Part 1: Revenue Optimization in Construction Bids

Revenue Optimization in Construction Bids

The construction industry faces significant uncertainty at every turn. Extended project deadlines, changing job requirements, and delivery challenges can transform comfortable profit margins into break-even scenarios. While every construction company develops their own tactics to handle these challenges, there are systematic approaches one can take to head against the uncertainty.

One important aspect of the construction industry littered with uncertainty is placing bids and quoting. Knowing how much to bid, how long to deliver the quote, hedging risk against cost estimates is difficult to optimally quantify. It is even more difficult when trying to optimize both long-run profits and customer retention.

In this blog series, we’ll provide an overview of how construction companies can place better bids. This article will introduce you to the key concepts in bidding models. Part 2 will demonstrate how a bidding model works in practice with a more detailed explanation of how such a model can affect your bottom line. Part 3 will then briefly discuss the challenges of implementing this model in practice and what data infrastructure should be put in place so that we can assure confidence in our bidding model across the business. At the end of this series, I hope that you have a better idea of what a bidding model can do for your construction company.

Why Models Matter in Construction Bidding

Statistical models provide a framework for making better decisions under uncertainty. Rather than relying solely on intuition or past experiences, models help quantify risks and opportunities in concrete terms. They serve multiple crucial functions:

First, models force us to think explicitly about the problems we’re solving. When we write down our assumptions mathematically, we can examine them more critically and refine them based on data.

Second, models give us a foundation to build upon. As we gather more data and understand our market better, we can improve our models iteratively, leading to increasingly better decisions over time.

Better Bidding

At its core, a good bidding model helps us answer a fundamental question: “What is the optimal bid amount given the characteristics of the job?” We can express this mathematically as:

\[ \pi = Pr(win | b)*(b - C) \]

Where:

• \(\pi\) is the expected profit

• \(Pr(win|b)\) is the probability of winning given the bid amount

• \(b\) is the bid amount

• \(C\) is the cost of doing the job. This can be the sum of all aspects of the job (materials, labor, transportation, etc.).

Characteristics of a job vary dramatically across different locations, customers, and time. Construction companies likely have a dedicated team to estimate these costs given the characteristics of the job. We can incorporate how well we estimate any job by thinking of \(C\) as a distribution that captures how well your team estimates jobs:

\[ C = c'*N(e,d) \]

Where:

• \(c'\) is the estimated cost

• \(N(e,d)\) is a normal distribution representing historical cost deviations

• \(e\) is the average ratio of estimated to actual costs (\(e = \frac{c'}{c^*}\) where \(c^*\) is the actual costs and \(e\) should be centered around 1)

• \(d\) is the historical deviation of estimates

When a new job is bid, it takes a sample from \(N(e,d)\). A value less than one is an under-estimate and more than one is an over-estimate. A better estimating team would have little variance \(d\) and centered around \(e = 1\).

In practice, we would use this model by first getting the cost distribution \(C\) . Then we would compare across markups from 0% markup and 100% (or some other high markup) and pick the markup that has the highest expected profit. The most difficult part in having the right model is estimating \(Pr(win | b)\). In Part 2, we will go through how to do this, step-by-step. In Part 3, we’ll go through the model’s core assumptions and what we can do to counter-act bias in our estimates.

This mathematical framework gives us a foundation for making better decisions. By understanding how our bid amounts affect win probability and how our cost estimates relate to actual costs, we can develop more refined strategies for different types of projects. Let’s look at how this translates into practical decision-making.

Making Optimal Decisions Under Uncertainty

The power of this model lies in its ability to quantify uncertainty and optimize decisions. For example, if we know that a 10% markup gives us a 50% probability of winning a bid, and the potential profit at that bid amount is $100, we can expect an average profit of $50 per bid at that markup level.

This statistical approach helps construction companies:

1. Make more informed bidding decisions based on quantifiable data
2. Better understand their cost estimation accuracy
3. Optimize markup strategies for different types of projects
4. Balance the trade-off between winning probability and profit margins

The key is to remember that while no model is perfect, a systematic approach to bidding helps reduce uncertainty and improve long-term profitability. By continuously gathering data and refining our models, we can make increasingly better decisions in an inherently uncertain industry.