In Agile Software Development (ASD), efort estimation is an activity that appears in the planning phase, and is impacted by a number of variables. Efort estimation in ASD is needed for eficient resource allocation, realistic scheduling, and improved predictability. Without it, teams risk mismanaging resources, missing deadlines, overcommitting, and disappointing customer and/or stakeholders. In this paper, through a semi-structured literature review, we identify and group twenty five major efort estimation challenges that occur in ASD into five categories. We give a summary of each challenge and examine how common it is in the found literature, emphasizing both commonly discussed and less wellknown problems. We additionally identified mitigation propositions for the least frequently mentioned challenges found. Our findings show that while certain issues, such as the value of team experience and the influence of biases, are well known, others - including unclear information, responsibilities and lack of a formal estimation technique, remain under-explored, despite their potential impact. The results of our paper aim to guide practitioners in identifying and managing challenges within their specific contexts, while also highlighting existing gaps that need further investigation. Additionally, our findings contribute to the foundation for improving estimation procedures and informing future research in this domain.
The Project Management Institute (PMI) defines the estimation objective as providing an
approximation (estimate) of the amount of resources needed to complete project activities and
deliver outputs – products or services – of specified functional and nonfunctional characteristics
[
Efort estimation in Agile Software Development (ASD) appears in the planning phase of the
Software Development Life Cycle (SDLC). When planning an iteration (e.g. a sprint in Scrum),
an Agile software development team estimates the efort of a work item (a task or a user story).
The team then chooses a set of tasks to be completed during the sprint based on the estimated
efort, making sure that the total efort of the chosen tasks does not exceed the sprint’s capacity.
The estimated efort of the delivered work items in previous sprints is used to determine the
sprint capacity. This therefore means, that the product of the efort estimation activity is an
estimate on how many functionalities (in a sprint) can be delivered, rather than predict time or
costs [
Depending on the requirements, circumstances, and preferences, efort estimation activities can produce diferent estimations. For example, because of their ease of use and reliance on past experiences, certain development teams may choose relative estimating activities, such as comparing new tasks or user stories to ones that have already been finished. Others could use group estimating activities, which involve cooperative discussions to reach a consensus about the estimated efort. These preferences frequently show how comfortable and knowledgeable the team is with diferent estimation techniques.
In ASD, efort estimation is needed for eficient resource allocation and planning. Teams
can better schedule and distribute the workload by using it to predict how much time and
efort a job will demand [
The aim of this paper is to present a thorough investigation of the challenges associated with efort estimation in ASD. We can learn more about the elements that lead to them by examining their categorization and frequency of appearance in the literature. We further enrich our findings with mitigation propositions for the least mentioned challenges found in the set of literature. Understanding these challenges is essential for developing more reliable efort estimation approaches that not only align with the Agile principles, but also enhance the outcomes of software projects. In the end, resolving these challenges may result in better resource allocation, enhanced decision-making, and a more eficient development process.
The research question that will guide this paper is:
RQ1: What are the challenges within efort estimation in ASD?
RQ1.1: How can these challenges be classified?
RQ1.2: Which challenges occur the least?
The rest of the paper is structured as follows: We start with an overview of related works in this area, which are presented in Section 2. In Section 3 we present the chosen research methodology, whose findings we present in Section 4 of this paper. In subsections 4.1 and 4.2 respectively, we delve deeper into the categorization of the identified challenges and their frequency mapping. Section 5 provides the identified mitigation propositions and in Section 6 we conclude with our overall findings and suggestions for future work.
Because of it’s inherent challenges and the requirement for precise forecasting in dynamic
situations, efort estimation in ASD has been the focus of a great deal of research, including
systematic literature reviews [
In order to classify various methods to increase estimation accuracy and find common causes
of imprecise efort estimations, Pasuksmit et al. [
Iqbal et al. [14] examine the dificulties of user story-based efort estimation in ASD, which also identifies important contributing reasons to inaccurate estimations. According to the paper, technological complexity and inconsistencies in user stories, as well as internal factors like communication, team composition, and competence, have a significant impact on estimating dependability.
