What Automation Actually Costs a Small Business

This is the first question almost everyone asks, and the honest answer — it depends — is useless on its own. So here is what it actually depends on, and how to work out the one number you can calculate today without talking to anybody.

Start With What You Are Already Paying

Manual work is not free. It is just invoiced as salary rather than as a project, which makes it invisible. Nobody circulates a monthly statement for the four hours a week someone spends copying figures between two systems, so that cost never gets compared against anything.

The arithmetic is not complicated. Take the task, and be specific — not “admin” but “pulling the weekly numbers into the report.” Then:

A worked example, with round numbers chosen for the arithmetic rather than from any real engagement: three people spending five hours a week each at a loaded cost of $45 an hour comes to fifteen hours weekly, about 690 hours a year, and roughly $31,000 annually. For one task. That is the figure a quote should be compared against — not against zero.

Do this before you ask anyone for a price. It changes the conversation from “is this expensive?” to “does this pay for itself, and how fast?” If you want to run the numbers without doing the multiplication yourself, there is a calculator on our homepage that does exactly this.

One caution about that number: it is real, but it is not usually cash you get back. Freeing fifteen hours a week rarely means a smaller payroll. It means three people spend those hours on work that actually needs a person. Whether that is worth $31,000 to you is a genuine business judgement, and it deserves to be made deliberately rather than assumed in either direction.

Manual work is not free. It is just invoiced as salary rather than as a project, which makes it invisible.

What Makes One Job Cost More Than Another

Two tasks that look identical from the outside can differ by a factor of ten in effort. These are the things that actually move the number, roughly in order of how much they matter.

How many systems it touches. A task that lives entirely in one spreadsheet is a different proposition from one that pulls from an accounting package, cross-references a supplier portal, and emails the result. Every boundary between systems is somewhere the work can go wrong, and each one has to be built and then handled when it misbehaves.

Whether the rules are written down. This is the one that surprises people. If the process lives in one person’s head — and it usually does — then a meaningful part of the work is getting it out of there. Half the value of a good process review is that, for the first time, somebody writes down what the business actually does.

How messy the data is. Consistent, well-formed data is straightforward. Data where the same supplier is spelled four ways, dates arrive in three formats, and someone has typed “N/A” into a number column is not. This is ordinary and fixable, but it is work, and it is usually invisible until someone looks.

How many exceptions there are. The main path is rarely the expensive part. The expense is in “except when it is a rush order,” “except for that one client,” and “except at month end.” A process with three exceptions and a process with thirty are not the same project, however similar they sound in a meeting.

Whether it depends on something you do not control. If the work requires reading a system belonging to someone else — a supplier’s portal, a customer’s system, a service with no proper interface for it — that dependency shapes both the cost and the risk, because it can change without warning and without asking you.

What This Usually Looks Like In Practice

Most of what we see falls into three rough shapes. These are about scope rather than price, because scope is the thing you can assess yourself before anyone quotes.

A single task, one or two systems, rules that are already clear. A report that assembles itself, a form that files itself in the right place, a weekly export that stops being manual. Days to a couple of weeks. This is where we suggest most people start, and not only for the cost — a small first project tells you whether you want to work with someone, on a scale where being wrong is cheap.

A process that crosses systems, with real exceptions. Several steps, more than one system, handling for the cases that do not follow the main path. Typically two to four weeks to something working. This is the most common shape of the work.

Replacing a spreadsheet that has become a system. Almost every business has one: a workbook that started as a convenience and now runs something important, held together by one person who knows which tabs not to touch. Replacing it properly is a larger piece of work, because the spreadsheet has quietly accumulated years of business rules nobody has ever listed.

Questions Worth Asking Anyone Who Quotes You

You do not need to be technical to tell a considered proposal from a hopeful one. Four questions do most of the work:

When It Is Not Worth Doing

Plenty of manual work should stay manual, and it is worth saying so plainly.

If the task takes twenty minutes a month, the arithmetic will not carry it, however irritating it is. If the process is about to change anyway — new system coming, the department reorganising — automating the current version buys something you are about to throw away. If nobody can articulate the rules, that is not a signal to build; it is a signal that the process needs deciding before it needs building. And if the judgement in the task is the task, then what looks like manual work is actually someone thinking, and removing it removes the point.

The work worth doing is the repetitive, rule-shaped, high-frequency kind that a person is performing because nobody ever got around to handing it to a machine.

The question is never “what does automation cost?” It is “what is this process costing me already, and does the fix cost less than that?”