Operations research
Definition
Operations research
Operations research is the use of mathematics to decide what an operation should do: which schedule, which route, how many staff, what stock level. It is a method, not a management job, and it answers the question before anybody runs it.
That boundary is the one readers get wrong most often. Operations management is the job of running the operation, covering procurement, capacity, scheduling, quality control and distribution.
Operations research is the body of mathematics that job borrows from when it has to choose. Put plainly, an operations manager runs the schedule; operations research decides what the schedule should be.
Key takeaways
- Operations research is a quantitative method, not a department and not a management role.
- Its core tools are optimisation, queueing theory, simulation and forecasting.
- It answers what an operation should do — operations management then does it.
- Pay and headcount data show a recognised occupation rather than a passing fashion.
How it works
Operations research works by turning a messy operational choice into a model: an objective to push up or down, a set of constraints you cannot break, and the variables you are allowed to move. Solve the model, then test the answer against reality.
The constraint list does most of the work. Skill mix, shift rules, vehicle capacity and opening hours are what separate a tidy theoretical answer from a schedule a supervisor can actually publish on Friday afternoon.
The working sequence is consistent across industries. Analysts identify the problem, pull data from systems and from the people doing the work, build and test a quantitative model, then write up what the numbers recommend for the managers who decide.
It is a sizeable profession. The US Bureau of Labor Statistics put the occupation at 113,100 jobs in 2025, with a median wage of $88,940 in May 2025 and growth of 12% projected to 2035.
Data quality decides whether any of it lands. A model fed by a system that records handling time inconsistently will give a confident answer to the wrong question — and that error turns invisible the moment it reaches a slide.
| Method | Question it answers | Where you meet it |
|---|---|---|
| Linear programming | Which mix of work fits the limits best | Production, transport, rostering |
| Queueing theory | How many servers hold the wait under target | Contact centres, clinics, checkouts |
| Simulation | What happens if Monday volume doubles | Warehouse, airport and ward design |
| Forecasting | How much arrives next Tuesday at 10am | Staffing and inventory planning |
Queueing is where most service businesses meet the method first. Erlang models turn arrival rates and handling times into a staffing number, which is why queue management in a bank branch and on a support line share the same arithmetic.
The lineage is older than the computers that made it cheap. Statistical theory only began to be applied effectively to quality control in the 1920s, once sampling theory had developed far enough to be useful.
The engineering statistics handbook published by the National Institute of Standards and Technology credits Walter A. Shewhart of Bell Telephone Laboratories with a 1924 memorandum sketching the modern control chart.
Examples
Four industries show the method doing visible work. In each one the mathematics sets the plan and a manager then runs it, which is the division this page exists to make clear.
Airlines were the first large customer. Crew pairing and aircraft rotation are enormous optimisation problems, and carriers have solved them with mathematical programming since the 1960s because small utilisation gains are worth real money.
Outsourced contact centres use the method daily without naming it. Call center forecasting predicts interval volume, an Erlang calculation converts that into required agents, and workforce management turns the requirement into real shifts.
Utilities schedule field engineers the same way. A day’s jobs, travel times, skills and appointment windows form a routing and assignment problem that no dispatcher can solve well by hand at scale.
Hospitals use simulation to size capacity. Emergency departments model arrivals by hour and day before setting triage and bed numbers, because staffing every day for the Monday peak is the expensive way to fix a Monday problem.
Retailers run the same mathematics on stock. Reorder points, safety stock levels and markdown timing are all optimisation problems, and a chain carrying thousands of lines cannot set them by judgement alone.
Parcel networks optimise routes overnight — a depot solving a vehicle routing problem before dawn is doing operations research, whether or not anybody on the floor uses the phrase.
Related terms
- Operations management: the job of running the operation, which consumes these methods rather than producing them.
- Erlang models: the queueing formulas used for staffing, one family of tools inside the wider method.
- Call center forecasting: predicting demand, which is the input a staffing calculation consumes.
- Staffing model: the arithmetic that converts forecast demand into required headcount and shifts.
- Judgmental forecasting: forecasting from expert opinion, used where data is thin or history has broken.
FAQ
Is operations research the same as data science?
They overlap but differ in purpose. Data science mostly predicts what will happen, while operations research prescribes what you should do about it under stated constraints. The same analyst often does both in one week.
Do you need a dedicated team to use it?
No. Most service firms get the benefit through forecasting and staffing tools that already package the mathematics, with specialists called in only for the harder design questions such as network or site layout.
What qualifications do analysts hold?
A bachelor’s degree is the typical entry point, often in operations research or applied mathematics, and many employers prefer a master’s degree. Some employers require that second degree outright rather than merely preferring it.
Where does it fail?
When the model is solved and the operation never changes — a recommendation that nobody owns is an expensive piece of arithmetic. The second failure is modelling a constraint that the floor quietly ignores.
How is it different from operations management?
Operations research produces the decision; operations management runs the result on a real floor with real people. One is a body of mathematics, the other is a daily job.
Shortlist providers with genuine planning and analytics teams in the Outsource Accelerator provider directory before you hand over your scheduling.
Independent