Fernandez-Diego et al. [
Lastly, Piñeros Rodríguez et al. [
As indicated by the existing body of literature, considerable efort has been dedicated to identify challenges associated with efort estimation in ASD. Numerous studies have explored various technical, organizational, and human-related factors that influence the estimation activity. However, what distinguishes this paper from prior work is it’s approach to collecting and categorizing the identified challenges. In addition to the collection of well-known challenges, in this paper we also address the least frequently mentioned challenges—those that are often overlooked yet can significantly afect estimation outcomes. By collecting mitigation strategies for these less-known challenges, the paper and ofers practical value for both researchers and practitioners seeking to improve efort estimation in ASD.
To efectively grasp the existing knowledge regarding efort estimation in agile settings, we conducted a semi-structured literature review, which aimed to provide answers for our set research question and it’s according sub-questions presented in Section 1.
The literature review covered the following points: • The pool of literature was limited to the results found in 3 academic digital libraries: IEEE
Xplore, ACM and SpringerLink. • The search string used for the literature search was the following: (("All Metadata":„efort estimation") AND ("All Metadata":„agile") AND ("All Metadata":„challeng*") OR ("All Metadata":„issue*")) • The results were limited only to peer-reviewed literature. • All relevant literature needed to be written in English. • The literature that was not directly related to answering the proposed research questions was excluded.
After identifying the initial set of papers through the presented search strings, we reviewed their references to find additional relevant studies (snowballing). Additionally, we incorporated literature that we were aware of, but did not appear in the initial search results. By following this semi-structured method, this paper tries to synthesize key findings while staying adaptable to new viewpoints. The results of the conducted research method are presented in Section 4.
As mentioned in the Introduction, the focus of this Section is on presenting the challenges
that were uncovered through the research methodology employed. In the following Sections
4.1 and 4.2, will look more closely at these challenges, which have emerged as significant
ifndings. Previous research [
The challenges associated with efort estimation manifest at various stages of estimation —namely, before, during, and after the actual estimation. Some of them are complex and encompass many diferent aspects, each of which poses particular dificulties. In the following of this Section, we will examine and evaluate these challenges, shedding light on the complexities that arise regarding the activity of efort estimation. By doing this, we hope to ofer a thorough overview of the variables afecting estimation as well as the challenges that arise during it. Note that all identified challenges are already present in the selected literature (see Table 1) and their categorization is done based on subjective assessment.
Organizations frequently face a number of challenges throughout the efort estimation activity. Eight key challenges that obstruct eficient efort estimation were found and are discussed in the following.
• O1: Methodology/formalism – Pertains to the lack of procedures and policies on how to deal with failures and avoid repeating mistakes by learning from past experiences. • O2: Time constraints - There is a lack of time for estimating requirements, and the activity of estimating efort and identifying dependencies can be very timeconsuming. • O3: Task details - Multiple aspects of size measurement, their adjustment, task sequence and priority. • O4: Pressure and control by management - A very common challenge resulting in inaccurate estimates to meet management expectations and feeling pressured to be faster than original estimates. • O5: Inappropriate tool support - Complexity and inflexibility of the (estimation) tool, as well as the lack of features that could support the estimation activity. • O6: Lack of organizational knowledge - Pertains to the lack of contacts (experts) required for the efort estimation activity, additional overhead such as meetings and explanations, neglection of relevant factors such as dependencies in initial estimates or non-functional requirements and resources in terms of people. • O7: Unforeseen changes - In terms of system or process-related problems during implementation, ad-hoc requirements, changes in the timeline, or people leaving the project. • O8: Lack of a formal estimation technique - In ASD, people and the interactions between them are valued more than processes and tools, but this doesn’t mean that the estimation activity should be completely left behind. Other challenges include lack of information about the estimation activity itself.
Accurately capturing customer requirements and expectations is a crucial challenge during the efort estimation activity. During our research, we identified four general challenges in this area.
• C1: Communication issues – that pertain to customer irresponsiveness. • C2: Unclear information - information deficit regarding the requirements and unclear . • C3: Unstable information - uncertainty of requirements and frequent requirements changes, after the estimation is done. • C4: Information quality - the amount and characteristics of the information provided by the customer.
The quality, availability, and interpretation of data are crucial components of efort estimation. As a result of our research, we found that this category presents three challenges.
• D1: Accuracy – Pertains to the accuracy of the estimated value in relation to actual efort spent. • D2: Lack of measures to improve and monitor estimations – In addition to measures/metrics that measure planned and actual eforts. • D3: Information deficit in the initial estimation of large, complex requirements – lack of historical data (in regard to previous projects or completed requirements, as well as the data quality and dataset characteristics) and changing requirements (especially in the beginning) due to size, dependencies, and uncertainty.
