Good Ideas, Bad Decisions: Reducing Risk in Construction Decision-Making
By Bill Tryon, Strategic Advisor | Partner Engineering and Science, Inc.
Originally published in the 2026 CLRM Journal.
During my 40+ years in business, both as a client and consultant, I’ve made what feels like millions of decisions. In the early years, my decisions had limited consequences, but as I moved up in the industry, decisions became more complex, sometimes even resulting in broad industry impacts. I’ve spent a lot of time studying decisions in order to try to make the best ones along the way. There are millions of pages of good advice on decision-making, and it’s impossible to dive very deeply into the literature in this short article. Instead, I’ll focus on a few areas that have had the biggest impact during my time in the industry.
Decision Fatigue
Estimates suggest the average adult makes roughly 35,000 remotely conscious (largely automatic) decisions each day.[1] According to McKinsey research, executives spend nearly 40% of their time making decisions.[2] This can drain our cognitive capital, making it difficult to devote the energy needed to make good decisions as the load increases. As an example, the Lazy Analyst study published in 2019 showed that as analysts churned through decisions over the course of a day, their accuracy declined linearly.[3] They don’t just make bad decisions; they make “safe” decisions. Instead of performing fresh, independent calculations, fatigued analysts begin to “herd,” simply copying the consensus opinion or recycling their own previous forecasts.
Triaging decisions is one way to deal with this fatigue. Steve Jobs went so far as to dress identically day after day to minimize his decision load, while Jeff Bezos[4] has described dividing decisions into irreversible (Type 1) and reversible (Type 2). Type 2 decisions can be delegated to others and made quickly, leaving more energy for Type 1 decisions. The 2-Minute Rule, popularized by David Allen,[5] was intended for dealing with emails, but can be useful for triaging decisions as well. To paraphrase the guidance:
- Delete: If a decision is not really needed or is not important, don’t bother.
- Delegate: If a decision needs to be made but not by you, or if someone else is responsible for making the decision, send it to someone else.
- Do (The 2-Minute Rule): If the decision will take less than 2 minutes and is of low consequence, do it immediately. The time it takes to track and manage the task is longer than the time it takes to just finish it.
- Defer: If the decision is very important or will take longer than 2 minutes, schedule it for later.
Which brings us to scheduling. Since research shows that our ability to make good decisions late in the day may decline, setting aside time early in the day or just after lunch can improve outcomes. The Hungry Judge study, published in 2011, underscores the importance of scheduling.[6] The study reviewed the results of 1,000 parole hearings conducted in 2009 and found that 65% of petitioners with early morning appointments were granted parole, but that the chances of parole dropped to near 0% right before lunch and then recovered to 65% following lunch. It may not always be possible to allocate decisions to specific times, but just being aware of the increased probability of failure under the wrong conditions can help overcome the issue.
Inconsistency
Inconsistencies can happen on a personal or organizational level. I’m sure we’ve all learned strategies for dealing with personal inconsistency, but organizational consistency is more complex. When experienced leaders or team members reach different conclusions based on the same data set, or, as discussed above, the same leader reaches different conclusions based on the same data set, we have a problem.
In 2015, Daniel Kahneman studied decision variability at a major insurance company.[7] Executives at the company were confident in the underwriting system and believed their process was disciplined and standardized. When asked to estimate the variance in quotes between professional underwriters assessing identical cases, leadership estimated around 10%. The study, however, revealed a variance of 55%. For the same risk profile, one customer might be quoted $9,500 and another $16,700, based solely on which underwriter handled the file. Not because of skill differences or judgment calls, but because of random factors: mood, recent experiences, time of day, or what case they saw last.
This kind of executive discretion creates challenges for the organization, but reframing the decision process can improve consistency. Instead of asking for an opinion, we can evaluate whether and how predetermined criteria are satisfied. In a lending situation, for example, we might ask an underwriter whether an application satisfies specific criteria such as targets for debt service coverage, the vacancy rate of the submarket, net worth, and performance history of the applicant. The right heuristics can guide decision makers toward more consistent results.
A review and feedback loop can also improve consistency by assuring accountability for decisions made.
Alignment of Objectives & Risk Tolerance
The objectives and risk tolerance of decision makers can also make decisions appear to be inconsistent. An analyst focused on the adequacy of a proposed budget, for example, may discount schedule considerations others may consider to be a deal-killer. Communication and focus are the key to aligning objectives. An analyst who reviews information to make sure it includes certain elements may not recognize the risks represented by those elements, either individually or in combination.
Differences in risk tolerance can also contribute to organizational inconsistencies. For a complex project on a tight budget, a mezzanine investor who will be expected to share cost overruns may want a higher contingency budget than a lender who can expect the investment group to fund overruns.
Alignment in these areas requires thoughtful and precise communication. Sherman Kent, a former CIA officer, is known as the father of modern intelligence analysis. In his 1964 guidance on Words of Estimative Probability, he discusses the “illusion of agreement,” which is an apparent agreement based on two parties using the same word.[8] The illusion comes in when that word means different things to each party. For example, some analysts interviewed assessed the likelihood associated with the word “possible” as a 10% chance of occurrence, while others assessed the probability to be as high as 80%. To avoid this kind of miscommunication, it can be useful to identify a percentage range of acceptable variance, loss, or other concerns up front.
The Illusion of Accuracy
It always seems as though more data will lead to better decisions, but this isn’t always the case. In 1973, Paul Slovic conducted a study of professional horse handicappers.[9] By providing them five pieces of information concerning each horse, the handicappers made predictions with 17% accuracy and 19% confidence. He then gave them 40 pieces of information for each horse; surprisingly, there was no improvement in the accuracy of their predictions, but their confidence nearly doubled to 34%. More data didn’t help them predict the winner. It just convinced them they were right.
