Nobody Gets the Time Back
Every systems consultant sells the same promise, and I am no exception: do this work, and you will get your time back.
The arithmetic is sound. A task takes twelve minutes, runs forty times a week, and after the build it takes none. Eight hours a week, recovered. The spreadsheet is honest.
The promise behind the spreadsheet is not, and I have watched it fail often enough to stop pretending otherwise. The hours are real. Almost nobody ends up with them.
What follows is an attempt to say why, and what to do about it. It is not an argument against the work. I still think the work is worth doing. It is an argument against the thing the work is usually sold as.
The escalator
Hartmut Rosa has spent a career on a problem that ought to be more famous than it is: every technology that saves time has produced a society with less of it.
Washing machines did not reduce the hours spent on laundry. They raised the standard of what counts as clean, and the hours stayed roughly where they were. Email did not reduce correspondence. It removed the friction that had been limiting it, and correspondence expanded to fill the space. Each innovation delivered exactly what it promised at the level of the individual task, and at the level of the week nothing was returned.
Rosa's point is that acceleration is not something that happens to us from outside. It is structural. A system in which everyone is getting faster is a system in which standing still means falling behind, so the time you save is immediately spent staying level. The escalator speeds up, and you walk faster to remain where you were.
This is the first thing that happens to the eight hours. They do not appear as eight hours. They appear as a slightly higher baseline — a few more enquiries handled, a slightly faster response time, a standard that quietly becomes the new normal and is never again described as an improvement.
The second shift, and who runs it
Byung-Chul Han describes the modern worker as somebody who has stopped needing a master, because the function has been internalised. We are no longer disciplined from outside. We are achievement subjects, and we exploit ourselves with enthusiasm, and we experience this as freedom, which is why it is so difficult to resist.
This lands with particular force on anyone running a small business, because the structure Han describes abstractly is their literal situation. There is nobody above them setting the pace. They set it. When the eight hours arrive, nobody takes them. The owner decides what happens to them, and the owner — who is tired, and conscientious, and has a list — decides to do more.
I have had this conversation many times. We free up a day a week, and three months later the person is working the same hours and is faintly embarrassed about it, as though the failure were one of discipline. It is not a failure of discipline. It is what the structure produces. A person who has internalised the demand to perform does not respond to recovered capacity by resting. They respond by raising the target, because that is what the achievement subject does, and because the alternative — stopping — feels like something that would have to be justified.
The automation did not fail. It did exactly what it was asked. It just was not asked for the thing the person actually wanted.
The worst case, which is more common than it should be
There is a failure mode worse than not getting the time back, and David Graeber named it.
Graeber's argument about bullshit jobs is often misremembered as a claim about bad jobs. It is narrower and more unsettling than that: there exists a substantial category of work which the person doing it privately believes to be pointless, and which they perform anyway, and about which everyone maintains a polite silence.
Small businesses have this too, at the level of the task rather than the role. The report that is produced monthly and read by nobody. The approval that has never once been withheld. The check that was added after an incident in 2021 and has outlived every condition that made it sensible. Everyone involved half-knows. Nobody raises it, because raising it would be strange, and because it only takes a few minutes.
Automate one of these and you have done something worse than waste the build. You have made it permanent and you have made it invisible. A pointless task performed by a person is at least irritating, and irritation is information — it is the pressure that might eventually cause somebody to ask why. A pointless task performed by a machine generates no irritation at all. It runs silently, forever, and the question never gets asked again.
I have built some of these. Not knowingly, but the outcome is the same, and the only defence is to ask, before any build, what would happen if this simply stopped. Sometimes the honest answer is: nothing. That answer is worth more than the automation would have been, and it is free.
What the hour was for
Hannah Arendt made a distinction that I find more practically useful than most things written directly about business.
