Most articles on custom software vs off-the-shelf are written for companies with 200 employees and a $150,000 budget. If you run a business with 4 to 20 people, that advice doesn't fit. You're not choosing between Salesforce and a six-figure enterprise build. You're choosing between a spreadsheet, a SaaS tool you've half-bent to fit, and something built specifically for how you work.
This is the version of that decision for your size of business.
The real question isn't "custom vs off-the-shelf"
It's "which parts of your business are standard, and which parts aren't."
Accounting, payroll, email: these are solved problems. Buy them. No small business should ever build its own invoicing system or its own email client. That's not where the decision gets interesting.
The decision gets interesting where your process doesn't look like anyone else's. A probate valuation. A prospecting workflow specific to your niche. A booking flow with rules a generic calendar tool can't hold. That's where off-the-shelf software starts asking you to change how you work to fit the tool, instead of the other way round.
Signs the tool is bending you, not the other way round
You'll notice a few things happening at once:
- You've built a spreadsheet next to the tool to track what it doesn't
- Someone spends part of every week re-entering the same data in two places
- You've paid for three subscriptions that each do a third of what you need
- The tool's "advanced" tier is priced for a team five times your size
- Onboarding a new person means teaching them the workarounds, not the tool
None of these appear on a budget line. The cost is time, spread thin across the team and never totalled up.
That's the real expense every build-vs-buy guide skips. Those guides assume a CFO who can model it. You don't have one. You just know Thursday afternoons disappear into admin.
What small businesses actually get wrong
Mistake 1: Assuming custom means enterprise. Most write-ups quote custom software costs starting at $50,000 and running past $500,000, with builds taking three to twelve months. That's enterprise pricing for enterprise scope.
A tool that fixes one constraint (a booking flow, an enquiry tracker, a valuation document) doesn't need that scope. It needs a day rate and a fixed price for a focused job.
Mistake 2: Comparing month one cost, not year two. A £30/month tool looks free next to a custom build.
Three tools at £30 to £150 a month, plus the hours spent stitching them together by hand, adds up faster than people expect. Run the maths over a year, not a month, before deciding.
Mistake 3: Waiting until the spreadsheet completely breaks. By the time a workaround system fails outright, it has already cost months of slow, quiet drag on the team. The better time to look is when you notice the workaround exists at all.
A real example: what "bending to fit the tool" looks like
Upper Floor, a social content agency, ran on Metricool for scheduling, a Google Sheet for prospects, and a different tab or tool for every client's research, analytics and ad spend. Nothing was broken. It was just spread out, and only one person knew where everything lived.
We built Signal: one web app with a staff view and a branded client login, covering the content planner, Instagram analytics, ad spend, a prospecting CRM, and a brand research engine that writes its own documents.
Every morning at 07:00, a scheduled job refreshes every client's numbers before anyone logs in, checks each metric against its own baseline, and posts an alert if something moves.
The agency now runs from one login instead of three subscriptions and a sheet. A second team member can take a client from onboarding to reporting without asking where anything lives, because there's only one place for it to be.
That's the pattern worth noticing: the individual tools weren't bad. The gap between them was the actual cost.
Another version of the same problem: paper and spreadsheets, not SaaS
Gorringe's, a long-established auction house, did probate valuations by hand across spreadsheets and paper. The same client and estate details typed in more than once. A document formatted manually every time. Totals checked that a system should have owned from the start.
We built a tool that takes the details once and produces the finished valuation document at the end, correctly formatted, totals handled automatically. Producing a valuation is now around 70% faster, and it comes out consistent every time, regardless of who's doing it.
No off-the-shelf valuation product existed for this. The workaround wasn't a bad SaaS choice. It was the absence of one. That's a different signal than "the tool doesn't fit," but it points to the same answer: build the specific thing.
