What It Costs to Build an App: The Numbers Nobody Puts in the Quote

Published ranges put a simple MVP at roughly £8,000–£30,000, a typical consumer app with accounts, payments and a backend at £30,000–£100,000, and a complex product with real-time features and integrations well past £100,000. Those numbers are broadly honest and they answer the wrong question. The build is a one-off; the app is a monthly bill for as long as it exists — and the second number is the one that decides whether the thing survives.

So this splits into what actually drives the build price, and the running costs that rarely appear in a quote at all.

What moves the build number
Who builds it
The running costs nobody quotes
Making the hosting number concrete
Spending less without building less
A worked sketch
How this fits the rest of the stack
FAQ

What moves the build number

Cost tracks features and complexity far more than platform choice or team location.

User accounts and authentication. Signup, login, password reset, sessions, permissions. Small individually, and the point at which you acquire a database, an email provider, and a security surface.
Payments. Integration, webhooks, refunds, failed charges, subscription state, tax. The unhappy paths are most of the work.
Real-time features. Chat, presence, live updates. A different architecture from request-response, not a feature bolted onto it.
Third-party integrations. Each one is an API to learn, credentials to manage, and a failure mode to handle when it goes down.
An admin interface. Almost always forgotten in the quote and always needed within a month of launch.
Native mobile. Two platforms, two review processes, two release cycles. Cross-platform frameworks reduce this without eliminating it.
Design. A bespoke interface costs meaningfully more than a component library, and the difference is often invisible to users.

The reliable way to reduce the number is to cut features, not corners. Every capability is design, build, test, and permanent maintenance — the fourth being the one nobody prices.

Who builds it

Freelancers — £25–£120 per hour depending on market and seniority. Cheapest, highest variance, and there is a bus-factor of one.
Offshore agencies — often half to a third of onshore rates. Works well with clear specifications and less well with evolving ones, where communication overhead eats the saving.
Onshore agencies — £600–£1,200 per day. More expensive, more process, easier accountability.
In-house — salary plus employer costs plus equipment plus recruitment. Expensive to start, cheapest over years, and the only option that accumulates knowledge you keep.
Building it yourself — free in cash, expensive in time, and time is the resource an early product has least of.

The pattern that fails most often is a fixed-price contract for a product nobody has fully specified. Fixed price means fixed scope, scope always changes, and every change becomes a negotiation instead of a decision. Time and materials with a trusted team is usually cheaper in practice, even though it looks riskier on paper.

The running costs nobody quotes

This is the part that gets discovered rather than planned, and where a launched product quietly becomes unaffordable.

Hosting — the application, the database, storage, bandwidth. Small at first and it grows with usage.
Domain and certificates — a domain is £10–£40 a year; certificates should be free and included.
Email — transactional email for signups and receipts. Usually cheap, and non-optional.
Error tracking and monitoring — you can start without it and you will not want to for long.
App store fees — $99/year for Apple, $25 once for Google, plus their cut of in-app purchases.
Third-party APIs — payments take a percentage; enrichment, mapping, and AI services bill per call and are the most likely to surprise you.
Maintenance — dependency updates, security patches, OS and framework upgrades. Budget 15–20% of the build cost annually to keep it working without adding anything.

That last item is the one that sinks projects. An app that cost £40,000 needs roughly £6,000–£8,000 a year just to remain functional. Skip it for two years and you have a codebase with known vulnerabilities and dependencies too old to upgrade incrementally — at which point the choice is a large remediation project or a rewrite.

Making the hosting number concrete

Hosting is the running cost most often hand-waved as “a few pounds a month”, and it is the one you can actually pin down before you build.

A typical small application is four things: a web service, a database, some storage, and bandwidth. On RunxBuild’s plan ladder those are separate, priced items rather than a bundle:

A Basic plan is 0.5 vCPU and 624MB at $6/month — enough for a low-traffic API or a small app.
BasicMini at 1 vCPU and 1GB is $13, BasicPlus at 1 vCPU and 2GB is $20.
BasicMax at 2 vCPU and 4GB is $65, and Standard at 4 vCPU and 8GB is $150.
A managed MySQL or Postgres instance sits on the same ladder, so a small database is another $6–$20.
Static sites include 120GB bandwidth, then $0.10/GB.

