Custom software vs off-the-shelf: which is right for your business?
Most advice on this question is written by people selling one of the two options. Here's the framework I actually use with clients, including the cases where I tell them not to build.
Almost every article on this topic is written by someone selling one of the two options. So let me be upfront: I build custom software for a living. That gives me an obvious bias, and it also means I have spent a lot of time watching businesses buy the wrong thing in both directions.
The honest answer is that off-the-shelf software is the right choice most of the time. It is cheaper, faster, and someone else maintains it. Custom software wins in a narrower set of situations than most agencies will admit. This post is about telling those situations apart.
Start with the real question
"Should we build or buy?" is the wrong opening question, because it jumps straight to a solution. The better question is: where is the work actually going wrong?
In practice the pain almost always falls into one of three buckets, and each points to a different answer:
- We don't have a tool for this at all. Buy something. Almost always. You do not need a bespoke system to send invoices.
- We have tools, but they don't talk to each other. This is an integration problem first. Sometimes a connector solves it; sometimes it needs a system built around the join.
- We have a tool, but our process doesn't fit it. This is the real custom-software case, and the one worth examining carefully, because the tool is sometimes right and the process is wrong.
When off-the-shelf is the right call
Buy when the job you need doing is a job thousands of other businesses also need doing, in roughly the same way. Payroll, accounting, email marketing, e-signatures, card payments. These are solved problems with mature, well-regulated products behind them. Building your own is not clever, it is expensive.
Buy when the regulatory surface is large and not your speciality. Payroll is the clearest example: the rules change every year and getting them wrong is costly. Let someone else absorb that.
And buy when you genuinely cannot yet describe your own process. If three people in your business would describe the workflow three different ways, custom software will simply encode the confusion at greater expense. Fix the process on paper first.
When custom actually wins
Custom software earns its cost when the way you work is genuinely different from the market default, and that difference is the reason customers choose you.
1. Your workflow is your competitive advantage
A clinic with an unusual triage process, a distributor with pricing rules nobody else runs, a manufacturer with a bespoke quality-control sequence. If a generic tool would force you to work like everyone else, and working unlike everyone else is why you win, you are paying to erase your own advantage.
2. You are paying humans to be middleware
This is the most common and most measurable case. Someone on your team exports a CSV from one system, reformats it, and pastes it into another. Every day. That person's salary is the real cost of not integrating your systems, and unlike a software licence, it grows as you grow.
It is also easy to quantify, which makes it a good place to start. Two hours a day of copy-paste, at a fully loaded cost of £18/hour, is roughly £9,000 a year. Ongoing. That is a budget.
3. Per-seat pricing has outgrown you
SaaS pricing is designed to scale with your headcount, not with the value you get. At five users, £40/user/month is trivial. At sixty users across three sites, it is £28,800 a year for software you do not own and cannot change. Run that number over five years before you assume buying is cheaper.
4. You need the data to be genuinely yours
If you handle patient records, financial data, or anything else where an export request or a compliance audit is a real possibility, owning the database matters. See our practical guide to UK GDPR for small business software for what that actually involves.
The comparison, honestly
| Off-the-shelf | Custom | |
|---|---|---|
| Time to first use | Days | Weeks to months |
| Upfront cost | Low or none | Significant |
| Cost at scale | Grows with headcount | Largely fixed |
| Fit to your process | You adapt to it | It adapts to you |
| Maintenance | Included | Yours to fund |
| Data ownership | Vendor-controlled | Yours |
| Risk if vendor fails | High, no exit | Low, you hold the code |
Notice that the trade is mostly time and maintenance in exchange for fit and ownership. If fit and ownership are not costing you anything today, do not buy them.
The hybrid answer almost nobody mentions
The framing of "build or buy" is itself a false binary. The most cost-effective systems I build are usually neither: they are a thin custom layer sitting on top of tools the business already pays for.
Keep Xero for accounting. Keep the payment provider. Keep the email platform. Then build the one thing that does not exist off the shelf: the workflow that ties them together and matches how your business actually runs. You get the fit without paying to rebuild solved problems.
The goal is not to own all of your software. It is to own the part that makes you different.
How to decide this week
- Write down the workflow as it actually happens, not as the manual says. Include the WhatsApp messages and the spreadsheet nobody admits to.
- Mark every step where a human moves data between systems by hand. Total up the hours per week.
- Price the off-the-shelf option properly, at your headcount, in three years, with the add-ons you will actually need.
- Ask whether the steps that do not fit are genuinely your advantage, or just habit. Be ruthless here.
- If the misfit steps are your advantage and the manual hours are material, custom is worth costing. Otherwise, buy.
If you get to step five and you are not sure, that is a perfectly reasonable place to stop and get a second opinion. It is a fifteen-minute conversation, and I would rather tell you to buy something than build you a system you did not need.
Frequently asked questions
Not over a long enough horizon. Custom software has a high upfront cost and a largely fixed running cost, whereas per-seat SaaS grows with your headcount. The crossover point depends on team size. For a small team it may never arrive, and for a growing multi-site business it often arrives within three to five years.
A focused first version that replaces one painful workflow typically takes four to eight weeks. Full ERP or CRM replacements are staged over months, and should be delivered in usable pieces rather than as a single launch, so the business gets value before the whole thing is finished.
This is the right question to ask, and the answer should be contractual. You should hold the source code, the repository, and the hosting accounts in your own name from day one. If a developer will not agree to that, treat it as a warning sign.