Operational research (OR) is the use of mathematical and scientific methods to help organizations make better decisions. It takes complex problems, like how to schedule thousands of hospital appointments or route a fleet of delivery trucks, and finds the most efficient solution using data, models, and algorithms. The field sits at the intersection of mathematics, computer science, and management, and it touches nearly every industry where resources are limited and choices have consequences.
Where Operational Research Came From
OR was born out of military necessity during World War II. British scientists were brought in to study the deployment and efficiency of new weapons, particularly radar systems, and their work helped tighten defenses against German air attacks. The approach quickly spread to other military challenges, improving antisubmarine campaigns and bombing strategy. Scientists weren’t designing weapons; they were figuring out the smartest way to use the ones that already existed.
After the war, the same thinking migrated into business and government. If you could optimize how a navy deployed its ships, you could optimize how a factory ran its production lines. By the 1950s and 60s, OR had become a recognized discipline with its own academic departments, professional societies, and a growing toolkit of mathematical techniques.
How an OR Project Works
A typical OR project moves through three broad phases: formulation, analysis, and interpretation. Within those phases, practitioners work through a series of steps that start with defining the problem and end with a plan for action. The process isn’t strictly linear. Analysts often loop back to redefine the problem or adjust their models as they learn more.
The steps generally look like this:
- Problem definition: Pinpointing exactly what needs to be improved and what constraints exist.
- Identifying what “better” means: Establishing how success will be measured, whether that’s lower cost, shorter wait times, or higher throughput.
- Generating alternatives: Laying out the possible courses of action.
- Building a model: Creating a mathematical representation of the real-world system, including its limits and variables.
- Optimization: Running the model to find the best solution among all feasible options.
- Decision making and implementation: Translating the model’s output into an actionable plan that real people can execute.
The power of OR is that it replaces gut instinct with structured analysis. A manager might have a hunch about the best way to allocate a budget across departments. An OR model can test every possible allocation against defined goals and constraints, then identify the option that performs best.
Core Techniques
OR draws on a wide range of mathematical tools. A few show up in almost every introduction to the field.
Linear Programming
Linear programming is one of the most widely used OR techniques. It finds the best outcome (maximum profit, minimum cost) when the relationships between variables are all proportional and the constraints can be expressed as straight-line equations. A manufacturer might use it to decide how much of each product to make, given limited machine time and raw materials. The model maximizes total profit while ensuring no factory exceeds its capacity and no material is used beyond what’s available. The same approach works for blending raw materials to meet quality standards at the lowest cost, or dividing a fixed budget across departments as fairly as possible.
Queuing Theory
Queuing theory is the mathematics of waiting in line. It originated with a Danish engineer studying how many telephone circuits the Copenhagen Telephone Company needed so customers wouldn’t wait too long for a connection. The core idea is straightforward: if you know how fast customers arrive and how long it takes to serve each one, you can calculate average wait times, the likelihood the system is idle, and how many servers you need.
The key ratio is the arrival rate divided by the service rate. When arrivals come in faster than the system can handle them, the queue grows without bound. When the service rate comfortably exceeds arrivals, wait times stay short. Queuing models let organizations find the sweet spot: enough staff or servers to keep waits acceptable without paying for capacity that sits idle most of the time.
Simulation
When a system is too complex for a neat equation, OR analysts build computer simulations instead. These models run thousands of scenarios with slightly different inputs to see how a system behaves under various conditions. Airlines use simulation to test how schedule changes ripple through their network. Hospitals use it to predict how a surge in emergency patients would affect bed availability.
Where OR Is Used Today
Healthcare
Healthcare is one of the richest areas for OR. Researchers have applied queuing models to transplant waiting lists, figuring out how to allocate donated kidneys and livers more effectively. Emergency departments use patient flow models to reduce overcrowding and prioritize care. Intensive care units use optimization to manage admissions when beds are scarce. Specialty clinics apply scheduling models to balance appointment capacity against patient demand, and radiotherapy departments use them to manage differentiated wait times for cancer patients. In every case, the goal is the same: get limited medical resources to the people who need them most, as quickly as possible.
Supply Chain and Logistics
OR is foundational to modern supply chains. Inventory management models help companies decide how much stock to keep on hand, balancing the cost of holding inventory against the risk of running out. Vehicle routing algorithms plan delivery paths for fleets of trucks so they cover all their stops in the fewest miles. Researchers at Cornell University, for instance, have built optimization tools to help the U.S. Navy and Marines transport people and supplies, including blood for transfusions, during overseas conflicts and humanitarian disasters. The same principles apply to any company shipping products: fewer wasted miles, lower fuel costs, faster deliveries.
Other Industries
Finance uses OR for portfolio optimization and risk management. Telecommunications companies use it to design networks that handle peak traffic without overbuilding. Airlines use it to set ticket prices, assign crews to flights, and build schedules that minimize delays. Energy companies use it to balance electricity generation across power plants. Government agencies use it to design public policy, from vaccine distribution plans to urban traffic management.
OR and Machine Learning
Traditional OR excels at finding optimal solutions when you can define the problem precisely. Machine learning excels at spotting patterns in messy, large-scale data. Increasingly, the two are being combined. Machine learning can handle the prediction side of a problem (forecasting demand, estimating arrival rates, identifying patterns in historical data) while OR handles the optimization side (deciding what to do given those predictions). In scheduling problems, for example, this integration improves both the robustness and efficiency of solutions compared to using either approach alone. The combination is especially valuable in fast-changing environments where conditions shift too quickly for static models but still require precise, constraint-respecting decisions.
Software Tools
OR problems are solved computationally, and a range of software exists for the purpose. On the commercial side, Gurobi and CPLEX are two of the most widely used optimization solvers, capable of handling problems with millions of variables. AMPL is a popular modeling language for formulating large-scale optimization problems. On the open-source side, GNU Octave provides a numerical computing environment similar to MATLAB, and ASCEND is used for mathematical modeling in process engineering. Many practitioners also use Python libraries like PuLP or SciPy for smaller problems or prototyping.
Career Outlook
The demand for people who can do this work is growing fast. The U.S. Bureau of Labor Statistics projects employment of operations research analysts to grow 21 percent from 2024 to 2034, which is much faster than average. That translates to roughly 24,100 new positions over the decade. The median annual salary was $91,290 as of May 2024. Most roles require at least a bachelor’s degree in mathematics, engineering, computer science, or a related field, though many employers prefer a master’s degree. The work spans industries from tech and consulting to healthcare and government, and the core skill set (modeling complex systems, optimizing under constraints, translating data into decisions) transfers readily between them.