One business I worked with was paying for 14 software subscriptions. Nine of them overlapped. Three tools were doing the same job in three different departments, and nobody had spotted it for two years.
That’s usually the moment the awkward question arrives. Build vs buy automation: should small businesses build their own tools, or keep paying someone else every month? Most owners assume buying is the safe choice and building is a vanity project for companies with in-house developers.
I think that’s backwards more often than people expect. Buy the boring things. Build the parts that actually make you money.
Most small businesses should buy commodity software and build only the workflows that connect it. Building becomes worthwhile when a process is specific to your business, when per-task pricing climbs faster than your revenue, or when the logic changes every month. In my experience, a typical small business loses 8 to 10 hours a week to manual admin.
Why the Build vs Buy Automation Debate Trips Up Small Businesses
Subscription sprawl is a genuine operating cost, not a rounding error. Every tool you buy comes with a licence fee, a login, an admin, an integration and a renewal date. The make-or-buy decision is one of the oldest problems in management, and Harvard Business Review has been writing about it for decades. The difference now is that the buying is frictionless and the sprawl is invisible.
The trap is treating it as a single yes or no. It isn’t. A modern automation setup has three layers, and each layer deserves its own answer.
The first layer is your systems of record: accounting, payroll, CRM, email, storage. Buy these. Always. The second layer is the connective tissue that moves data between them. The third layer is your business logic: how you score a lead, quote a job, chase a payment or prioritise a build.
Most of the value sits in layers two and three. Most of the off-the-shelf products are aimed at layer one. That mismatch is why so many owners feel like they’re paying a lot and still doing the work by hand.
Where the build or buy decision applies across the three layers.
The 4C Test for Deciding What to Build
The 4C test is a four-question filter for any automation decision: Core, Cost curve, Change and Capability. Run a process through all four before you commit budget. A process that fails on Capability will fail no matter how good the business case looks, because nobody will maintain it after launch.
Core: does this process make you money?
If the process is how you win or keep customers, it’s probably worth owning. A recruitment agency’s candidate matching logic is core. Its payroll is not.
I once worked with a firm whose quoting rules were unusual and genuinely better than their competitors’. No off-the-shelf tool understood them. Forcing that logic into a generic CRM made the quotes worse, so I built the rules instead.
The reverse also holds. If your competitors do it exactly the same way you do, buy it and move on. Nobody wins market share with a custom expenses tool.
Cost curve: what happens at ten times the volume?
This is the question owners skip, and it’s the expensive one. Ask how the price behaves when your volume triples, not what it costs today.
Zapier is genuinely simple and a good starting point, but its per-task pricing punishes growth. At 2,000 tasks a month it’s cheap. At 200,000 tasks a month, a workflow that saves you $400 of admin can cost more than the admin did.
n8n has a steeper learning curve, but you get control, the option to self-host, and one system to learn as you scale. If you want a broader comparison, I keep a list of the best workflow automation tools for small business updated.
Change: how often will this need rewiring?
Processes that change monthly are painful to rent. Every change means a support ticket, a feature request, or a workaround that nobody documents.
When you own the logic, a change is a 20 minute edit. That speed matters more than the licence fee.
One warning: automating a broken process just creates faster chaos. Map it first. Very often the mapping reveals the real problem was the process itself, and no tool fixes that. My TRACE method for mapping a process before automating it walks through how to do that in an afternoon.
Mapping a process on the whiteboard before automating it.
Capability: who owns this in 12 months?
Self-hosting swaps subscription fees for maintenance responsibility. That’s a real trade, not a free win. Someone has to patch the server, watch the logs and fix the workflow when a supplier changes their API.
If that person doesn’t exist and you have no partner, buy. A dead workflow that nobody can debug is worse than a manual process, because at least the manual process complains.
Before you launch anything custom, test it properly. My four-stage guide to testing n8n workflows covers the checks that catch problems before your customers do.
