What does bespoke software actually cost in the UK?
It genuinely does depend on the project. That is not a reason to tell you nothing, so here is the floor and what moves the number above it.
Ask five agencies what a custom system costs and you will get five versions of "it depends", which is true and useless. It genuinely does depend on the project, more than most people expect, but that is not a reason to tell you nothing.
So here is the one number we can give you honestly: simple projects start from around £1,500. A single workflow, one type of user, one job done properly. That is the floor, and it is worth knowing because the most common reason people never ask is a fear that the answer starts with a five-figure sum.
Above that, the price follows the shape of what you are building rather than any price list. What follows is what actually moves it, so you can work out roughly where your own project sits before you speak to anybody.
What actually moves the number
Five things, roughly in order of how much they matter.
How many different kinds of user there are. Not how many people, how many kinds. A system with one type of user is a fraction of the cost of one where a customer, a field worker, an office administrator and a manager each need a different view with different permissions. This is the single biggest driver and the one clients underestimate most.
How much it has to talk to. A system that stands alone is straightforward. A system that reads from your accounts package, writes to your CRM and files to a government endpoint costs more, because each connection is somebody else's rules, somebody else's downtime and somebody else's version changes.
Whether there is a mobile app. A web application that works well on a phone is much cheaper than a native app in the app stores. Native is right when the job needs the camera, offline working or location in the background. It is an expensive way to get something a responsive website would have done.
How exact the requirements are. Not how detailed the document is, how many decisions are genuinely settled. Projects run over when nobody has decided what happens in the awkward cases, and the awkward cases are most of the work.
What happens to the data you already have. Migrating twelve years of records out of an old system, cleaning them and proving nothing was lost is frequently a fifth of the project and is almost never in the initial estimate.
Where the money goes
Whatever the total, the shape of a project is fairly consistent, and it is useful for sense-checking any quote you are given:
| Stage | Roughly | What it buys |
|---|---|---|
| Discovery and design | 15% | Working out what it does, screen designs, decisions written down |
| Core build | 55% | The application itself, all user types and permissions |
| Integrations | 12% | Connecting to the systems you already run |
| Data migration | 8% | Moving and cleaning what is already there |
| Testing and launch | 10% | Fixing what testing finds, deploying, training |
If a quote has almost nothing against discovery, that work has not been removed, it has been moved to the middle of the build where changing your mind is far more expensive.
Then the part people forget: a live system needs a budget to keep running. Hosting, security updates, the changes browsers and phone operating systems make without asking you, and the small improvements that keep people using it. Software is not a thing you buy once, and anyone who lets you believe otherwise is storing up a bill for year two.
Where the money is wasted
Three patterns, all of them expensive and all avoidable.
Building everything at once. The version that does all twenty things costs four times the version that does the six things people actually do daily, and it takes long enough that the requirements change before it ships. Build the six, use them, then decide.
Paying to rebuild what you could buy. Accounting, payroll, email marketing and general CRM are solved problems with mature products. Custom software earns its money on the parts of your business that are genuinely yours, and the parts nobody sells off the shelf. We would rather talk you out of the wrong project than take the money, and the choosing between bespoke and off the shelf question is worth settling before you brief anyone.
Cheap builds you cannot maintain. A system built quickly by whoever quoted lowest, with nothing documented and no tests, does not save money. It moves the money to the next agency who has to work out what the first one did.
What to ask before you commit
- What happens in month thirteen, and what does it cost?
- Who owns the code and the data, and how do I get them if we part company?
- What is deliberately not in this quote?
- Show me something similar you built and still support.
That last one matters more than the others. Plenty of firms can build something. Fewer are still answering the phone about it three years later, which is the question worth asking about any agency.
How we quote
We look at what you are trying to build, tell you what drives the cost in your particular case, and put a number in writing before anything starts. If the budget you have cannot build the thing you want, we will say so at that point rather than after the deposit, and usually suggest what the smaller first version looks like.
You can see the range of what that has meant in practice across our client work, from an operating platform grown over several years to a platform going through HMRC approval. If you want a number for your own project, tell us what you are trying to build.
Sounds like your situation?
A 20-minute conversation, in plain English. We’ll tell you honestly what it would take.