Pick a job you finished last month. Not a category, not an average. One job, with a name on it.
Now tell me what it cost you.
Most owners can answer the first half of that instantly. The revenue is on the invoice. It is one number, it is in the accounting system, and nobody argues about it. The cost side is where it goes quiet. Somebody will say they can get it, and what they mean is that they can assemble an estimate over the next few days out of three systems that were never designed to talk to each other.
That pause is not a finance problem. It is an operating problem wearing a finance costume, and it shapes decisions you are already making.
Why the number does not exist
Revenue is easy to capture because it happens once, at a moment somebody cares about, and a document gets created. Cost is hard because it arrives in pieces, from different places, at different times, and nobody owns the assembly.
Labor is usually the first break. Hours get recorded against a day or a department rather than against the job, or they get recorded against the job but at a blended rate that was set two years ago and has not been revisited. Overtime lands somewhere else entirely.
Materials are the second. What was pulled for the job and what was consumed on the job are frequently different numbers, and the difference lives in somebody's memory. Consumables often are not tracked to the job at all, because individually they are too small to bother with, and collectively they are not.
Then there is the equipment, the rework, the second trip nobody logged, and the overhead. Overhead is the one that does the most damage, because most companies allocate it as a flat percentage. That means the simple job and the difficult job absorb the same burden, which is another way of saying the simple job is subsidizing the difficult one and you cannot see it happening.
None of these are failures of effort. Each system is doing its own job correctly. The problem is that no one system was ever asked to produce the sentence you actually want, which is: this job earned this and cost this.
What you do instead, without noticing
When the real number is not available, decisions still get made. They just get made on substitutes.
Pricing gets set by feel, or by what the last quote was, or by what a competitor charges. It usually works, because experienced operators have good instincts. The trouble is that instincts are calibrated on an average, and averages hide the jobs at both ends. Somewhere in your work there is a category you are proud of that loses money, and a category that feels unglamorous and carries the business.
Capacity decisions go the same way. When you cannot see cost per job, you optimize for volume, because volume is visible. More work looks like more health. It is not always.
And the moment the market tightens, this becomes urgent rather than academic. In a good year, an unknown margin gets absorbed by growth. In a flat year, you need to know which work to protect and which to reprice, and that is precisely when you discover the number was never there.
Costing one workflow, not the whole company
Here is where most attempts die. Somebody decides to fix this properly, which means a company-wide costing initiative, which means every system, every department, and a project plan. It gets six weeks in, collides with the actual work, and stalls. Then nobody wants to raise it again for a year.
The version that works is much smaller. Pick one workflow. Cost it end to end. Then stop.
Choose your most common job type, not your most complex. You want the one you do many times, because that is where a small error repeats itself into a real number, and because you will get enough examples to see a pattern rather than an anecdote.
Then trace one instance of it, all the way through, with the people who actually touched it. Where did the hours get recorded, and against what. What materials left the shelf, and how do you know. What got consumed and never recorded. What came back and had to be done twice. What sits in a different system that nobody thought to mention because it is not their department.
You are not building a model yet. You are finding out where the truth currently lives and which parts of it are not being written down anywhere.
What to do with what you find
Almost always, the gaps fall into three groups, and they need three different responses.
Some are capture problems. The data exists, it is just being recorded at the wrong level, usually against a department or a day instead of against a job. This is the good kind of problem, because it is fixed by changing where a field gets filled in, not by buying anything.
Some are definition problems. Two people record the same thing differently because nobody ever decided what counts as this job. These do not get solved by tooling either. They get solved by making the decision and writing it down.
And some are genuine absences. Nobody is recording it and nobody ever was. Here you have to be honest with yourself about whether it is material. Some things are genuinely too small to chase, and a costing model that demands perfection will collapse under its own weight. Decide deliberately what you are choosing not to track, and write that down too, so the number carries a known boundary rather than a false precision.
The output of all this is not a dashboard. It is a defensible sentence about one type of work: this job earned this, cost this, and here is what we know we are excluding.
Once you have that for one workflow, the second one is dramatically faster, because the hard part was never the arithmetic. It was agreeing what counts.
What changes when you have it
The obvious change is pricing, and it is real. But the more useful change is that a category of argument disappears.
Cost versus revenue stops being a debate between operations and finance, each holding a different number and each convinced the other is missing something. It becomes a report. People can disagree about what to do about it, which is a productive argument, instead of disagreeing about what is true, which is not.
You also stop being surprised. Not because the numbers improve on their own, but because you can see a job going wrong while it is still going, rather than at the end of a quarter when it is already history.
Where to start this week
You do not need a project or a budget for the first step.
Take your most common job type and try to cost one instance of it end to end. Give yourself an afternoon, not a month, and pull in the people who touched it rather than the people who report on it.
If you get to a number you would defend in front of the person who did the work, you are in better shape than most companies and you should go do the second workflow.
If you cannot get there in an afternoon, you have learned something more valuable than the number itself. You have learned exactly which part of your operation is invisible, and that is a much smaller and more solvable problem than the one you thought you had.
If you have tried to answer this question in your own business and it turned out harder than it should have been, I would be interested to hear where it broke down.