The real cost of off-the-shelf software isn't the monthly fee. It's the hour a day your team spends working around it. The spreadsheet that bridges two systems. The copy-paste between your CRM and your accounting package. The training you do every time someone joins because the software doesn't match how the job actually works.
That cost used to be the price you paid for not having a six-figure budget for custom software. Not any more. Bespoke software is now realistic for small businesses and charities, and the number one thing it buys you is software that fits your process instead of the other way round.
The Hidden Tax of Off-the-Shelf Software
Off-the-shelf tools look cheap on paper. A subscription here, a subscription there, a free plan for the small stuff. But the real bill shows up somewhere else.
Research from Pendo and others has found that around 80% of features in the average SaaS product are rarely or never used. Software vendors build for the broadest possible market, which means every customer gets a product shaped around an average nobody actually is. That's roughly $29.5 billion a year of R&D aimed at features most people ignore, and passed on in your subscription.
The cost you pay isn't just in licences. It's in:
- Workarounds. Spreadsheets, sticky notes, and "just for now" processes that quietly become permanent.
- Re-keying. Data typed into one system, then into another, then into a third for the trustees' report.
- Training. New starters learning your workarounds, not just the software.
- Errors. Numbers that don't match between systems because someone missed an update.
- Opportunity cost. All the work your team isn't doing because they're wrangling the tools.
One recent study put the average productivity loss from badly fitting software at around 15%. For a team of five, that's most of a person's salary disappearing into nothing.
What Actually Changed
Custom software used to start at £80,000 and take six months. It genuinely did. A couple of things have shifted at once, and they compound.
AI Coding Tools (Used Well)
The big one. Experienced developers using AI coding assistants are getting through repetitive work much faster. GitHub's own research put the speed-up at 55% on relevant tasks. McKinsey's study pegged the reduction in routine coding time at 46%.
The savings are biggest exactly where projects used to bleed hours: integration glue code, form plumbing, data validation, database queries, admin screens. The boring 60-70% of most builds. That's where your bill used to come from. Now it doesn't.
Mature Off-the-Shelf Parts
Ten years ago, building a secure login, a payment flow, or a file upload meant weeks of careful work. Today those are proven components your developer drops in. The custom bit is the logic that makes your business different, not the plumbing around it.
Cloud Hosting That Costs Almost Nothing
A typical small business application running on modern serverless hosting costs under £50 a month. Some of the ones we run come in under £10. You no longer need a server, a sysadmin, or a dedicated IT person to keep the lights on.
A Real Example: Getting Donorfy to Talk to Xero and Your CRM
Here's a pattern we see constantly with charities. You've got Donorfy for fundraising and Gift Aid, Xero for accounts, and a supporter engagement tool (or a custom CRM, or a mailing platform). On paper they all "integrate." In practice, someone on your team spends half a day a week making sure the numbers match.
The pain usually looks like this:
- A donation comes into Donorfy. Great.
- It needs to end up in Xero, coded to the right fund, with the right tracking, matched against the bank feed.
- The donor's details and engagement history need to live in your CRM so fundraisers know who's who.
- Gift Aid claims need to be ready to submit without anyone reconstructing them from three places.
- The trustees want a report on Monday that pulls from all of it.
Donorfy's built-in Xero integration handles the basics, and it's genuinely good. But the moment you need slightly different tracking codes, split allocations between restricted and unrestricted funds, sync donor data into a CRM that isn't on the standard partner list, or trigger a thank-you workflow based on gift size, you're outside what the ready-made connectors do.
A custom integration service that sits between the three, listens for new donations, codes them correctly, and keeps everything in step, used to be a £25,000 to £40,000 project. Today, for most charities, we'd expect it to land in the £5,000 to £12,000 range, delivered in weeks rather than months. It pays for itself in the first year on the admin time alone, and that's before you count the fundraising you actually get to do instead.
The question isn't "can we afford custom software?" any more. It's "can we afford to keep losing half a day a week per person to software that nearly works?"
The Catch: AI Only Helps People Who Know What They're Doing
This part matters, because the internet is full of breathless claims about "anyone can build software now." That's not true, and the research is starting to back it up.
A 2025 study from METR, the most careful one done so far, found that experienced developers using AI on complex codebases were actually 19% slower than without it. They felt faster, but the data said otherwise. The reason: you have to review everything the AI writes, because it will confidently produce code that looks right and isn't.
So how is AI making projects cheaper? Because the same study, and many others, show the gains are concentrated where it matters: well-defined, repetitive work done by people who can spot bad output immediately and correct course. On that kind of work, experienced developers are significantly faster. On the clever bits that actually matter to your business, they're slower but more careful. That's exactly what you want.
What doesn't work is handing the job to someone without the experience to tell good code from AI-generated nonsense. We get a steady stream of projects that started on Fiverr or with an in-house "vibe coder" and ended up as a pile of code nobody can maintain. We call that service AI Code Rescue, and it exists because cheap AI-assisted builds done badly are more expensive, not less.
The short version: the cost of custom software is coming down, but only when experienced people are in the driving seat.
When Custom Actually Makes Sense
Not every problem needs custom software. Off-the-shelf is still the right answer for most of what a small business or charity does. Basic accounting, email marketing, video calls, CRM for a standard sales process. If the tool fits, use it.
Custom becomes worth it when one or more of these is true:
- Someone on your team says "if only it did X" at least once a week.
- You've got a spreadsheet quietly propping up the gaps between two or three systems.
- You're paying for features you don't use to get the one you do.
- A subscription is going up every year and you can't leave because your data's trapped in it.
- You've outgrown the "one size fits all" tool and your process is now a competitive advantage you want to double down on, not flatten.
Rough rule of thumb: if your team is losing four or more hours a week per person to software workarounds, a well-scoped custom build almost always pays for itself inside a year.
The Bottom Line
For years the advice was "bend your process to the software, because the alternative is too expensive." That advice is out of date. Software that fits your business, talks to the tools you already use, and stops you paying a productivity tax every week is now within reach of organisations that would never have considered it five years ago.
If you've got a pair of systems that should be talking to each other, a process that's outgrown its off-the-shelf tool, or a productivity drag you've been tolerating for too long, we're happy to have a look and tell you honestly whether it's worth fixing. Drop us a line and we'll set up a chat.