She separated labour, work and action. Labour is the cyclical, consuming, never-finished activity that keeps life going — it produces nothing that lasts, and it must be done again tomorrow. Work is fabrication: making something durable, something that outlasts the making of it. Action is the part that happens between people, unpredictable and unrepeatable, where something genuinely new can begin.
Automation removes labour. That is precisely what it is for, and it is good. The repetitive, cyclical, consuming part of running a business is the part that should be given to a machine, and giving it away is not a loss.
The difficulty is what fills the space. In almost every case I have seen, the space is filled with more labour. The hours recovered from one cyclical task are spent on another cyclical task, because that is what is visible, and because labour is the category of activity that always presents itself as urgent. The business becomes more efficient at the thing it was already doing and no more capable of anything else.
Arendt's categories give you the question to ask before the build, and it is not what will this save. It is: what will the saved hour be, and in which category. If the answer is more labour — and the honest answer usually is, until somebody decides otherwise — then what you have bought is throughput, which is a real thing and worth money, but it is not what was promised and it should not be sold as such.
Why this is not a motivation problem
It is tempting to treat all of the above as a matter of personal resolve. Decide to use the time well. Protect it. Be disciplined.
Donella Meadows explains why this almost never works.
Her best-known contribution is a ranked list of places to intervene in a system, ordered by how much leverage each one gives you. At the weak end sit parameters: numbers, rates, efficiencies. Adjusting them feels like action and changes very little, because the system reorganises around the new value and continues doing what it was doing. Near the strong end sit the system's goals, and above those, the paradigm the goals arise from.
Efficiency is a parameter. It is close to the weakest intervention available.
If the goal of the system is do as much as possible, then every efficiency gain is immediately converted into more output. Not through anyone's weakness — through the system functioning correctly. The goal is doing what goals do. You cannot defeat a goal with willpower applied at the level of a parameter; you can only change the goal, and changing the goal is a different kind of decision from buying an automation, made in a different register, usually not by a consultant.
This is the part of my own work I find hardest to say to a client, because it sounds like a refusal to sell. We can make your operations substantially more efficient. Whether that produces anything you actually want depends on something upstream of me, which I cannot build and which you may not have decided.
So what should anyone do
I promised this was not an argument against the work, so here is the practical residue.
Decide what the hour is for before you free it, and write it down. Not as an intention — as a specific commitment with a shape. Friday afternoons are for the thing that has no deadline. An unassigned hour has already been assigned; it just hasn't told you yet.
Ask what happens if the process stops, before asking whether it can be automated. This single question kills a meaningful fraction of candidates, costs nothing, and is the only reliable defence against making a pointless task permanent.
Treat the recovered time as a decision, not a consequence. The build produces capacity. What the capacity becomes is a separate choice, made by a person, usually not made at all — and a thing that is not decided is decided by default, in favour of whatever is loudest.
Be suspicious of the version of this that promises rest. Including my version. The work reliably produces capacity. It has no mechanism whatsoever for producing rest, and anyone implying otherwise is describing a hope.
And occasionally, consider doing less rather than doing it faster. Nassim Taleb's observation about subtraction — that we know far more reliably what to remove than what to add — applies to operations more cleanly than to almost anything else. Removing a step is free, immediate, reversible and certain. Automating one is none of those. The cheapest improvement available to most businesses is not a build. It is deciding that something does not need to happen.
What I actually sell
Not time. The hours are real, and the arithmetic holds, and then the hours go somewhere, and where they go is not something I control or can honestly promise.
What the work produces is a business that is less dependent on one person's memory, that can be handed over, that does not quietly lose the things it was supposed to do, and that has capacity it did not have before.
Capacity is not the same as freedom. Capacity is an option, and an option that nobody exercises expires exactly like one that was never issued.
The decision about what to do with it is yours, it is more consequential than anything I will build you, and it is not a decision you can buy.
This is the first in an occasional series of longer pieces — the ones that argue rather than instruct. The practical material lives in the Knowledge Base, where it belongs.