Engineering the Future of Cost Estimation
In the public sector, cost estimation is the bedrock of fiscal responsibility. It determines budgets for defense systems, infrastructure projects, and critical government programs. While private organizations have rapidly adopted cloud computing, big data, and artificial intelligence, government cost estimating has been slower to modernize.
We are currently witnessing a widening gap between the technological capabilities available to modern enterprises and the tools actually being used by public sector estimators. This lag creates inefficiencies that cost taxpayers significantly and slows down critical decision-making. To modernize, we must first acknowledge the three pillars of our current stagnation: data isolation, inconsistent application of standards, and an outdated technology stack.
The Problem: A Legacy of Silos and Spreadsheets
Data Analysis and Availability
Good estimates are founded upon solid data. Be it historical contractor performance, validated cost estimating relationships, or information about analogous programs. However, in the public sector, data is notoriously siloed. Agencies rarely communicate effectively with one another, and even within a single organization, data access can be a extremely cumbersome. Some of these access issues are driven by warranted security concerns but much of it is self-imposed. Valuable program actuals sit on hard drives, inaccessible to the wider community of analysts who could learn from them.
Compounding this, the analysis of that data is equally restricted. When a new cost estimating relationship (CER) or a novel analytical approach is discovered, it is typically shared only with immediate stakeholders. Even when presented at a conference, the relative infrequency of such events makes best practices slow to propagate. Because there is no centralized mechanism to share these discoveries, the wheel is constantly being reinvented. Novel approaches don’t scale; they are forgotten, leading to a cycle of repetitive, stagnant methodology.
Inconsistent Standards
Often, the GAO Cost Estimating and Assessment Guide is viewed as a gold immutable standard. While these best practices are robust on paper, they are not applied universally in practice. As discussed by Michael Nosbisch at the 2025 International Cost Estimating and Analysis Association (ICEAA) Workshop, many organizations struggle to operationalize these standards.
The result is a large amount of variability in output reliability. When standards are treated as optional guidelines rather than mandatory frameworks, the confidence level of an estimate becomes dependent on the cost team rather than the process itself. One program might have a rigorously risk-adjusted estimate, while a similar program next door relies on point estimates with ill-defined uncertainty. This inconsistency makes portfolio-level decision-making precarious, as leadership cannot trust that the numbers they are comparing were derived with the same level of rigor.
An Outdated Tech Stack
Perhaps the most visible symptom of this lag is the technology itself. For the vast majority of public sector estimators, the “tech stack” begins and ends with Microsoft Office. From data gathering in Excel to presenting results in PowerPoint, the process is manual, error-prone, and static.
Even specialized tools like ACE (Automated Cost Estimator) tell this story well. ACE has been a workhorse of the cost estimating community for decades, providing a structured, repeatable environment that many practitioners have come to rely on.. However, the broader ecosystem has evolved considerably, and integration has become an increasingly important need. While ACE offers an API, its connectivity with modern platforms has not kept pace, with documentation that appears not to have been significantly updated since 2014. Consequently, creating live connections to external results remains difficult or impossible for most users. This forces a static workflow: if an input changes, the estimator must manually re-run the model, export the data, update the spreadsheet, and re-paste the chart into a slide deck. This manual latency is incompatible with modern agile program management, which demands real-time data integration.
The Solution: Building a Modern Estimating Ecosystem
Modernization does not mean simply buying newer software; it requires a fundamental shift in how we treat data and methodology.
Centralized Data and Analysis Repositories
We need to move from local hard drives to secure, cloud-native data lakes. Agencies must develop centralized repositories that not only store raw program data but also index and share the analysis derived from it. Imagine a tool like GitHub for cost estimation, where vetted CERs, normalization techniques, and risk distributions are stored, version-controlled, and accessible to authorized personnel across agencies. This would democratize data, ensuring that a breakthrough in one department lifts the capabilities of the entire sector. In turn this makes future cost estimates more transparent and audit ready.
Living Standards and Automated Governance
Changes are needed across the board—from the governing bodies that write the standards to the organizations that implement them. First, standards bodies like the GAO must treat their guides as living frameworks that evolve alongside modern business processes, rather than static reference manuals.
Simultaneously, organizations must stop relying on manual checklists to enforce these standards. Instead, best practices should be “baked in” to the software itself. Modern estimating tools must act as guardrails, walking users through required process steps—such as normalizing for inflation or documenting assumptions—to ensure that a markedly complete and compliant estimate is the default output, not the exception.
Leveraging Machine Learning and AI
Cost estimation is, at its core, a predictive science, making it a perfect candidate for Machine Learning (ML). Instead of relying on an estimator’s memory to find similar historical programs, Natural Language Processing (NLP) can scan thousands of program descriptions to identify the best analogs instantly. Furthermore, ML algorithms can rapidly test thousands of potential cost drivers to find the most statistically significant relationships, far surpassing human capacity for trial-and-error in Excel. Most importantly, these models can serve as a powerful validation tool by instantly comparing a new estimate against the actual costs of thousands of completed programs, flagging specific parameters that fall outside historical norms.
True Interoperability
Finally, we must break the “copy-paste” cycle. Modern cost tools must support robust, modern APIs that allow for bidirectional integration across the entire program lifecycle—not just for visualization, but for digital engineering and program management.
Cost data should not live on an island. Data from schedule and risk management tools should flow seamlessly into the software being used for cost analysis and just as easily flow out into visualization platforms like Tableau or Power BI. When a program manager adjusts a schedule assumption, the cost impact should be immediately calculable and visible across the ecosystem. This eliminates the need for an estimator to spend days manually updating decks and ensures that cost is a continuous variable in the decision-making process, rather than a periodic report.
Conclusion
The public sector cannot afford to manage 21st-century programs with 20th-century estimating tools. By breaking down data silos, enforcing consistent standards through automation, and embracing the power of machine learning, we can transform cost estimation from a back-office administrative task into a dynamic, strategic asset. The technology exists; the challenge now is cultural adoption. It is time to stop estimating in the past and start engineering the future.
