, Utrecht, The Netherlands, November 18–20, 2024, Proceedings
- series title
- Lecture Notes in Business Information Processing
- editor
- Papatheocharous, Efi ; Farshidi, Siamak ; Jansen, Slinger and Hyrynsalmi, Sonja
- volume
- 539
- pages
- 16 pages
- conference name
- 15th International Conference on Software Business, ICSOB 2024
- conference location
- Utrecht, Netherlands
- conference dates
- 2024-11-18 - 2024-11-20
- external identifiers
-
- scopus:105001396970
- ISSN
- 1865-1348
- 1865-1356
- ISBN
- 978-3-031-85849-9
- 978-3-031-85848-2
- DOI
- 10.1007/978-3-031-85849-9_26
- language
- English
- LU publication?
- yes
- id
- 10fa257e-0c74-4abf-8f37-a2ac074f5972
- date added to LUP
- 2025-04-10 08:23:18
- date last changed
- 2025-06-05 14:55:08
@inproceedings{10fa257e-0c74-4abf-8f37-a2ac074f5972, abstract = {{<p>Continuous experimentation (CE) is a software development approach where product decisions are data-driven. Large global internet-facing companies such as Microsoft, Google, and Facebook apply the practice by leveraging their massive user bases to obtain high statistical significance in experimentation results. However, companies with smaller user bases, such as small- and medium-sized enterprises (SMEs) and early-stage software startups, struggle to adopt CE due to limitations in user data. Our goal is to increase understanding of situations in CE where data limitations occur, as observed in the research literature. We investigate data limitations and challenges related to them, including their characteristics, and solutions and practices that may address these. We conducted a rapid review of CE papers from a previous systematic literature review, and analysed these to identify scenarios that exhibit data limitations in CE. We present a framework that illustrates different dimensions of data-limited CE (DLC) and connects scenarios to challenges and their potential future solutions. Most challenges and potential solutions that we found are related to the amount of data. We also note examples of other limitations related to, e.g., evaluation metrics, that provide interesting avenues for further research.</p>}}, author = {{Chren, Stanislav and Fagerholm, Fabian and Bjarnason, Elizabeth and Linåker, Johan and Rafi, Saima and Lehtelä, Bettina and Runeson, Per and Kauppinen, Marjo}}, booktitle = {{Software business : 15th International Conference, ICSOB 2024, Utrecht, The Netherlands, November 18–20, 2024, Proceedings}}, editor = {{Papatheocharous, Efi and Farshidi, Siamak and Jansen, Slinger and Hyrynsalmi, Sonja}}, isbn = {{978-3-031-85849-9}}, issn = {{1865-1348}}, keywords = {{Continuous experimentation; Software engineering; User data}}, language = {{eng}}, month = {{03}}, pages = {{324--339}}, series = {{Lecture Notes in Business Information Processing}}, title = {{Data-limited Continuous Experimentation (dlCE) : a literature review}}, url = {{http://dx.doi.org/10.1007/978-3-031-85849-9_26}}, doi = {{10.1007/978-3-031-85849-9_26}}, volume = {{539}}, year = {{2025}}, }