Explainer
What operational research actually is.
In plain English, for someone running a business rather than a research group.
Operational research is the study of how to make a system work better, using maths that describes how the system actually behaves. It was invented in Britain during the Second World War to solve problems like where to put radar stations and how to route convoys. After the war it moved into industry, and it is now the reason your flight has a crew rostered against it, your supermarket has stock on the shelf, and your hospital has beds sized against expected demand.
It is not data analytics. Analytics tells you what happened last quarter. Operational research tells you what will happen if you change something, before you spend money finding out.
The methods, and what each one answers
Queuing theory
Answers: why is everything waiting, and will hiring fix it? Queues form when work arrives faster than it can be cleared. The counterintuitive part is that waiting time does not rise smoothly as you get busier. It rises gently, then explodes. That is why a business can feel fine at seventy percent utilisation and be drowning at ninety five, and why the person who seems to be the problem usually is not.
Forecasting
Answers: what is coming, and when should I act? Most owners can feel their busy season. Fewer can put a number on it three months out. Methods like exponential smoothing separate the trend from the seasonal pattern from the noise, using records you already have. It is the difference between hiring in response to pressure and hiring ahead of it.
Capacity and constraint modelling
Answers: what actually limits this business? Every operation has one binding constraint at any given moment. Everything else has slack. Adding resource anywhere except the constraint changes nothing at all, which is why so much investment in growing businesses produces no visible result.
Variation analysis
Answers: is this a bad month, or a real decline? Every process varies. The skill is telling ordinary variation apart from a genuine change in the system, so you react to the second and not the first. Reacting to noise is one of the most expensive habits in a small business.
Decision analysis
Answers: how do I choose between options that are not comparable? When one option is cheaper, one is faster, and one keeps a client happy, the choice is usually made on instinct. Multi criteria methods make the trade off explicit, so the decision can be explained afterwards and repeated.
Problem structuring
Answers: what is the actual problem? Often the hardest part. Soft operational research methods exist for situations where people inside a business disagree about what is wrong. They surface the disagreement rather than papering over it.
Why a business of thirty people has never been offered this
Not because the methods do not scale down. They do. The reason is cost. Traditionally the analysis carried enterprise consulting fees, so it only made sense above a certain size of prize.
Two things changed that. Software made the modelling itself far cheaper, and a small practice with an offshore delivery bench can carry a lower cost base than a large firm. That is the entire commercial logic behind Ashtara.
What it does not do
It does not remove judgement. A model is a description of a system, and every model is wrong at the edges. The value is in narrowing the decision, not in outsourcing it.
It also does not replace knowing the business. The maths tells you where the constraint is. Sitting in the room tells you why it is there, and that part is not analytical.
Where this fits with AI
Operational research decides what is worth automating. AI is one way of doing the automating. Taking those in the wrong order is how businesses end up with a tool nobody uses, automating a process that should have been removed.
Fix the process first. Then automate it.