Issues pertaining to the team itself present another challenge category during efort estimation. Estimates can be greatly impacted by factors like team composition, experience levels, and communication styles. We identified seven key challenges that belong in this category, which are more elaborated in the following.
• T1: Unclear responsibilities - Lack of knowledge about the responsibility in terms of person, team, or workstream for certain requirements. Another sub-challenge is working on parallel projects, therefore producing the same outcome. • T2: Lack of team commitment - General resistance to the program’s way of working and team members not participating in the estimation activities. In some cases, there also may be resistance of the estimation technique used, as well as ignorance of documentation. • T3: Team changes and history - Frequent team changes and a lack of previous work history may hinder the estimation activity. • T4: Common understanding of requirements - A correct and team-wise understanding of requirements may in some cases negatively afect the estimations produced. • T5: Subjectivity of estimates (biases) - Estimates are based on subjective criteria such as the individual knowledge of the estimators. There are some diferent aspects to this challenge, namely: a) anchoring - the first numbers “played” or vocalized serve as an implicit starting point, influencing later estimates. b) groupthink - dominant personalities or the desire for quick consensus causes team members to converge on estimates without fully debating underlying uncertainties. c) Optimism & pessimism - voicing the very extreme estimates. d) Skill disparities - team members may not have the same experience as their more senior counterparts, afecting their judgement. • T6: Eficient communication - Language barriers challenging the correct understanding of requirements, spatial distribution and dificulty to reach a consensus all lie within this category. • T7: Lack of knowledge and experience - Teams often lack experience and knowledge regarding efort estimation.
The software project being developed plays a big role in the potential challenges that may arise during efort estimation. In this category, we identified three major challenges. • P1: Technological factors – The technological complexity of the project and lack of knowledge on the required technologies. • P2: Impact of existing system factors – The set of chosen factors, such as database used, development platform an programming language. • P3: Project setting - A fixed, large time frame requiring initial estimates and the tight tying of estimates to the budget limit the flexibility of estimations. Other issues that arise in this category are (too) big projects or backlogs, complexity of tasks and project environmental settings.
Following the classification of the identified challenges in the previous subsection, we now turn to their frequency mapping. This step involved aggregating the frequency with which each challenge was mentioned in the literature. An overview of all references is given in Table 1, which links each of the challenges to the appropriate source from which it was found.
The presence of T5: Subjectivity of estimates, which was found in twelve distinct references, is an important finding from our research. This implies that a significant issue facing ASD is subjective biases in efort estimation. During efort estimation, expert judgment is often necessary, which, by it’s nature, involves subjectivity. Furthermore, this challenge is often brought to light due to diferences in experience, biases, and diferent perceptions of project needs. Similarly, C1: Communication issues with the customer were mentioned eleven times, making it another frequently mentioned challenge. This high number is explained by how important stakeholder engagement is to estimations. The importance of this challenge is further highlighted by the fact that inconsistency in efort estimation is frequently caused by misaligned expectations, unclear requirements, and insuficient communication.
On the other hand, challenges including O4: Pressure and control by management,
estimation activities received less references. Their specificity to certain organizational contexts, or their minor importance in relation to more dominant challenges (like subjectivity), may be the reasons for the smaller amount of attention they receive.
On the other end of the spectrum, the least mentioned challenges are mostly related to the customer (C2: Unclear information, C3: Unstable information and C4: Information quality) and O5: Inappropriate tool support. These challenges seem to be less present in the literature, as presented with less than three mentions. One possible explanation may be that organizations prioritize other challenges that can be solved in a known way. On the other hand, this finding may potentially reference a lesser know gap in the literature, that can be addressed in the future.
To further enrich the insights from our research, we selected the least frequently discussed challenges in the set body of literature discussed in Section 3. This decision was motivated by the understanding that, despite their limited representation in the literature, these challenges may have important implications for both theory and practice. By highlighting them, we want to consolidate potential solutions to address them, as well as encourage more research into these understudied but potentially significant areas.
To mitigate O5 - Inappropriate tool support organisations should utilize tool support to automate estimation using available and historical data, thereby reducing time and incomparability while enhancing transparency. In this case, it is recommended that such tools are user-friendly and capable of integrating relevant data [19]. The incorporation of analogies with similar projects and initial estimates can further improve estimation. Additionally, tools can be enhanced with features such as Post-its or Scrum Boards to support and visualize the estimation activities [19].