A simple way to decide
Skip the 40-point scoring frameworks. For a business your size, three questions do most of the work:
1. Is this a solved problem, or your problem? If every competitor in your space could buy the same tool tomorrow and run it the same way, it's a solved problem. Buy it. If the workflow is specific to how you sell, deliver or value your work, that one is yours to own.
2. How much of your week goes into working around the gap? Count it honestly for one week. Re-entering data, exporting and importing, chasing the same information across two tools. If it's more than a few hours across the team, that's the real cost of staying off-the-shelf, and it compounds every week you don't fix it.
3. Could a focused build remove the constraint, not replace everything? You don't need to replace your whole stack. One tool (a booking flow, an enquiry workspace, a document generator) removes the actual bottleneck. Keep the SaaS tools that work. Build the piece that doesn't exist yet.
If the answers point to "this is ours, it costs us real time, and one tool would fix it," that's the signal to look at a custom build. If not, keep buying.
Where no-code platforms fit into this
No-code and low-code tools (Bubble, Retool, Glide) sit in an awkward middle ground. They're faster to start than a full custom build, and cheaper than three separate SaaS subscriptions. But they come with their own version of the bending-to-fit problem.
Most no-code platforms handle the standard 80% of a workflow well, then hit a wall on the edge cases: the exception, the odd approval step, the calculation that doesn't fit their data model. And because you're building inside someone else's platform, you don't own the code at the end.
If the platform changes its pricing, or you outgrow what it can do, migrating off it is its own project.
If your workflow is simple and stable, a no-code tool can be a reasonable stopgap. If it has real edge cases, or you're already picturing the workaround you'll need in six months, that's the same signal as with any off-the-shelf tool: build it once, properly, and own it.
What this costs at your size
This is where guides go vague. Here's what it actually looks like for a focused build: £500 a day, one day minimum, fixed price agreed before work starts. If the build overruns, that's absorbed, not passed on.
Most focused tools (a booking system, an enquiry tracker, a valuation generator like Gorringe's) land somewhere between one and five days. You own the code at the end. No licence, no per-seat pricing, nothing to renew.
Compare that against the SaaS stack it replaces over a year, not a month. The maths is closer than people expect. In some cases it's cheaper from day one.
Where to start
You don't need to guess which category your problem falls into. A free systems audit finds the actual constraint (the one thing capping how much work your team gets through) and tells you honestly whether it's worth building something or better to buy an existing tool. Half the time the answer is "go buy X."
If you want that answer for your own business, start with an Operations Check. It's a 30-minute call, no commitment to build anything after it.
You can also see more of what a focused build looks like on our work page, including how Signal and the Gorringe's valuation tool were scoped and shipped. And if cost is the open question, the pricing page has the actual day rate, not a "contact us for a quote."
Common questions
How much does custom software cost for a small business? For a focused build that solves one problem (a booking tool, an enquiry tracker, a document generator), expect somewhere between £500 and £2,500 at a £500/day fixed-price rate, with most jobs landing in the one-to-five day range.
Larger, multi-stage builds cost more but are scoped and priced stage by stage, so you're never committing to the full number upfront.
Is custom software worth it for a 4-person team? It's worth it when one specific process is capping how much work the team gets through, and no off-the-shelf tool fits it without heavy workarounds. It's not worth it for solved problems like accounting or email. The size of your team matters less than whether the process is genuinely yours.
How long does a custom software build take? A focused build removing one constraint ships in one to five days. Larger builds are staged, with something working in the first week and each stage priced and delivered before the next begins.
What happens if I outgrow off-the-shelf software later? Most businesses start with SaaS tools and migrate specific processes to custom as the gaps get expensive. That's a normal path, not a failure of the original decision. The businesses that get stuck are the ones that never revisit the decision at all.
The short version
Buy the solved problems. The ones that are genuinely yours, build once and own. Most businesses don't need an enterprise custom software programme. They need one tool that removes the thing capping their week, built at a price that matches a 4-person team, not a 400-person one.