Luís Ramalho bio photo

Luís Ramalho

Product, Engineering, Business & Management

Twitter LinkedIn Github Stackoverflow Youtube

Accurate estimates are notorious difficult to make. Parkinson’s law states that “work expands so as to fill the time available for its completion” (Parkinson 1968). In fact, this law might explain the reason why projects rarely underrun. However, the major problem in many projects is that the time available for its completion is often underestimated, thus, causing the potential cancellation or overrun of the project both in terms of time and cost. In the sections that follow, there will be a discussion of reasons why estimates are often wrong and in another post give suggestions of ways to improve them.

Reasons why estimates are often wrong

A notorious industry where project estimates are extremely often wrong is the construction industry (Chan and Park 2005). The Olympic Games is a great example, where more often than not the budget is surpassed. Although one of the reasons for this is that the time constraints forces the project to invariably be concluded on time. There are nonetheless other examples, such as the software industry. Here, Laird and Brennan (2006) state that approximately one quarter of all projects are cancelled ahead of their completion. In fact, of the percentage of projects that actually finish, only 28% deliver on the initial estimated cost and time. These inaccurate estimates, one could argue, are spread over all industries and a substantial amount of research has been done in order to identify the root cause and possible explanations for this recurring issue of imprecise estimates.

However, it is important to note, before listing a few possible reasons that may cause estimates to be often wrong, that the project schedule is usually negotiated at the beginning of a project (Daughtrey and Carroll 2007). Aside from that, in most cases projects start with a date that is decided in advance and with a poor definition of the requirements needed.

A reasonable way to list these reasons would be to categorise them into different groups. Siemiatycki (2008) divided the reasons an estimate might be inaccurate into four main categories, which are: (1) technical; (2) economic; (3) psychological; and (4) political. These sub-groups are logical and will be pointed to during the discussion.

In the literature, multiple authors (Cantarelli et al. 2013; Ramgopal 2003; Siemiatycki 2008) suggest that uncertainties are possibly the major cause of wrong estimates. These are the technical reasons an estimate might not be accurate. Ramgopal (2003) argues that the resources and methods choices are often based on assumptions that might not include any contingency for unpredictable events. Furthermore, this assumptions are not due to rational decision-making but rather a result of opportunism, bounded rationality and subjectivity (Firmenich 2017). Thus, the individuals in charge of the estimates might be limited by the scope of the project they are aware of or the time left to make a decision.

Another technical issue that might cause estimates to be wrong is that they are usually taken from historical data.

Novakova (2013) pointed out that since costs in the past are smaller than in the present, these early estimations are often undervalued. In addition, there are also estimates that go over the actual cost (Moyo and Huisman 2014), although less frequent, these are also the result of similar factors that cause estimates to go under the actual cost. For example, these factors could be uncertainties or bounded rationality. Furthermore, considering Parkinson’s law (Parkinson 1968) it could be argued that estimates are created based on the time available to complete the tasks, instead of the actual time that would take to finish them. This would cause projects to either finish on the time stipulated or simply to overrun. Also, Daughtrey and Carroll (2007) state that the task interdependencies are not recognised in the early stages which might cause further deviations from an accurate estimate. In fact, this is one very important point because if task interdependencies are not planned properly and a contingency solution is not in place then most likely the estimate will fail to be correct.

Humans are not perfectly rational; they are emotional and therefore prone to misjudgement and biases. Flyvbjerg, Holm and Buhl (2002) state that in the beginning of a project, cost reports consistently underestimate the budget because they anticipate a best-case scenario instead of a sensible and more realistic expectation. This, also called “optimism bias” (Siemiatycki 2008), occurs when on one hand organisations underestimate time and cost but on the other hand overestimate the benefits of a project. It is a case of a psychological reason for inaccurate estimates. Cantarelli et al. (2013) add that these initial cost estimates besides the fact that they include prior uncertainties, they are also influenced by possible political pressures, extra modifications and with limited direction during the process.

However, one of the limitations of the factors described in the previous paragraphs is the assumption that the people making the estimates are honest and integrous. Unfortunately, the real world is very complex and often the person doing the estimate will try and disregard on purpose the actual cost of the project so that they increase their chances in successfully promoting a project due to its cheapness (Flyvbjerg, Bruzelius and Rothengatter 2003). That is an example of both an economic and political argument sub-grouped by Siemiatycki (2008) as a cause for possible imprecise estimates. Furthermore, even though they might be honest and integrous, they might lack risk management competence in order to effectively estimate a complex project. In fact, if they simply do not have the necessary competence they will in effect produce a poor risk analysis and as a consequence produce a flawed estimate. In addition, it is not uncommon for companies to over-commit and later under-deliver. This can occur as a result of multiple factors, such as competitive pressure from other companies, having an excessively eager sales team, a misunderstanding of the consequences of the commitment or merely for failing to include the team in the estimation process (Daughtrey and Carroll 2007).

In summary, there are endless reasons why estimates are often wrong. They are extremely hard to make precisely and there are simply too many variables to be counted that might cause the estimates to be inaccurate. The goal would be, therefore, to try and quantify as much as possible the risks and have a contingency plan in case the project almost without exception does not occur as predicted. As an old adage from Benjamin Franklin states: “If you fail to plan, you are planning to fail”


Cantarelli, Chantal C et al. (2013). ‘Cost overruns in large-scale transportation infrastructure projects: Explanations and their theoretical embeddedness’. In: arXiv preprint arXiv:1307.2176.

Chan, Swee Lean and Moonseo Park (2005). ‘Project cost estimation using principal component regression’. In: ConstructionManagement and Economics 23.3, pp. 295–304.

Daughtrey, Taz and Sue Carroll (2007). Fundamental concepts for the software quality engineer. Vol. 2. ASQ Quality Press.</small>

Firmenich, Jennifer (2017). ‘Customisable framework for project risk management’. In: Construction Innovation 17.1, pp. 68–89.

Flyvbjerg, Bent, Nils Bruzelius and Werner Rothengatter (2003). Megaprojects and risk: An anatomy of ambition. Cambridge University Press.

Flyvbjerg, Bent, Mette Skamris Holm and Soren Buhl (2002). ‘Underestimating costs in public works projects: Error or lie?’ In: Journal of the American planning association 68.3, pp. 279–295.

Laird, Linda M and M Carol Brennan (2006). Software measurement and estimation: a practical approach. Vol. 2. JohnWiley & Sons.

Moyo, Benson andMagda Huisman (2014). ‘Empirical investigation of Systems Cost EstimationModels in the Limpopo province of South Africa: A requirements modelling problem’. In: Proceedings of the International Conference on Software Engineering Research and Practice (SERP).WorldComp, p. 1.

Novakova, Helena (2013). ‘Methodology of transportation project management’. In: Journal of Systems Integration 4.3, p. 30.

Parkinson, C Northcote (1968). Parkinson’s Law on the Pursuit of Progress. Penguin Books.

Ramgopal, Maruboyina (2003). ‘Project uncertainty management’. In: Cost Engineering 45.12, pp. 21–24.

Siemiatycki, Matti (2008). ‘Managing optimism biases in the delivery of large-infrastructure projects: A corporate performance benchmarking approach’. In: Infrastructure Systems and Services: Building Networks for a Brighter

Future (INFRA), 2008 First International Conference on. IEEE, pp. 1–6.