You’ve outgrown your spreadsheets. The copy-paste workarounds aren’t cutting it anymore. Someone on your team (maybe you) is spending hours every week moving data between apps that should talk to each other.
So you start looking for solutions. And you land on two options: hire developers to build something custom, or use an automation tool to connect what you already have.
This is a bigger decision than it looks. The cost difference between these two paths can be €20,000 or more in the first year alone. And once you commit to one, switching is expensive.
Here’s what I’ve learned from helping small businesses navigate this choice.
What custom software actually costs
Most business owners dramatically underestimate this. When someone says “I want an app that does X,” the mental estimate is usually a few thousand euros. The reality is very different.
For a small to mid-sized business, a custom software project typically falls into one of these ranges:
| Project type | Cost range | Timeline |
|---|---|---|
| Simple internal tool (dashboard, form, data viewer) | €10K to 25K | 4 to 8 weeks |
| Business process app (CRM, inventory, order management) | €30K to 80K | 3 to 6 months |
| Customer-facing app or platform | €50K to 200K+ | 4 to 12 months |
These are estimates for working with a small development firm or experienced freelancers. Enterprise agencies charge more. Offshore teams charge less, but the coordination overhead often eats the savings.
Developer rates vary wildly. According to FullStack’s 2025 price guide, a quality freelance developer charges €60 to 120/hour. A small agency charges €90 to 160/hour. An enterprise firm starts at €250/hour and goes up. For a 500-hour project (a modest business app), you’re looking at €30K to 80K in labor alone, before hosting, design, or project management.
And that’s just the build. Maintenance starts the day you launch.
Year 2 is where custom software gets expensive
Building the app is the easy part (financially). Keeping it running is where the costs compound.
The industry standard for software maintenance is 15 to 25% of the original build cost per year. For a €50K application, that’s €7,500 to €12,500 annually, just to keep it working. Not to add features. Just bug fixes, security patches, server costs, and keeping up with third-party API changes.
| Year | Custom software (€50K build) | Automation tools |
|---|---|---|
| Year 1 | €50K + €5K maintenance | €3K to 6K total |
| Year 2 | €10K to 15K | €3K to 6K |
| Year 3 | €12K to 20K | €3K to 6K |
| 3-year total | €77K to 90K | €9K to 18K |
Maintenance costs tend to increase over time, not decrease. Years 1 to 2 run at the low end (10 to 25% of build cost). Years 3 to 5 climb to 15 to 30%. By year 6 and beyond, you’re often looking at 20 to 40%, especially if the original developer is no longer available.
That last point matters a lot. If the freelancer who built your app moves on, a new developer needs to understand the entire codebase before they can fix a bug. That ramp-up time costs money, and it happens every time you switch developers.
The O'Reilly 60/60 rule. Roughly 60% of all software lifecycle costs go toward maintenance, not the initial build. Some studies put this as high as 80% for long-lived systems. The build is the down payment. Maintenance is the mortgage.
Most custom projects don’t go as planned
The Standish Group’s CHAOS report has tracked IT project outcomes across 50,000+ projects globally. The numbers haven’t improved much:
- 31% of projects are delivered on time, on budget, with the right features
- 50% are “challenged” (late, over budget, or missing features)
- 19% fail outright (canceled or never used)
More than half of projects run 189% over the original cost estimate. A project quoted at €30K ends up costing €57K. This isn’t unusual. It’s the median outcome.
Small projects (under €1M) have a much better success rate, around 90%. But even “small” in software terms means something different than most small business owners expect. What feels like a simple request (“I want a dashboard that shows my orders and inventory in one place”) turns into a complex project once you factor in authentication, permissions, data migration, testing, error handling, and the inevitable “can we also add…” requests.
What automation tools can actually do
This is where I should be transparent about my bias: I build automations for small businesses using n8n, so I’m naturally going to favor this approach. But I also turn down projects where automation genuinely isn’t the right answer. Here’s where it works and where it doesn’t.
Automation tools work well for:
- Connecting apps you already use. Your Shopify store talks to your accounting software, your CRM syncs with your email, your inventory updates across channels. This is the bread and butter.
- Moving data on a schedule. Nightly reports, weekly summaries, daily syncs. Anything that runs at a set time and follows the same steps.
- Responding to events. A new order triggers an invoice. A form submission creates a CRM contact. An overdue payment sends a reminder.
- Data transformation. Reformatting, filtering, enriching, merging. Taking data from one format and putting it into another.
A typical automation setup for a 20-person company might include: order processing from Shopify to Xero, inventory sync across sales channels, automated invoicing, Slack notifications for key events, and a weekly report compiled from three different data sources. Total cost: €50 to 300/month on tools, plus setup from someone like me.
