Buying advice11 March 20267 min read

How to choose a software development company in the UK

Portfolios are curated and testimonials are selected. Twelve questions that reveal how a supplier will behave in month four, plus the warning signs worth walking away from.

A team reviewing project work together in an office

Choosing a development company is difficult because the thing you are buying does not exist yet, and everybody selling it has a portfolio of the projects that went well.

What you are really assessing is how a supplier will behave when something goes wrong — because on a project of any size, something will. These questions get at that.

The twelve questions

Ask every supplier the same list and write down the answers. The differences will be more informative than the quotes.

01Who will actually write this code? Get names and roles. If the answer involves a partner agency or an offshore team, you are buying a different product from the one being demonstrated.
02Is the price fixed or an estimate? And if it is an estimate, what happened to the last three estimates they gave?
03Who owns the code and the repository on final payment? The correct answer is you, in full, in writing.
04Am I required to host with you? A yes is a business model built on lock-in rather than on the work being good.
05What is included after launch? How many weeks of fixes, and where is the line between a fix and a chargeable change?
06What happens when I change my mind in week six? Listen for a process — priced and dated change requests — rather than reassurance.
07How will I see progress? A working link that updates continuously beats a monthly status report every time.
08Can I speak to a client whose project went badly? The willingness to answer matters more than the answer.
09What is your response time if it breaks at 2am, and is that in a contract?
10What documentation and training do I get? Software nobody understands does not get used.
11What is your test and release process? You want automated tests and staged deployment, not a manual click-through before a Friday afternoon push.
12What would make you turn this project down? Anybody who has never turned work away is selling capacity, not judgement.

Warning signs

None of these are automatically disqualifying. All of them deserve a straight answer.

No price is discussed until several meetings in. Sales process, not engineering process.
The quote is dramatically below the others. Either something has been misunderstood, or it will be recovered later through change requests.
They agree to everything. A supplier who never says "that is a bad idea" is not bringing judgement to the table.
The portfolio has no live links. Screenshots are cheap and stock mockups are cheaper.
The contract is silent on intellectual property. This is never an accident.
They cannot explain a technical decision in plain English. Either they do not understand it or they do not want you to.

The most reliable predictor of a good project is a supplier who tells you something you do not want to hear during the sales process.

Big agency, small studio, or freelancer?

All three work. They fail differently, and that is what should decide it.

Large agencies bring process, cover for absence and the ability to absorb a big scope — at a higher rate, and with the person you met unlikely to be the person who builds it. Small studios give you direct access to the people writing the code and faster decisions, with less resilience if somebody is ill. Freelancers are the cheapest and the fastest for well-defined work, and the highest risk on anything long-running.

Match the shape of the supplier to the shape of the risk you can carry. For most small and mid-sized companies, a small studio that owns its own products is the best value — enough resilience to be safe, small enough that you talk to the builder.

Before you sign

Get four things in writing regardless of who you choose: a scope specific enough that both sides would agree whether it was met, a fixed price with named triggers for change, an intellectual property clause transferring everything to you on final payment, and a defined support arrangement for the first three months.

A supplier who resists any of the four is telling you something important, and it is cheaper to hear it now than in month five.

The short version

Ask every supplier the same twelve questions and compare the answers, not the price.
Confirm in writing who writes the code and who owns it afterwards.
Compulsory hosting is a lock-in model, not a technical requirement.
The strongest signal of a good partner is willingness to disagree with you before you pay.
Get scope, fixed price, IP transfer and post-launch support in the contract before signing.

Questions people ask about this

Should I choose a company with experience in my industry?

It helps but it is rarely decisive. Sector knowledge can be transferred in a fortnight of good discovery; engineering judgement cannot be acquired that quickly. Prioritise the quality of the questions they ask you over the logos on their site.

Is it safer to use a bigger agency?

Safer against absence and turnover, riskier on cost and on distance from the person doing the work. It depends on whether your bigger risk is continuity or the project drifting away from what you asked for.