This is the trap of modern decision-making. We believe that with one more report, one more week of analysis, one more round of feedback, the right answer will reveal itself. But research from MIT Sloan Management Review shows that beyond a threshold, additional information doesn’t improve decisions; it just delays them and increases false confidence.[10]
There’s a tricky balance here, though. Additional data can sometimes reveal something that wouldn’t have been found by reviewing fewer variables. Where risk tolerance is low or the stakes are high, the time and cost to collect additional information can be trivial.
Bias
We like to believe that our high-stakes decisions are the result of cold, hard logic applied to objective data, but research shows that our brains are riddled with systemic errors—mental shortcuts that evolved for survival but often misfire. These cognitive biases can distort how we perceive value, assess probability, and weigh risk. I’ve provided a few examples below.
- Anchoring Bias: Placing too much value on the first piece of information, narrowing the range of options. For example, properties in an area may typically trade at an identified cap rate, which can lead to an assumption that a property under consideration should trade in a similar range.
- Confirmation Bias: Overvaluing information that confirms what we believe or hope to be true rather than seeking out contrarian information or opinions.
- Sunk Cost Fallacy: Continuing to pour resources into a failed project in an attempt to recover time, money, and effort already spent, ignoring ways in which additional resources might be put to better use.
- Loss Aversion Bias: Studies show that the emotions associated with losing are nearly twice as powerful as those resulting from a win, potentially leading us to overvalue safety vs. return.[11]
- Availability Bias: Overestimating the importance or likelihood of events that are recent or emotionally charged (and therefore “available” in memory). For example, one bad experience with a contractor on a prior project may lead you to avoid them in the future despite countless good experiences.
- Groupthink: The tendency of a team to latch onto and build on an early idea without critically thinking through alternatives. This acts like a “social anchor,” where participants suppress their own doubts to go along with the consensus rather than risk being the lone dissenter.
There are more tools to control these biases than there are biases themselves, but the following provide a starting point:
- Brainwriting: Setting aside time for independent thought prior to discussion of a proposal. Brainstorming meetings can be great for solving problems, but they can also be subject to anchoring. Brainwriting can avoid this by setting aside a few minutes for each person in a meeting to independently brainstorm before diving into discussion. This can help avoid anchoring because each person experiences their own thoughts first. To increase the benefit, notes can be collected or shared among participants to be addressed systematically during discussion.
- Devil’s Advocate/Red Team: Appoint a specific person or group whose sole responsibility is to challenge the prevailing consensus. By explicitly authorizing someone to dismantle the plan, you remove the social friction of being the “negative voice.” This can ensure that silence is never mistaken for agreement and that the strategy is stress-tested against active opposition.
- Second-Order Thinking: “And then what?” Asking this question refocuses discussion to downstream consequences such as implementation, acceptance, and controls necessary for success.
- Six Thinking Hats: This imposes a thought process on decision-making that cycles through factual information, emotions, devil’s advocate, optimism, alternatives, and controls. Everyone involved in the decision assumes the same role at the same time. This can help blunt adversarial positions and encourage collaboration.[12]
The Constant Variable
Technology has changed the business world immensely since I entered it in 1980, but human psychology is much the same. I used to think that good decisions were made by smart people working hard, but even smart people can make better, more consistent decisions by implementing safeguards. It can feel awkward or inefficient to implement some of these ideas, and they may not be warranted for low-stakes decisions. If you think they’d be helpful, I would encourage you to try one or two of the ideas going forward. Aim to shift just 5% of your decision-making and evaluate the results. We can’t eliminate bad decisions, but by acknowledging our flaws and structuring our thinking, we can build systems that protect our best ideas from our own worst instincts.
[1] Psychology Today. (2018). How many decisions do we make each day?
[2] McKinsey & Company. (2021, June 12). Making better decisions. https://www.mckinsey.com/featured-insights/themes/making-better-decisions
[3] Hirshleifer, D., Levi, Y., Lourie, B., & Teoh, S. H. (2019). Decision fatigue and heuristic analyst forecasts. Journal of Financial Economics, 133(1), 83–98. https://www.sciencedirect.com/science/article/abs/pii/S0304405X19300054
[4] Bezos, J. (2015). 2015 letter to shareholders (Exhibit 99.1, Amazon.com, Inc. Form 10-K). U.S. Securities and Exchange Commission. https://www.sec.gov/Archives/edgar/data/1018724/000119312516530910/d168744dex991.htm
[5] Allen, D. (2001). Getting things done: The art of stress-free productivity. Viking.
[6] Danziger, S., Levav, J., & Avnaim-Pesso, L. (2011). Extraneous factors in judicial decisions. Proceedings of the National Academy of Sciences of the United States of America, 108(17), 6889–6892. https://www.pnas.org/doi/full/10.1073/pnas.1018033108
[7] Kahneman, D., Sibony, O., & Sunstein, C. R. (2021). Noise: A flaw in human judgment. Little, Brown Spark.
[8] Kent, S. (1964). Words of estimative probability. CIA Sherman Kent School for Intelligence Analysis.
[9] Slovic, P. (1973). Behavioral problems of adhering to a decision policy. Institute for Quantitative Research in Finance, Napa, CA.
[10] De Langhe, B., & Puntoni, S. (2021). Leading with decision-driven data analytics. MIT Sloan Management Review, 62(3), 1–9.
[11] Kahneman, D., & Tversky, A. (1979). Prospect theory: An analysis of decision under risk. Econometrica, 47(2), 263–291.
[12] de Bono, E. (1985). Six thinking hats. Little, Brown and Company.
