Journal · Operations
The questions you stopped asking
Every business runs on the data it can still see. The interesting part is what sits underneath that, and what it cost you to stop looking.
Every business has a report that everyone quotes. It comes out of one system, it holds one number, and it is the only part of the operation anybody can point at without asking a person first. Up at that level the light is free and everything agrees.
The agreement is what makes it the surface. A number that is cheap to produce gets produced often, and a number that gets produced often is a number people learn to trust. None of that is evidence that it is the most important number you have. It is evidence that it is the cheapest one.
Underneath it, in the dark, sits everything else. What the job really cost. Where the order actually is. The answer nobody can get to any more. This piece is about how a business loses access to its own information, why every individual loss is sensible at the time, and what it takes to get it back. It is not about software. The software is the last part and the easiest part.
The descent is an economic gradient
Picture the information in your company as water, with you at the surface. The depth of a fact is not how complicated it is. It is what it costs to retrieve.
At the surface, nothing. The report is already written and already believed.
One layer down, the answer stops being a number and becomes a person. Who knows the real price for that account. Who remembers what the last run actually cost. Who can tell you whether that customer is on the old terms. The information exists and it is accurate. It just has to be fetched by hand, and only some hands can reach it. The price of this layer is a favour, which is cheap once and expensive as a habit.
Below that, records stop being written for anyone to read. Exports nobody opens. A sheet maintained by one person who has not been asked about it in a year. A system that got replaced and never switched off, still quietly accumulating. All of it is still there. Nothing above it can see any of it. The price of this layer is an excavation.
Lower again, you have the data and it will not answer the question. The order sits in one place, the cost in another, the delivery in a third, and no two of them share a key. The facts are all present and they cannot be put in the same sentence. The price of this layer is a morning of reconciliation, every time, with a different answer each time depending on who did the reconciling.
And at the bottom are the questions nobody asks. Which customers are actually worth keeping. What a job really costs by the time it ships. Where the margin went this quarter. These are not hard questions. They are not even unusual questions. They are simply expensive enough to ask that the business learned to run without the answers.
Nobody decides this
That last step is the one worth sitting with, because it is the one that does the damage and it never appears on anybody’s list of problems.
Every question has a price, paid in somebody’s hours. When the price of asking exceeds what the answer seems to be worth, the question stops being asked. There is no meeting about this. No decision gets recorded. Somebody asks twice, gets a morning of reconciliation and two different answers, and quietly stops asking. That is the whole mechanism. It is individually rational every single time.
What makes it hard to catch is that you cannot miss a question you stopped asking. There is no error log for abandoned questions. Nothing fails. The business looks like it is running perfectly well, and in an important sense it is, because the questions that survive are precisely the ones that are cheap to answer. They all agree with each other, they always have, and that agreement feels like health.
So the symptom of a business running in the dark is not confusion. It is a strange calm about a narrow set of numbers.
Three complaints, one problem
Owner-operators tend to describe this in one of a few sentences, and they sound like separate problems.
We do not know our true cost. Nobody knows where the job is. Our lead times are guesses. The data is not trustworthy. Our two sites do not sync.
Those are not five problems. They are one problem described from five depths. Each of them is a question whose retrieval price went above its perceived value, so the organisation routed around it: a spreadsheet here, a phone call there, a person who just knows. The workarounds are competent. That is why they last. A business can run on them for a decade and many do.
The cost does not show up as a failure. It shows up as decisions made on the cheap numbers, because those were the ones available.
A diagnostic you can run this week
You do not need a project to find out how deep your own water goes. You need a count.
Pick three fields that something real depends on. A weight that a shipping quote uses. A cost
that a price is calculated from. A date that a promise is made against. Export the list. Then
count how many rows carry a real value and how many carry a placeholder: a zero, a one, a blank,
a TBD, a copy of the row above.
Write the result as a fraction with its denominator, and resist the urge to round it into an adjective. “Most of them are fine” is worth nothing. A fraction out of a stated total is worth a great deal, because the denominator is the part that tells you whether you are looking at a cleanup or a crisis. Date the count, because it will drift, and keep the command you used so somebody can run it again in a month and get a comparable number.
That is the entire method. It takes an afternoon, it needs no new software, and it converts a feeling about data quality into a measurement that two people can argue about productively.
Run it on the field you are most confident about first. That is usually the instructive one.
Coming back up, in the only order that works
Fixing this is mostly a question of sequence, and the sequence is counterintuitive.
The old records come up first. Not the dashboard. Migrate what you have, reconcile it against what actually happened, and look at the difference before anybody is asked to rely on it. This is the step that gets skipped, because it is the least satisfying and the most likely to produce an uncomfortable conversation. Skipping it means building a trustworthy system on figures nobody has checked, which buys you a faster way to be wrong.
You will not like every number that surfaces in that step. You will know them, which is the point.
Then cost, recorded where it happens. Labour, materials, machine time and the jobs they belong to, captured as the work occurs rather than reconstructed at month end from memory and a stack of tickets. When cost is recorded at the point it is incurred, margin stops being an opinion.
Then one place a question goes. Orders, inventory, purchasing, invoicing and the floor on one system, so that the number on your screen and the number in the shop are the same number. Nobody has to be asked. Nothing has to be rebuilt in a spreadsheet to be read.
Then the surface again, which is the part that looks like nothing happened. The report still shows one number and people still quote it. What changed is everything underneath it: current, reachable, and answerable at a price low enough that people start asking again.
That last clause is the actual return. Not the dashboard. The recovery of the questions that got abandoned, because the only reason they were abandoned was that they cost too much.
What this does not do
It does not make every question free. It makes the expensive ones affordable, which is a smaller claim and a true one.
It does not fix a process that is broken for reasons that have nothing to do with information. A system that records a bad workflow faithfully gives you a precise account of a bad workflow.
And it is not for everyone. We work with operating businesses that have outgrown spreadsheets and cannot justify an enterprise platform. We do not take on multi-country consolidation, statutory reporting across several jurisdictions, or deep discrete-manufacturing planning, and we will say so on a first call rather than three months in. Saying what we decline is cheaper for both of us than discovering it together.
If the questions in this piece sound like your own, the useful next step is the count in the diagnostic above. Run it yourself. It will tell you more about your position than any conversation with us would, and you will own the result either way.