What automation costs a small business, and why quotes differ

August 3, 20269 min read

Ask what it costs to automate a process and you will get numbers between $1,500 and $50,000, often on the same afternoon. None of the pages quoting those figures explains how they were reached, which makes them useless for planning a budget.

This is an attempt to answer the question properly: what the money actually buys, which part of it is published and checkable, which part depends on your business, and how to decide whether the whole thing is worth doing. Where a number cannot be verified, it is left out rather than repeated.

Why do quotes for the same job differ so much?

Short answer: "automate my invoicing" describes a wish rather than a piece of work. Two suppliers hearing that sentence can quote for two different jobs, ten times apart in price, and both can be honest about it.

Say you want overdue invoices chased automatically. One reading of that is a reminder email on a schedule, wired to the accounting tool you already pay for. Another reading is a system that watches payments land, works out which client is late and by how much, adjusts tone by how long the invoice has been sitting, stops the moment somebody replies, and hands anything awkward to you before it damages a relationship you spent years building.

Same sentence. Different jobs. The gap between them is where every wide quote range comes from, and it closes only when somebody looks at how your invoicing actually works today.

What are you actually paying for?

There are three costs, and confusing them is what makes budgets go wrong.

Software. The platforms the automation runs on. Published pricing, roughly predictable, paid monthly for as long as the automation lives.

The build. Somebody working out what should happen, in what order, with what exceptions, then wiring it together and testing it against real cases. Paid once, though "once" usually means a few rounds as reality corrects the plan.

Upkeep. Your bank changes an export format. A supplier renames a field. The automation breaks quietly, and quiet is the dangerous part: nothing announces that reminders stopped going out three weeks ago.

Most quotes cover the second cost and mention the first. The third gets discovered later, which is why a project that looked cheap in month one can feel expensive in month eight. Budget for it from the start, even roughly.

What does the software itself cost?

This is the one layer you can check yourself before talking to anyone, because the vendors publish their prices.

Platform	Entry price	What you get	Self-hosting
Zapier	$19.99/mo billed annually	750 tasks	No
Make	$9/mo	10,000 operations	No
n8n	~$20/mo cloud, free self-hosted	2,500 executions	Yes

All three have free tiers big enough to prove an idea works: 100 tasks a month on Zapier, 1,000 operations on Make, unlimited on a self-hosted n8n if you can run a small server.

Those prices are current as of August 2026 and worth re-checking with the vendor, since tiers move. What matters more than the numbers is the last column in each vendor's pricing table, which almost nobody reads before signing up.

Why does the same workflow cost ten times more on one platform?

Because the three platforms count three different things, and the unit they count decides your bill far more than the headline price does.

Zapier counts steps. Every action in a Zap is one task. A five-step Zap that fires 150 times a month consumes 750 tasks, which is the entire Professional tier. Go over and the overage bills at 1.25 times your base rate.

Make counts operations. A scenario with a trigger, three filters, a lookup and two actions burns seven operations every single time it runs.

n8n counts executions. One complete run of a workflow is one execution, whether it contains four nodes or forty.

Take one automation firing 50,000 times a month. On Make that runs to somewhere around $48-96. Self-hosted on n8n it is $10-15, because you are paying for a server rather than for operations.

The practical consequence for a small business: ask any supplier which platform they intend to use, how many steps the automation will have, and how often it will run. Multiply those together yourself. A quote that specifies the build price but leaves the running cost as "plus platform fees" is hiding the number that grows with your success — the busier you get, the more it costs, and on a step-counting platform it grows faster than your volume does.

What makes a build expensive?

Five things, in the order they usually bite.

How many systems have to talk. Two tools with proper APIs is straightforward work. Five tools, one of which only exports a spreadsheet by email, is a different project entirely.
How many exceptions exist. The main path is rarely the problem. The cost sits in "unless the client is on a payment plan", "except in December", "but not for the two accounts my business partner handles personally".
How clean your data is. If the same customer exists three times with different spellings, the automation will faithfully send three reminders. Cleaning that up is real work and it comes before the automation, not after.
Whether anything is custom. A standard connector is cheap. An integration with the booking system a local developer wrote for your industry in 2014 is not.
How often it runs. This drives the monthly bill rather than the build, but it compounds, and over two years it can exceed what the build cost.

Nothing on that list is about your industry. A dental practice and a law firm chasing the same number of documents through the same number of systems land in much the same place, which is why the four industries on our front page are illustrations rather than a filter.

How do you work out whether it is worth it?

The arithmetic is simple enough to do on paper, and doing it before you talk to anyone protects you from being sold something.

Take one task. Count how many times it happens in a month, and how long it takes each time. Multiply. That is the hours you would get back — the honest ceiling on what the automation is worth, before you subtract anything.

An example. Forty-five minutes on a working day, chasing invoices and re-sending reminders, comes to roughly fifteen hours across a twenty-day month. Value your own time at $50 an hour and that task is costing $750 a month in attention you could spend elsewhere. Against that, a build in the low thousands with $30 a month running cost pays back inside a year and keeps paying afterwards.

Three corrections to that number, all of which push it down:

You will not get all of it back. Automation that handles 80% of cases and hands you the rest is normal and good. Count 80%, not 100%.
Your time has a real price only if you fill it. Hours saved are worth money when they go into work that earns; if they go into inbox refreshing, the automation bought you nothing measurable.
Add the upkeep. Something breaks a couple of times a year and needs attention.

Run those corrections and plenty of automations still pay for themselves. Some do not, and finding that out on paper costs an afternoon rather than a project.

We do not quote before the audit, because a number given without understanding the work is only a guess. If you want the arithmetic above done properly against your business — every repeated task mapped, priced and ranked — that is what the process audit is for. It ends in a written map you keep, whether or not you hire us.

When is automating the wrong answer?

More often than anyone selling automation will tell you. Four cases where the answer is no.

The process is broken. Automation multiplies whatever it is given. A messy process becomes a fast messy process, and the mess arrives at your customers sooner. Fix the process first, on paper, by hand. Then automate the version that works.

Nobody can explain it. If you cannot describe how the task happens today to a new employee in five minutes, it is not ready. That is not a failure on your part; most processes live in somebody's head, and getting them onto paper is genuinely difficult. It is simply work that has to happen before any software.

It barely happens. A task you do twice a month saves you minutes and costs you a project. The maths does not care how annoying the task feels.

The human part is the point. Some conversations are the product. A client who has just been told bad news does not want a well-timed sequence, and automating that particular moment costs more in trust than it saves in time.

We would rather say this at the start than sell a build for its own sake. If an audit shows automation would not help, or that the honest fix is a change to how you work rather than software, you will hear that.

Why nobody can quote you properly over email

Everything above should make the reason obvious: the price depends on how many systems are involved, how many exceptions live inside the process, how clean the data is and how often the thing will run. None of that is visible from the outside.

A supplier who quotes a firm number from a two-line email is guessing, and the guess protects them rather than you. It will either be padded to cover what they cannot see, or it will be low enough to win the work and grow later, once you are committed.

The reasonable version is to charge a small amount for the looking, produce something you keep, and only then talk about the build. That is why our first step is a paid audit at $299 over three business days: a written map of what can be automated in your business, what each piece is worth, and what it would cost. The fee comes off the project if you go ahead, and the map is yours either way.