Somewhere in the business, a person copies information from one screen to another. An enquiry arrives by email and gets typed into a spreadsheet. A booking comes in and somebody adds it to the calendar, then sends a confirmation, then tells the person doing the job. Each step takes two minutes. Together they take an afternoon a week, and they go wrong on the busy days.
There is a category of tool built for exactly this, and most small businesses have heard of at least one of them without being sure what it is for.
What a workflow tool actually does
It waits for something to happen in one place and then does something somewhere else. A form is submitted, so a row is added, an email is sent and a message lands in the team chat. A payment clears, so an invoice is created and filed. Nobody writes a program. You pick a starting event, pick the steps, connect the accounts, and it runs.
The well-known names are Zapier, Make and n8n. They overlap a great deal and differ in ways that matter more after a year than on the first day. Power Automate is the same idea inside Microsoft 365, and if most of your work already happens there, that question is worth answering first.
Why this is often the right answer
It is cheap to try. Most of these tools have a free tier or a low monthly price at small volumes. Building the first automation takes an afternoon, not a project.
It connects things that were never designed to talk. Your booking system, your accounts package, your email marketing and your spreadsheet were made by four different companies. Each one connects to these tools. That is the whole value.
It is priced by use, not by headcount. A process touched by fifteen people costs the same to run as one touched by two. That is the opposite of most business software, and it is the main reason a workflow tool can be dramatically cheaper than the platform alternative.
Where it stops being the right answer
When the logic gets complicated. One trigger and four steps is a workflow. Forty steps, six branches and a lookup table is a program drawn as a diagram, and it is harder to read and harder to change than the program would have been.
When the volume climbs. Pricing is based on how much runs, counted as tasks, operations or executions depending on the tool. At small volumes it is trivial. A process that fires thousands of times a day can turn a cheap tool into a significant monthly bill, and the bill grows with the business rather than staying flat.
When it holds sensitive data. Every step passes your data through the tool’s servers. For a newsletter sign-up that hardly matters. For health information, financial records or anything about children, it is a question to answer deliberately rather than by default.
When nobody owns it. These tools make it easy for anyone to build something, which is the appeal. It also means a business can end up with thirty automations, built by five people, under three personal accounts, documented nowhere. That is how automations end up failing quietly.
The difference between the tools that matters
Not features. They all connect to most things you will use. Two differences are worth knowing.
Hosted, or run yourself. Zapier and Make are services: you sign up and they run everything. n8n can be used the same way, or installed on your own hosting. Running it yourself changes the cost from “per run” to “the cost of a server and of keeping it updated”, which at high volume can be much cheaper, and keeps the data on hardware you control. It also makes you responsible for it, which is only a saving if someone actually does the looking after.
How readable it is to the next person. Whatever you build, somebody else will one day have to understand it. A tool with clear naming, comments and a visible history of runs is worth more than one with a few extra connectors.
Three situations, worked through
A physio clinic copying web enquiries into a spreadsheet. Ideal. One trigger, two steps, low volume, and a person no longer retyping. A hosted tool on its free or entry tier does this well, and it is running by the end of the day.
An events company sending reminders, invoices and supplier briefs for every booking. Still a good fit, with care. The volume is moderate and the steps are many, so the monthly cost is worth estimating against a realistic busy season rather than an average month. It also earns a page of notes on what runs, because three people will depend on it.
A business processing thousands of orders a day through a chain of automations. The tool is now doing the work of an application, priced as if it were an occasional convenience. Either running the workflow tool on its own hosting or moving the core of it into a small application usually costs less within the year, and is easier to reason about.
Five things worth doing
- Write the process down as steps before opening any tool. If it does not fit on a page, it may not be a workflow.
- Estimate the volume at your busiest, not your average, and price it on the plan you would actually need.
- Build under a business account, never a personal one, so it survives someone leaving.
- Turn on failure alerts to a shared address. A workflow that stops raises no complaint, only an absence.
- Keep one list of what exists: what it does, what it connects, whose account it uses. Ten minutes now, hours saved later.
Common misconceptions
- “No code means no expertise.” It means no programming. Designing a process that handles the awkward cases, the duplicates and the failures is the same skill it always was.
- “It is only for tech companies.” The best uses are dull and very common: enquiries, bookings, invoices, reminders. Exactly the work small businesses do most.
- “Once it is built, it is done.” Accounts get disconnected, forms change, and tools update their connectors. Something has to notice.
- “Self-hosting is always cheaper.” Cheaper per run, more expensive per hour of attention. It pays off at volume or where data must stay put, not otherwise.
The honest summary
For most small businesses, a workflow tool is the cheapest way to remove the copying and pasting that eats a week. It is quick to start, priced fairly at modest volumes, and it connects software that was never meant to be connected.
It stops being the answer when the logic grows into a program, when the volume turns a small subscription into a large one, or when nobody is responsible for what has been built. Those limits are worth knowing at the start, because they are much cheaper to plan for than to discover.
If there is a process in your business that is mostly people moving information between screens, describe the steps. The answer is often a workflow that can be running within the week, and sometimes it is that the product you already pay for can do it without one.
Common questions
Is n8n better than Zapier for a small business?
Neither is better in general. Zapier is the simplest to start with and has the widest range of connections. n8n is more flexible and can be run on your own hosting, which suits higher volumes or data that needs to stay put. Make sits between them. Choose based on volume, data and who will maintain it.
Do we need a developer to set these up?
Not for simple automations. Someone comfortable with the software the business already uses can build them. A developer, or someone experienced, earns their fee on the design: handling failures, duplicates and edge cases, and on anything running at volume.
Is it safe to connect our business accounts to these tools?
The established tools are used by very large organisations and take security seriously. The questions to ask are about your own setup: whose account each connection uses, who can edit the automations, and whether the data passing through is sensitive enough to need a closer look.
What does it cost to run?
At small volumes, often nothing or a modest monthly fee. Costs rise with the number of runs, so estimate your busiest month rather than your average one before deciding.