Should Small Businesses Build Their Own Tools? Try the Middle Path First
The middle path means buying the platform and building the workflows on top of it. You rent the infrastructure, the connectors and the interface, then write your own logic inside it. This is where most small businesses should land, because it removes the two big risks: no developer team needed, and no vendor owning your process.
In practice that looks like this. You buy Xero, your CRM and your email. You build the workflows that connect them, using a platform like n8n or Zapier. The n8n documentation is open and detailed enough that a technical operations person can get productive in a week.
You own the logic and the data. No vendor can hike prices or pull the rug on a feature you depend on. But you’re not writing software from scratch either.
The low-code market has matured fast, and analyst firms like Gartner publish regularly on where it’s heading. The short version: the gap between “buy a product” and “hire developers” has narrowed to something a small business can actually cross.
Build costs more upfront; buy costs more as it scales.
What Building Actually Costs
Custom automation costs more upfront and less over time. A typical connected workflow build for an SMB might cost $5,000 to set up, then around $50 a month to host and maintain. The equivalent stack of subscriptions and per-task fees often runs $300 a month with no upfront cost, which means the two paths cross at roughly month 20.
At month 12, buying looks cheaper: $3,600 spent versus $5,600. At month 24, the picture flips: $7,200 versus $6,200. By year three the gap keeps widening, and unlike a subscription, your build doesn’t get repriced at renewal.
The honest caveats matter here. Build costs run over when the process wasn’t mapped. Subscription costs run over when nobody audits the stack, which is how firms end up saving $8,500 a year by cutting software sprawl.
If you want to model this properly for your own numbers, my guide to calculating true AI automation costs includes the line items most people forget.
Three Times You Should Always Buy
Some decisions don’t need a framework. Buying wins outright when a process is regulated, commoditised, or dependent on data you don’t have. Payroll, tax filing and card payments fall into the first group, because the compliance burden alone outweighs any advantage from owning the logic.
The second group is commodity infrastructure: email, calendars, file storage, video calls. Nobody should build these. The vendors have thousands of engineers and you have a Tuesday afternoon.
The third group is anything you don’t have the data to support. Most chatbots fail not because the technology is poor, but because they’re launched without useful data and without an easy route to a human. Sometimes the honest answer is not to build one at all.
I apply that to my own work. I once built an automated moodboard generator, found the aesthetics needed a human eye, and deliberately pulled it back to a human-in-the-loop process. Automation should take people out of repetitive internal work; anything a customer feels stays with a person.
Making the Call
So, on build vs buy automation, should small businesses build their own tools? Buy your systems of record, build the logic that makes you distinctive, and rent the platform in between. Run every candidate process through the 4C test before you spend anything.
The cheapest option in year one is usually the most expensive in year three. And the best build decisions often start with deleting tools rather than adding them.
If you’re not sure which of your processes fall on which side, an automation assessment will tell you in weeks rather than months.
Frequently Asked Questions
Is it cheaper to build or buy automation for a small business?
Buying is cheaper in the first year, building is usually cheaper after roughly 20 months. A typical custom build costs around $5,000 upfront plus $50 a month, versus $300 a month for equivalent subscriptions. The crossover depends heavily on your transaction volume and how often the process changes.
Can a small business build automation without a developer?
Yes, for most workflows. Platforms like n8n and Zapier let a technically confident operations person connect systems without writing much code. You’ll still want expert help for anything touching payments, personal data, or a process that would break the business if it silently failed at 3am.
What should you never build in-house?
Never build payroll, tax filing, card payments, email hosting or file storage. The compliance risk and maintenance load are far higher than the cost of a licence. Build the workflows that connect these systems instead, because that’s where the wasted hours and the competitive advantage actually sit.
How do I know if a process is worth automating at all?
Map it first, then count. If the process happens weekly or more, follows clear rules, and costs a person more than an hour each time, it’s a candidate. If the steps change every time or need human judgement, fix the process before you automate anything.