Given the similar nature and underlying causes of challenges C2 - Unclear information,
C3 - Unstable information and C4 - Information quality a combined set of mitigation
strategies is proposed to address them collectively. Common issues such as unclear, unstable,
or inaccurate data, should be addressed particularly in user stories, acceptance criteria, and
requirements. A proposed approach is through detailed analysis and stakeholder validation prior
to estimation. Practitioners should apply structured approaches such as the INVEST criteria,
estimation checklists, and developer stories, as well as explore techniques for identifying
uncertainty, missing information, or relevant quality attributes. Team members should be
equipped with adequate domain and technical knowledge [
Another rarely mentioned challenge is O8 - Lack of a formal estimation technique. To mitigate this challenge, a structured process for developing and evolving estimation checklists tailored to agile teams can be adopted. Such checklists could support expert judgment by ensuring critical factors are considered consistently, thereby reducing the risk of underestimation and improving the reliability of efort estimates [22]. Training sessions and workshops can be implemented to enhance development team and stakeholders’ understanding and application of estimation techniques, to promote consistency and overall understanding. This challenge can be improved with the integration of estimation tools (O5) to streamline adoption and ensure continuous usage.
Since challenges P1: Technological factors and P2: Impact of existing system factors share a similar underlying nature, their mitigation proposals will be discussed jointly. A potential solution lies in investing in continuous training and upskilling initiatives, that are tailored to the project’s tools and platforms. Technical evaluations conducted early on can assist in identifying knowledge gaps and directing resource allocation appropriately. To further lower technical risk, criteria should be established for choosing technologies based on team knowledge, scalability, and project needs. Cooperation with external experts can ofer additional assistance in handling challenging technology settings.
The mitigation propositions discussed above are specifically tailored to address the least frequently mentioned challenges identified in the reviewed literature. While these proposals ofer potential solutions, it is important to acknowledge that their efectiveness may vary across diferent contexts, team structures, and software characteristics.
In this paper, through a semi-structured literature review, we shed light on twenty five major challenges regarding efort estimation in ASD, along with explanations of their definitions. Additionally, we grouped them into five categories, each focusing on a distinct area. Furthermore, we examined how commonly they appear in the literature, emphasizing both the most and the least frequently discussed challenges. Our findings highlight that certain challenges, like the importance of team experience and the impact of biases, are well known, others, including information quality and tooling support, are not as well-discussed but nevertheless afect the estimation activities. Another significant contribution of our paper lies in the identification and formulation of mitigation propositions for the least frequently mentioned challenges in efort estimation in ASD. These challenges, while not widely discussed in the existing body of literature, can have a considerable impact on project outcomes if left unaddressed. The proposed mitigation strategies serve as potential solutions and are proposed based on findings from the reviewed literature.
Using the results of our paper, the proposed strategies are intended to guide practitioners in proactively identifying and managing these challenges within their own organizational and project-specific settings, as well as have a better understanding of existing gaps and areas that need more investigation.
Even though this paper provides a structured perspective on the challenges in efort estimation in ASD, there are still a number of future research directions that can be explored further. During our research, we noticed that there is a lack of empirical studies done in this area. Therefore, one potential research direction lies in the empirical validation of the impact and occurence of these challenges in actual agile projects. Further research into less-studied challenges, such as information quality, unclear and unstable information, may reveal some new viewpoints and solutions.
Future research can help develop more accurate and eficient methods for efort estimation by filling in these research gaps. This will, in turn ultimately help in the successful planning and implementation of agile software projects.
The authors acknowledge financial support from the Slovenian Research and Innovation Agency (Research Core Funding No. P2-0057).
During the preparation of this work, the authors used QuillBot, Grammarly and Thesaurus in order to: Paraphrase and reword, check grammar, search for synonyms, analyze tone, and improve fluency. After using these tools and services, the authors reviewed and edited the content as needed and take full responsibility for the publication’s content. [27] M. Jorgensen, S. Grimstad, Software Development Estimation Biases: The Role of Interdependence, IEEE Transactions on Software Engineering 38 (2012) 677–693. URL: https://ieeexplore.ieee.org/abstract/document/6193066. doi:10.1109/TSE.2011.40, conference Name: IEEE Transactions on Software Engineering.