July 2024
by Karl Vantine, Chief Customer Officer at Contruent
A lot is happening where artificial intelligence (AI) and project controls meet. One area within that space worth exploring is Earned Value Management (EVM).
Historically, the value of applying EVM has been limited by subjective or inaccurate progress data, which can lead to inaccurate or overly optimistic forecasts and corrective actions focused on the wrong problems. AI stands to play a direct role in changing all this. We’re going to look at three subfields of AI—computer vision, natural language processing and machine learning—that can help EVM deliver more accurate, actionable insights into lifecycle cost management project performance and progress.
How AI Can Transform EVM
EVM is a good structure for measuring performance, forecasting and predicting a project’s future. It’s essentially a big model—you plug in numbers and, using time-tested formulas, EVM calculates a view of the future for you.
However, the accuracy of EVM calculations relies heavily on the quality of data fed into the system. If you give it bad data, you get useless results. In other words, garbage in, garbage out.
This happens far too often. Every formula in EVM leads back to a fundamental starting point—a reliable measure of progress (% complete). But often people don’t use quantitative or objective measures to determine the percent complete; they use their best judgment. Unfortunately, judgement can be subjective and influenced by biases. Most project teams get it wrong by relying on subjectivity rather than data and facts.
How do you overcome human subjectivity that can lead to inaccurate data and skewed forecasts?
First, you must make sure you’re measuring the right things. How you measure contract progress for a supplier is different from how you measure the engineering progress during the design phase, or how you measure field progress during construction. So, first get your metrics right.
Then, you need an objective, quantitative measure of progress. This is where AI can help. Think of it as taking on the role of a garbage filter. AI screens the inputs and asks, “Is this data accurate? Is it realistic based on what data sources indicate?”
It’s all about improving the quality of what gets fed into the EVM engine. AI can help here by combing through large data sets more quickly to weed out the bad and keep the good. With more accurate data, you get more realistic performance and progress results. This way, you know the actual problems and can use AI to evaluate corrective action options.
Computer Vision for Field Data Capture
With computer vision, seeing is believing. Using cameras and drones equipped with advanced imaging, this subset of AI can quickly and efficiently capture objects, people and locations directly from the jobsite. In short, not only can computer vision help gather progress information more quickly, but it can help improve the fidelity and reliability of that progress data.
What’s remarkable about this technology is that each 2D image it captures is analyzed for usable field data. For instance, it recognizes material components, their measurements and installed quantities. It visually checks and confirms physical task progress. It verifies equipment location and usage. Plus, multiple images of a structure can be recorded to create 3D virtual models, a process called photogrammetry, to compare progress against planned models.
Computer vision delivers data that is less subjective and more fact-based. By feeding this collected data into cost management or project controls software that calculates EVM metrics, the results are further refined. This translates to more comprehensive insights into project performance and progress, ultimately leading to more accurate cost and schedule forecasting.
Natural Language Processing for Extracting Insights
Capital projects generate an incredible amount of text data through reports, change orders, meeting notes and risk assessments. However, this data often exists in unstructured formats, making it difficult to glean valuable insights for EVM analysis.
Natural language processing (NLP) comes to the rescue. It understands, interprets and analyzes unstructured text, giving it new life and extracting derivative value that can be fed into EVM metrics.
NLP’s superpower is that it can call out any unrealistic subjective measures. For example, let’s say someone claims a project is 75% complete. NLP can comb through and evaluate project artifacts to verify this. Past meeting minutes may indicate this level of progress isn’t possible. NLP is neutral, seeing through the optimism bias. In challenging the claim, it brings progress claims and percent complete back to reality and a more accurate forecast.
Machine Learning for Data Analysis
Now that you have more accurate data, what do you do with the outcome? Machine learning (ML) helps you see what you think is going to happen versus what is actually going to happen. Its algorithms analyze vast, complex datasets—including past project information, industry benchmarks, project reports and external sources—and can compare them to historical trends. ML’s mission is to discover correlations among data points or predict outcomes, learning what worked and didn’t work.
For example, your CPI indicates 80% efficiency against the budget. You want to know if and how you can improve it. Based on past projects, AI will look at this and ask if proposed mitigation strategies or corrective actions will actually improve the CPI going forward. In short, based on historical application of similar strategies, AI will evaluate whether your corrective actions will have the anticipated effect.
ML continues scouring through data as it pours in, self-improving along the way. Over time, it becomes more adept at picking up on nuanced patterns and anomalies. This helps ML flag data inconsistencies that can compromise EVM calculations and, ultimately, project outcomes.
The Potential of AI-Powered EVM
AI-powered EVM is about what it enables you to do. It doesn’t replace the human element or the decision-making process. Rather, it’s an enabler, supplementing human expertise with exceptional data analysis and forecasting capabilities. Being able to track and predict performance and risks with greater accuracy—especially for an industry as dynamic and risk-prone as construction—empowers you and your team to make decisions more confidently and proactively.
Leveraging these artificial intelligence-based technologies allows you to harness real-time, data-driven insights to better understand and manage project outcomes. A solid earned value management process and cost management technology are excellent places to start if you want to improve your capital project outcomes. Let us know what you’re curious about; we’d be happy to talk with you. And if you’re interested now, request a demo.