Automation tools don’t work well for:
- Complex user interfaces. If you need a custom dashboard with interactive charts, filters, and role-based views, automation tools aren’t the right fit. They’re backend connectors, not frontend builders.
- Customer-facing products. If the software IS the product you sell, you need custom software.
- Heavy computation. Machine learning models, real-time video processing, large-scale data analytics. These need dedicated infrastructure.
- Highly specific business logic. If your process has 50 branching conditions based on factors that change weekly, an automation workflow will become unmaintainable. At some point, you need code.
A simple framework for deciding
Before you commit to either path, answer three questions:
1. Do your apps already exist?
If you’re connecting Shopify, Google Sheets, Slack, Xero, HubSpot, or any other mainstream app, automation is almost certainly the right choice. These tools already have integrations built. Custom software would mean rebuilding connections that already work.
If you need an app that doesn’t exist (a unique workflow, a proprietary process, a customer-facing tool), you might need custom software.
2. Is the process mostly the same every time?
Automation excels at repeatable processes: the same steps, in the same order, with minor variations. If your order processing follows the same flow for 95% of orders, that’s a perfect automation candidate. You can handle the 5% edge cases manually or with branching logic.
If every instance is unique and requires human judgment at multiple steps, automation will be frustrating to build and maintain.
3. What’s your total budget for the first 3 years?
| 3-year budget | Best option |
|---|---|
| Under €5K | DIY automation (Zapier, Make, or self-hosted n8n) |
| €5K to 20K | Automation + consultant setup |
| €20K to 80K | Depends on complexity (see above) |
| €80K+ | Custom software makes financial sense |
The option most people miss
There’s a third approach that doesn’t come up enough: automation tools for the core logic, with a thin custom layer where you need it.
Here’s what this looks like in practice. A client needs order data from their e-commerce platform to flow into their accounting software, with specific formatting rules, duplicate detection, and a weekly summary report. The “all custom” version of this is a €30K+ application. The automation version is an n8n workflow that handles the data flow, with a simple Google Apps Script or lightweight web app for the parts that need a user interface.
The total cost is usually €2K to 5K for setup, plus €100 to 300/month for maintenance. That’s a fraction of the custom software price, and it’s built on tools that are already tested, documented, and maintained by large teams.
The key insight: most businesses don’t need a custom application. They need their existing applications to work together. That’s a fundamentally different problem, and it has a fundamentally cheaper solution.
Both paths have lock-in risks
People worry about vendor lock-in with automation tools, and that’s fair. If you build everything on Zapier and Zapier doubles their prices, you’re stuck migrating.
But custom software has its own version of lock-in: developer dependency. If one freelancer or agency built your entire system, and they become unavailable, you’re locked in to finding someone who can understand their code. That’s often harder and more expensive than migrating between automation platforms.
Automation tools at least have each other as alternatives. A workflow built on Zapier can be rebuilt on Make or n8n in days. The logic is visual and documented. Custom code can take weeks or months for a new developer to understand, and that’s assuming the original developer left good documentation (they usually didn’t).
Self-hosted tools like n8n reduce the automation lock-in risk further. Your workflows run on your server, your data stays with you, and the tool is open source. Nobody can change the pricing or shut it down on you.
Three scenarios, three different answers
Scenario 1: “I want my Shopify orders to automatically appear in Xero.”
This is automation. Every automation tool on the market can do this. A consultant can set it up in a day for €500 to 1,000. Custom software would cost €15K+ and take weeks. There’s no reason to build this from scratch.
Scenario 2: “I need a portal where my clients can log in, upload documents, and see their project status.”
This is custom software (or a low-code tool like Retool or Softr). Automation tools can handle the backend (processing uploads, sending notifications), but the client-facing interface needs to be built. Budget €15K to 40K.
Scenario 3: “I want to automate our entire order-to-invoice pipeline, including tax calculation, multi-currency handling, and stock level updates across three warehouses.”
This is the middle path. The data flow and logic can be automated. The warehouse systems and accounting software already exist. What you need is an automation layer that connects them, with maybe a simple dashboard to monitor the pipeline. Budget €3K to 8K, mostly setup and configuration, with €150 to 300/month ongoing.
What to do next
If you’re leaning toward custom software, get three quotes. Not one. Three. Compare timelines, ask about maintenance costs explicitly, and ask what happens if the original developer becomes unavailable. If the answer to that last question is vague, that’s a red flag.
If you’re leaning toward automation, start by listing every app you currently use and every manual process you want to eliminate. That list is the scope of the project. I wrote a practical guide to business automation costs that breaks down what each approach actually costs, from DIY to consultant to agency.
And if you’re not sure which path fits, that’s a normal place to be. It’s worth a 30-minute conversation with someone who builds both. I work with small businesses on automation, and part of that work is telling people when automation isn’t enough. Send me an email and we’ll figure it out together.