Somebody on your team is doing a computer's job
It works fine. It just costs somebody's Monday morning, and it falls apart the week they take vacation.
Somewhere in your business, on a Monday morning, a person opens a laptop and starts moving numbers by hand.
They log into the ad account and download a file. They open the customer database and download another one. They paste both into a spreadsheet that has a tab nobody else understands, fix the four rows where the campaign names do not match, and update the numbers in a slide that gets shown on Wednesday. Maybe they also retype a dozen contact forms into the customer database, because the website form and the database have never been on speaking terms.
It takes two or three hours. Sometimes more, if something changed upstream.
That person is doing a computer’s job. They are the piece connecting two programs that cannot connect to each other, and they have been doing it so reliably, for so long, that nobody has ever thought of it as something missing. It is just what Dana does on Mondays.
Why this never looks like a problem
When something is broken, it tells you. A contact form stops working and the phone stops ringing. A credit card processor goes down and somebody calls you, upset.
This does not do that. It produces a finished spreadsheet, on time, every single week. From the outside, everything looks like it is working. And it is working, in the narrow sense that the numbers show up before the meeting starts.
So it never gets on the list. Nobody asks you to fix a process that is currently succeeding. It sits below the line where things get scheduled or paid for, in that big pile of work that is not broken enough to fix and not small enough to ignore.
There is a second reason it survives. The person doing it becomes valuable in a way that is hard to describe and easy to feel. They are the only one who knows why the March numbers got changed, or which of the two lead counts is the real one. That knowledge is genuinely useful to you. It is also a quiet argument against ever replacing the thing that produces it.
What it actually costs
The obvious cost is hours, and the hours are the least interesting part.
Your numbers are always a little bit old. If the connection between two programs is a person, your information only updates as often as that person sits down to update it. Every decision you make off that spreadsheet is a decision made against a picture of last week. In a business where a new lead goes cold in three days, weekly is not a reporting schedule. It is a blindfold with a calendar.
Mistakes do not announce themselves. When software fails, you get an error. When a person mistypes a number, or copies a column one row short, or pulls the wrong date range because the screen reset to a default, you get a spreadsheet that looks exactly like a correct spreadsheet. Nobody catches it in the meeting. It becomes the official record. Six weeks later somebody notices the numbers do not add up and there is no way to tell which week went wrong.
The whole thing depends on one person, and that person takes vacations. This is the version most owners eventually meet face to face. Somebody is out for two weeks, and the report either does not happen or happens badly, and it turns out nobody ever wrote down the four judgment calls that made it right. That is not a documentation problem. That is a process that only ever existed in one person’s head.
It puts a ceiling on what anybody bothers to ask. This is the cost that does the most damage and gets noticed the least. When getting a number costs three hours of somebody’s week, people quietly stop asking for numbers. “What did that campaign actually cost us per closed sale” becomes a question nobody raises, because everybody knows what it would take to answer it. Your business does not get less curious on purpose. It gets less curious because curiosity got expensive.
How to find yours
You do not need to audit anything. You need to ask two questions.
The first: what would stop if this person were out for two weeks? Not what would be harder. What would actually stop happening. Ask it about four or five people and the answers will point straight at the gaps nobody ever filled.
The second: which of our numbers only exist after somebody makes them? Your revenue is in the accounting software whether anyone looks or not. Your open deals are in the customer database. But cost per lead by channel, or close rate by source, or how many of last quarter’s sales came from the thing you spent forty thousand dollars on, may not exist anywhere until a person sits down and builds it. Any number in that second group has a human being underneath it.
There is a third signal, less precise but reliable. Listen for the phrase “I’ll pull that.” Every time somebody says it, there is manual work hiding behind the sentence.
What to do about it
The instinct is to automate all of it at once. That is usually the wrong first move, and it is why a lot of these projects stall halfway and get quietly abandoned.
Not every manual handoff should become automatic. Some of them happen rarely enough, or take enough real judgment, that a person is the right answer and always will be. The work is figuring out which is which.
Three things separate the ones worth building from the ones worth leaving alone.
How often it happens. Something that happens every week is part of how your business runs, whether you built it that way on purpose or not. Something that happens twice a year is just a task. Build the first. Write down the second.
Whether the judgment is real. If the person is matching up two lists by applying a rule they could explain to you in four sentences, that is not judgment. That is a rule nobody has built yet. If they are looking at a record that does not match anything and deciding based on context that exists nowhere else, that is real judgment, and automating it will produce confident nonsense.
What depends on it. A manual step that feeds a slide nobody acts on is not urgent. A manual step that decides how you spend your ad budget, or which leads your salesperson calls first, is holding weight. Fix the ones holding weight and let the rest wait.
Start with one. Pick the handoff that scores highest on all three, build the real connection, and then, this part matters, have the person who used to do it by hand check the automatic version against their own for three weeks running. Not because you distrust the work. Because those four judgment calls living in their head are the actual instructions, and this is the only reliable way to get them written down.
The point
Every business past a certain size runs on connections nobody planned. Some of them are software. Some of them are a person with a spreadsheet and a good memory, quietly holding two programs together on a Monday morning because three years ago somebody had to, and then nobody ever went back and looked at it.
That is not a failure of anybody’s work ethic. It is what happens when a business grows faster than the plumbing underneath it, which is what growing businesses do.
But you should know where those spots are. Not because all of them need to be replaced. Because right now they are the parts of your business with no backup, no warning system, no history, and nothing written down, and you cannot make a decision about a risk you have never named.
Go find out what stops when Dana takes a week off.
If you suspect there are three or four of these quietly holding your business together and you are not sure which ones are holding real weight, that is exactly what a Systems Audit maps. We trace where your information actually moves, where a person is standing in for a connection that was never built, and which two or three of those are worth fixing first.