So a small product — a service on Basic, a database on Basic — starts around $12/month and scales up the ladder as traffic grows. That is a number you can put in a spreadsheet before writing any code, which is worth more than a range.

The reason to itemise rather than take one figure is that it tells you which part grows. If bandwidth is the line that doubles, that is a different problem from the database plan being the constraint, and you can only see that when they are separate numbers.

Spending less without building less

Ship the smallest version that answers the real question. Most features in a v1 are built on an assumption about users that has not been tested.
Use boring technology. Familiar frameworks with large communities cost less because problems are already solved and people are easy to find.
Buy the commodity parts. Auth, payments, email, and search are solved. Building them yourself is expensive and rarely better.
Web before native. A responsive web app skips two app stores and two review cycles. Go native when there is a reason beyond preference.
Do not over-provision. Start on a small plan and scale when the metrics say to. Paying for capacity you might need in a year is money spent on a guess.
Own the code and the accounts. Repository, hosting, domain, and third-party accounts in your name. Getting this wrong turns a routine handover into leverage.

That last point is not a cost item until it is, and then it is the most expensive one on the list.

A worked sketch

A subscription web app: accounts, payments, a dashboard, an admin area, some email.

Build — £25,000–£45,000 with a small agency or two experienced freelancers over three to four months.
Hosting at launch — a service and a database on small plans, roughly $12–$30/month.
Email — free to £20/month at low volume.
Payments — around 1.5–3% plus a fixed fee per transaction, so it scales with revenue rather than being fixed.
Monitoring — free tier initially, £20–£50/month later.
Maintenance — £4,000–£7,000 a year to keep it current, before any new features.

So: a mid-five-figure build, under £100 a month to run at launch, and a low-five-figure annual commitment to keep it alive. The hosting is genuinely small early on — the maintenance is not, and it is the line most often left out.

The costs that grow are hosting with usage and third-party APIs with volume. Both are predictable if you itemise them at the start rather than discovering them from an invoice.

How this fits the rest of the stack

Build costs run from roughly £8,000 for a genuine MVP to well past £100,000 for a complex product, driven by features rather than by platform. The number that decides whether the app survives is the annual one: hosting, third-party services, and 15–20% of the build cost each year in maintenance.

Cut features rather than quality, buy the commodity parts, start on small plans, and make sure the accounts are in your name. Before committing to a build, put the running costs in the spreadsheet alongside it — the RunxBuild hosting calculator itemises the service, the database, the storage, and the bandwidth separately, so you can see which line grows as the product does.

Useful related references:

FAQ

How much does it cost to build an app?

Roughly £8,000–£30,000 for a genuine MVP, £30,000–£100,000 for a typical consumer app with accounts, payments and a backend, and beyond £100,000 for complex products with real-time features and many integrations. Feature count and complexity drive the number far more than technology choice.

What are the ongoing costs of running an app?

Hosting for the service, database, storage and bandwidth; a domain; transactional email; error tracking; app store fees if you publish natively; third-party API usage; and maintenance. Budget 15 to 20 percent of the build cost annually just to keep it working, before adding any features.

Is it cheaper to build a web app or a native mobile app?

A web app, usually by a wide margin. Native means two platforms, two review processes, and two release cycles, and cross-platform frameworks reduce that without eliminating it. Go native when you need device capabilities or app store presence for a specific reason, not by default.

How much does hosting an app cost per month?

For a small application, less than most people expect. A service on a small plan plus a managed database starts around $12 a month and grows with traffic. Itemising the service, database, storage and bandwidth separately matters more than the total, because it shows which part scales as usage grows.

How can I reduce the cost of building an app?

Cut scope rather than quality, use familiar technology with large communities, and buy the commodity parts — auth, payments, email, search — instead of building them. Ship a web version first, start on small hosting plans, and make sure the code, repository, and accounts are in your name from day one.